CN101477641A - Demographic method and system based on video monitoring - Google Patents

Abstract

The invention discloses a person number counting method and a system based on video monitoring. Based on video monitoring, the person number counting method can count the person number by detecting heads and estimating and tracking the movement of the heads, and can obviate counting error of manual counting due to the lack of energy. Because a video monitoring camera is usually arranged at a higher position and faces downwards slantwise in the monitored scene, the heads of all persons are visible basically and can be detected on a real-time basis even in a people-concentrated scene, so that the accurate counting can be achieved by continuously tracking the heads in the video, thereby obviating missing counting by manual counting in the people-concentrated scene.

Description

Demographic method and system based on video monitoring

Technical field

The present invention relates to the video monitoring technology, particularly a kind of demographic method and a kind of passenger number statistical system capable based on video monitoring based on video monitoring.

Background technology

In for example scenes such as gateway of supermarket, office building, subway, be provided with the camera of video monitoring system usually and the scenes such as gateway of supermarket, office building, subway carried out video monitoring so that realize.

Simultaneously, for some specific needs, in above-mentioned each scene, the common number that also needs to add up discrepancy.And video monitoring of the prior art can't be realized demographics, thereby above-mentioned statistics need be finished by the people usually.

Yet, because the people who finishes above-mentioned statistics is difficult to concentrate one's energy for a long time to keep the accurate counting to the number of coming in and going out, and leaks the meter number easily when crowded always, thereby realize not only labor intensive of demographics, and the precision of demographics is not high yet by the people.

Summary of the invention

In view of this, the invention provides a kind of demographic method and a kind of passenger number statistical system capable, can realize demographics based on video monitoring based on video monitoring based on video monitoring.

A kind of demographic method based on video monitoring provided by the invention comprises:

A1, in present image, carry out the number of people and detect, determine each number of people in the present image;

A2, utilize the position of each number of people in present image and the present image, estimate the translation vector speed of each number of people in the former frame image;

A3, according to the translation vector speed of each number of people in the former frame image, each number of people in the former frame image is carried out predicting tracing, determine the corresponding respectively number of people in present image of each number of people in the former frame image, also determine newly to appear at the number of people in the present image simultaneously, use when the next frame image is carried out described step a2 and a3;

A4, according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people in the former frame image and/or the former frame image, determine the number in the present image.

Before the described step a1, this method further comprises: a0, utilize the background area of former frame image, detect the foreground area that comprises moving object from present image;

And, only in the foreground area of present image, detect the number of people among the described step a1.

Detect from present image and comprise after the foreground area of moving object, described step a0 further comprises:

A01, pixel matching is carried out in each moving object in each moving object in the former frame image and the present image, and, estimate the translation vector speed of each moving object in the former frame image according to the alternate position spike of moving object in former frame image and present image of pixel matching;

The translation vector speed of each moving object is determined the predicting tracing position of each moving object in the former frame image in a02, the basis former frame image that estimates, and the physical location of each moving object in the predicting tracing position of each moving object in the former frame image and the present image mated, to determine the corresponding respectively moving object and newly appear at moving object in the present image in present image of each moving object in the former frame image;

The moving object of all moving in former two field pictures in a03, the present image is set to the background of present image, uses when detecting the foreground area that comprises moving object from the next frame image.

Described step a1 comprises:

A11, search obtains candidate's number of people window in the foreground area of present image;

The first order sorter that a12, utilization obtain by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows that obtain from search extract little feature of Haar and gray average feature respectively, and carry out the first order according to all candidate's number of people windows that the little feature of Haar that extracts and gray average feature obtain search and detect and filter;

A13, the first order detect is filtered the remaining candidate's number of people window in back carry out gray scale normalization and handle;

The second level sorter that a14, utilization obtain by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows after handling from gray scale normalization extract the little feature of Haar respectively, and all the candidate's number of people windows after according to the little feature of Haar that extracts gray scale normalization being handled carry out the second level and detect and filter;

A15, the second level detect is filtered in remaining all the candidate's number of people windows in back, adjacent a plurality of candidate's number of people windows merge;

All the candidate's number of people windows that a16, calculating merging obtain and the similarity of default number of people feature rule;

A17, similarity is defined as the number of people greater than candidate's number of people window of preset first threshold value.

Among the described step a11, only in the part foreground area of present image, carry out described search according to position, the size and dimension of default counting subregion, and/or when carrying out described search, only search for candidate's number of people window of default people's area of bed.

Described step a2 comprises: each number of people in each number of people in the former frame image and the present image is carried out pixel matching, and, estimate the translation vector speed of each number of people in the former frame image according to the alternate position spike of the number of people in former frame image and present image of pixel matching.

Described step a3 comprises: the predicting tracing position of determining each number of people in the former frame image according to the translation vector speed of each number of people in the former frame image that estimates, and the physical location of each number of people in the predicting tracing position of each number of people in the former frame image and the present image mated, to determine the corresponding respectively number of people and newly appear at the number of people in the present image in present image of each number of people in the former frame image.

In described step a4, the number in the determined present image only comprises: the quantity of the number of people that in the N continuous two field picture, all occurs, and wherein, N is the positive integer more than or equal to 2;

And/or the number in the determined present image only is the position of default counting subregion, the number in the size and dimension.

In described step a4, the number in the determined present image only comprises: in the N continuous two field picture, all occur and the described similarity sum total in the N continuous two field picture greater than the quantity of the number of people of default second threshold value, wherein, N is the positive integer more than or equal to 2.

In described step a4, the further translation vector speed that obtains according to described step a2 is determined in the present image number in the different motion direction respectively.

A kind of passenger number statistical system capable based on video monitoring provided by the invention comprises:

Number of people detection module is used for carrying out the number of people at present image and detects, and determines each number of people in the present image;

The image memory module, the number of people testing result that is used for storing former frame image and each number of people of expression former frame image;

Motion estimation module is used for utilizing the position of present image and each number of people of present image, estimates the translation vector speed of each number of people in the former frame image;

The predicting tracing module, be used for translation vector speed according to each number of people of former frame image, each number of people in the former frame image is carried out predicting tracing, determine the corresponding respectively number of people in present image of each number of people in the former frame image, also determine newly to appear at the number of people in the present image simultaneously, use during for described velocity estimation module and described predicting tracing resume module next frame image;

The quantity determination module is used for determining the number in the present image according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people of former frame image and/or the former frame image.

This system further comprises: the foreground detection module, be used to utilize the background area of former frame image, and from present image, detect the foreground area that comprises moving object;

And described number of people detection module only detects the number of people in the foreground area of present image.

Described foreground detection module comprises: the foreground extraction submodule is used for detecting the foreground area that comprises moving object from present image;

And described foreground detection module further comprises:

The estimation submodule, be used for pixel matching is carried out in each moving object in each moving object of former frame image and the present image, and, estimate the translation vector speed of each moving object in the former frame image according to the alternate position spike of moving object in former frame image and present image of pixel matching;

The predicting tracing submodule, be used for determining the predicting tracing position of each moving object in the former frame image according to the translation vector speed of each moving object of former frame image that estimates, and the physical location of each moving object in the predicting tracing position of each moving object in the former frame image and the present image mated, to determine the corresponding respectively moving object and newly appear at moving object in the present image in present image of each moving object in the former frame image;

The context update submodule is used for the background that present image all not mobile moving object in former two field pictures is set to present image, uses when detecting the foreground area that comprises moving object for described foreground extraction submodule from the next frame image.

Described number of people detection module comprises:

The window search submodule is used for obtaining candidate's number of people window in the foreground area search of present image;

Utilize the first order sorter that obtains by positive sample of some numbers of people and anti-sample training in advance, be used for extracting little feature of Haar and gray average feature respectively, and carry out the first order according to all candidate's number of people windows that the little feature of Haar that extracts and gray average feature obtain search and detect and filter from all candidate's number of people windows that search obtains;

The gray scale normalization submodule is used for that the first order is detected the remaining candidate's number of people window in filtration back and carries out the gray scale normalization processing;

Utilize the second level sorter that obtains by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows after being used for handling from gray scale normalization extract the little feature of Haar respectively, and all the candidate's number of people windows after according to the little feature of Haar that extracts gray scale normalization being handled carry out the second level and detect and filter;

Window merges submodule, is used for remaining all the candidate's number of people windows in filtration back are detected in the second level, and adjacent a plurality of candidate's number of people windows merge;

The similarity calculating sub module is used to calculate and merges all the candidate's number of people windows that obtain and the similarity of default number of people feature rule;

Decision sub-module as a result is used for similarity is defined as the number of people greater than candidate's number of people window of preset first threshold value.

Described window search submodule is only carried out described search according to position, the size and dimension of default counting subregion in the part foreground area of present image, and/or only searches for candidate's number of people window of default people's area of bed when carrying out described search.

Described motion estimation module is carried out pixel matching with each number of people in each number of people in the former frame image and the present image, and, estimate the translation vector speed of each number of people in the former frame image according to the alternate position spike of the number of people in former frame image and present image of pixel matching.

Described predicting tracing module is determined the predicting tracing position of each number of people in the former frame image according to the translation vector speed of each number of people in the former frame image that estimates, and the physical location of each number of people in the predicting tracing position of each number of people in the former frame image and the present image mated, to determine the corresponding respectively number of people and newly appear at the number of people in the present image in present image of each number of people in the former frame image.

Number in the determined present image of described quantity determination module only comprises: the quantity of the number of people that in the N continuous two field picture, all occurs, and wherein, N is the positive integer more than or equal to 2;

And/or the number in the determined present image only is the position of default counting subregion, the number in the size and dimension.

Number in the determined present image of described quantity determination module only comprises: in the N continuous two field picture, all occur and the described similarity sum total in the N continuous two field picture greater than the quantity of the number of people of default second threshold value, wherein, N is the positive integer more than or equal to 2.

The translation vector speed that described quantity determination module further obtains according to described motion estimation module is determined in the present image number in the different motion direction respectively.

As seen from the above technical solution, the present invention is based on video monitoring, and detect by the number of people, and to the estimation of the number of people with follow the tracks of and realize demographics, when avoiding realizing demographics by the people, the counting error that causes owing to the energy that is difficult to keep enough, and consider that the video monitoring video camera is according to being arranged on usually than higher position in monitoring scene, and obtain video under oblique, even it is thereby crowded in the scene, all numbers of people also all are visible basically, can detect all these numbers of people in real time, and in video, carry out continuous tracking to realize precise counting, in the time of can avoiding realizing demographics like this by the people owing to the crowded leakage meter number that causes.

Further, owing to can in image, can detect the foreground area that comprises moving object, therefore, the present invention can only carry out the number of people and detect and need not to carry out the number of people in the background area that the number of people can not occur and detect in foreground area, thereby can get rid of unnecessary testing process, to improve the efficient of demographics.

Detect in the background area that utilizes the former frame image under the situation of foreground area of present image, the present invention can also be by estimation and the predicting tracing to foreground area, judge each moving object that foreground area comprised in the present image, promptly can this judge that stationary object in monitoring scene is to upgrade the background area then, so that the foreground area that detects can be more accurate, thereby can improve the precision of demographics in the next frame image.

Again further, the present invention can utilize the two-stage classification device that search in the foreground area is obtained the similarity that candidate's number of people window carries out two-stage detection filtration and calculates two-stage detection filtration remaining all candidate's number of people windows in back and default number of people feature rule, with the detection of realization to the number of people.

In this case, because detecting the candidate window of filtration, the first order of carrying out first order sorter do not carry out the gray scale normalization processing, thereby can detect and filter out the comparatively complicated non-face image of a large amount of intensity profile, thereby can reduce the processing of second level sorter, improving the efficient that the number of people detects, and then can further improve the efficient of demographics; Because before the similarity of calculated candidate number of people window and default number of people feature rule, can also merge adjacent a plurality of candidate's number of people windows earlier, thereby the corresponding a plurality of candidate's number of people windows of the same number of people have been avoided, further improve the accuracy that the number of people detects, and then can further improve the precision of demographics; And, the appearance of false-alarm is often more isolated owing to the just possible corresponding a plurality of candidate's number of people windows of the real number of people, therefore, the candidate's number of people window after if the present invention only calculates and merges and the similarity of default number of people feature rule, then can avoid the false-alarm flase drop in the image is surveyed is the number of people, thereby improved the accuracy that the number of people detects again further, and then can improve the precision of demographics more further.

In addition, the present invention can be earlier pixel matching by the number of people realize estimation, the location matches by the number of people realizes predicting tracing again, thus the number of people that can avoid being blocked in the monitoring scene omit, and then can improve the precision of demographics again further.

Be actually empty scape and the demographics precision is exerted an influence for fear of once the number of people only occurring, alternatively, the present invention can not consider only to occur the number of people once when determining number, thereby can improve the precision of demographics again further.

And when applying the present invention to the video monitoring of the scenes such as gateway of supermarket, office building, subway for example, can also determine in the present image number respectively in the different motion direction, adapting to actual needs, thereby can make technical scheme of the present invention be with a wide range of applications.

Description of drawings

Fig. 1 is the exemplary process diagram of demographic method in the embodiment of the invention;

The exemplary block diagram of the little feature of Haar that people's face testing process that Fig. 2 adopts for demographic method in the embodiment of the invention is extracted;

Fig. 3 is the employed first order of people's face testing process of demographic method employing in the embodiment of the invention and the composition synoptic diagram of second level sorter;

Fig. 4 is the exemplary process diagram of people's face testing process that demographic method adopts in the embodiment of the invention;

Fig. 5 is the exemplary block diagram of passenger number statistical system capable in the embodiment of the invention;

Fig. 6 is the exemplary block diagram of the foreground detection module of passenger number statistical system capable in the embodiment of the invention;

Fig. 7 is the exemplary block diagram of the number of people detection module of passenger number statistical system capable in the embodiment of the invention.

Embodiment

For making purpose of the present invention, technical scheme and advantage clearer, below with reference to the accompanying drawing embodiment that develops simultaneously, the present invention is described in more detail.

In the present embodiment based on video monitoring, and detect by the number of people, and to the estimation of the number of people with follow the tracks of and realize demographics, when avoiding realizing demographics by the people, the counting error that causes owing to the energy that is difficult to keep enough, and consider that the video monitoring video camera is according to being arranged on usually than higher position in monitoring scene, and obtain video under oblique, even it is thereby crowded in the scene, all numbers of people also all are visible basically, can detect all these numbers of people in real time, and in video, carry out continuous tracking to realize precise counting, in the time of can avoiding realizing demographics like this by the people owing to the crowded leakage meter number that causes.

Fig. 1 is the exemplary process diagram of demographic method in the embodiment of the invention.As shown in Figure 1, based on each two field picture in the demographic method of video monitoring receiver, video monitoring image successively, and successively every two field picture is carried out following steps as present image in the present embodiment:

Step 101 is utilized the background area of former frame image, detects the foreground area that comprises moving object from present image.

In this step, can adopt existing any foreground detection mode, give unnecessary details no longer one by one at this.

Step 102 is carried out the number of people and is detected in the foreground area of present image, determine each number of people in the present image.

In this step, can realize that the number of people detects according to existing any number of people detection mode; Certainly, also can realize that the number of people detects according to a kind of mode that proposes in the present embodiment based on the two-stage classification device.What wherein, present embodiment proposed realizes that based on the two-stage classification device mode that the number of people detects please see for details hereinafter.

Need to prove, because step 101 has detected the foreground area that comprises moving object in present image, and present embodiment realizes that the number of people of demographics institute foundation must belong to moving object, therefore, can only in foreground area, carry out the number of people to detect in this step and need not to carry out the number of people and detect, thereby can get rid of unnecessary testing process, to improve the efficient of demographics in the background area that the number of people can not occur.

Certainly, above-mentioned steps 101 only is optional step, if execution in step 101 not, then this step need be carried out the number of people and detected in the whole frame of present image, but so only be additionally to have carried out unnecessary testing process, and can not cause substantial influence number of people testing result.

In addition, alternatively, the user can be provided for representing only the pre-interior number of people in this district to be counted the size of the number of people of effectively counting subregion and/or being used to count according to the actual conditions of monitoring scene arbitrarily.At this moment, this step then can only be carried out the number of people and detect, and/or only detect the number of people that meets default people's area of bed when the execution number of people detects according to position, the size and dimension of default counting subregion in the part foreground area of present image.Wherein, above-mentioned any setting is meant that the counting subregion that counting subregion and setting are set at an arbitrary position can have any shape.

Step 103 is utilized the position of each number of people in present image and the present image, estimates the translation vector speed of each number of people in the former frame image.

In this step, can adopt existing any one estimation mode; Certainly, a kind of pixel matching mode that also can adopt present embodiment to provide.Wherein, the pixel matching mode that present embodiment proposed please see for details hereinafter.

Step 104, translation vector speed according to each number of people in the former frame image, each number of people in the former frame image is carried out predicting tracing, determine the corresponding respectively number of people in present image of each number of people in the former frame image, also determine newly to appear at the number of people in the present image simultaneously, at 104 o'clock for the next frame image being carried out described step 103 and being used.

In this step, can adopt existing any predicting tracing mode, the mode that a kind of position-based that also can adopt in the present embodiment be provided mates realizes.The mode of the position-based coupling that wherein, present embodiment proposed please see for details hereinafter.

Step 105 according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people in the former frame image and/or the former frame image, is determined the number in the present image.

Alternatively, the user can be provided for expression according to the actual conditions of monitoring scene arbitrarily only to the effective counting of the pre-interior number of people counting in this district subregion.At this moment, this step then can only be determined the interior number of counting subregion of present image.

So far, this flow process finishes.

Below, each step in the above-mentioned flow process is elaborated respectively:

1) about step 101:

With first two field picture during as present image execution in step 101, entire image is foreground area; And will be to follow-up other two field pictures except that first two field picture during as present image execution in step 101, only some then is the background area for the remaining another part of foreground area usually.

Like this, because every two field picture that will be except that first two field picture is during as present image, all need to utilize the background area execution in step 101 of the former frame image of this two field picture, therefore, step 101 in the present embodiment can be after detection comprises the foreground area of moving object from present image, further alternatively to detected foreground area carry out estimation and predicting tracing, identifying the stationary object that in multiple image, occurs continuously and to upgrade the background area, thereby improve the precision of demographics.

Specifically, detected foreground area is carried out estimation can adopt existing any estimation mode, a kind of pixel matching mode that also can adopt present embodiment to propose, this mode comprises: based on the mode of block of pixels, pixel matching is carried out in each moving object in each moving object in the former frame image and the present image, and, estimate the translation vector speed of each moving object in the former frame image according to the alternate position spike of moving object in former frame image and present image of pixel matching.

And carry out predicting tracing for detected foreground area, then can adopt existing any predicting tracing mode, the mode that the position-based that also can adopt present embodiment to propose mates realizes, this mode comprises: according to the translation vector speed of each moving object in the former frame image that estimates, directly determine, or determine the predicting tracing position of each moving object in the former frame image according to the clustering processing result of each moving object in the former frame image, and the physical location of each moving object in the predicting tracing position of each moving object in the former frame image and the present image mated, to determine the corresponding respectively moving object in present image of each moving object in the former frame image, and newly appear at moving object in the present image.

After this, the background, M that the moving object of all moving in preceding M two field picture in can present image is set to present image used during for execution in step 101 from the next frame image for more than or equal to 1 positive integer.

2) about step 102:

Present embodiment proposed a kind ofly to realize the mode that the number of people detects based on the two-stage classification device, and all candidate's number of people windows that this mode obtains search from the foreground area of the whole frame of present image or present image adopt three grades to detect and filter.Wherein, the first order detects to filter utilizes the two class sorters of " number of people/non-number of people " to realize, and at candidate's number of people window of handling without gray scale normalization; The second level is detected to filter and is also utilized the two class sorters of " number of people/non-number of people " to realize, but at candidate's number of people window of handling through gray scale normalization; The similarity that the third level then is based on candidate's number of people window and number of people feature rule realizes.The two class sorters of " number of people/non-number of people " abbreviate " sorter " in this article as.

Above-mentioned first order sorter and second level sorter can utilize the Adaboost theory of maturation in the existing human face detection tech to realize, first order sorter extracts the microstructure features (abbreviating the little feature of Harr as) and the gray average feature of Ha Er (Haar) small echo from candidate's number of people window, second level sorter then only extracts the little feature of Harr from candidate's number of people window; First order sorter and second level sorter can determine based on the feature that is extracted whether the rectangle number of people candidate window of certain yardstick is the number of people.

Preferably, present embodiment has adopted 6 kinds of little features of Haar and the a kind of gray average feature shown in Fig. 2 rightmost side shown in Fig. 2 left side.For 6 kinds of little features of Haar as shown in Figure 2, the difference of corresponding black region and white portion interior pixel gray average obtains feature in the computed image of the present invention; For the gray average feature, the present invention then calculates the average of all pixels in the rectangle frame.

Wherein, the background image of the above-mentioned black region ordinary representation number of people, above-mentioned white portion be people's face in the ordinary representation number of people, especially number of people front then; And in 6 kinds of little features of group as shown in Figure 2, the length and width of black region or white portion can be selected arbitrarily, and the size that only need be no more than candidate's number of people window gets final product.

Certainly, the actual little feature of using of Haar can be not limited to as shown in Figure 26 kinds, but comprises one of the following or combination in any:

The equal value difference of pixel grey scale between black region that the left and right sides is adjacent and the white portion, promptly be not limited to black region and white portion which on a left side, which is on the right side;

The equal value difference of pixel grey scale between a neighbouring black region and the white portion, promptly be not limited to black region and white portion which last, which is down;

The equal value difference of pixel grey scale between black region two white portions adjacent with its left and right sides;

The equal value difference of pixel grey scale between white portion two black regions adjacent with its left and right sides;

The black regions that two diagonal angles link to each other with the equal value difference of pixel grey scale between the white portion that adjacent two diagonal angles link to each other, are not limited to the relative position relation of black region and white portion;

The equal value difference of pixel grey scale between black region that the diagonal angle links to each other and the white portion is not limited to the relative position relation of black region and white portion.

And on the specific implementation of first order sorter and second level sorter, present embodiment consists of a strong classifier based on Adaboost is theoretical with a plurality of Weak Classifiers based on single feature, then a plurality of strong classifiers are cascaded into two class sorters of complete " number of people/non-number of people ", first order sorter promptly required for the present invention, second level sorter.Referring to Fig. 3, first order sorter, second level sorter are formed by the above-mentioned strong classifier cascade of n layer, at first order sorter, when second level sorter detects, if it is (False) vacation that certain one deck strong classifier in the n layer strong classifier is differentiated candidate's number of people window, then get rid of this window and further do not differentiate, if it is true to be output as (True), the more complicated strong classifier of one deck is differentiated this window under then using.That is to say that each layer strong classifier can both allow almost whole number of people samples pass through, and refuses most of non-number of people sample.Candidate's number of people window of input low layer strong classifier is just many like this, and the high-rise candidate's number of people window of input significantly reduces.

In addition, for the first order sorter and the second level sorter of said structure, also need to utilize positive sample of a large amount of numbers of people and the anti-sample of the number of people to train in advance.Wherein, the number of people image that the positive sample of the number of people can comprise from number of people front, side, the back side and top are taken has promptly covered different attitudes, different hair, has worn the true number of people image of different caps; And the anti-sample of the number of people comprises for example any image that does not comprise the number of people such as landscape, animal, literal; And concrete training method can realize based on existing Adaboost theory, does not repeat them here.

And, in order to guarantee to the positive sample of all numbers of people and the anti-sample of the number of people to be in equal conditions when processed, before training, the present invention can set the size of sample searches window earlier, for example 13 * 13, sample searches window by first order sorter and second level sorter utilization setting size carries out cutting and size normalized to positive sample of all numbers of people and the anti-sample of the number of people then, obtains positive sample of the measure-alike number of people and the anti-sample of the number of people.

Like this, as shown in Figure 4, the concrete processing procedure of step 102 just can comprise:

Step 102a, search obtains candidate's number of people window in the foreground area of the whole frame of present image or present image.

Preferably, in order to guarantee that as far as possible all possible candidate's number of people window can not omitted in the input picture, the processing procedure in this step can specifically comprise: earlier to the image of input carry out mirror image, for example 1.05 multiple lengths cun amplify or 0.95 multiple length such as cun dwindles at the convergent-divergent of preset ratio, for example ± 10 rotation of predetermined angle such as degree; Then the input image and carry out in described convergent-divergent, the described postrotational image, obtain some candidate's number of people windows of different size with exhaustive mode search; At last, the some candidate's number of people windows with different size carry out the size normalized again, obtain some candidate's number of people windows of preset standard size.

Like this, can avoid candidate's number of people window of different angles or different sizes to be omitted to greatest extent; Also can guarantee in follow-up processing procedure, all candidate's number of people windows be adopted the processing of equal conditions.

In addition, for as mentioned before, the user only can be provided for arbitrarily according to the actual conditions of monitoring scene representing to this district pre-in the effective size of counting subregion and/or the number of people that is used to count of number of people counting.At this moment, this step then can only be carried out search according to position, the size and dimension of default counting subregion in the part foreground area of present image, and/or only searches for the candidate's number of people window that meets default people's area of bed when carrying out search.

The first order sorter that step 102b, utilization obtain by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows that obtain from search extract little feature of Haar and gray average feature respectively, and carry out the first order according to all candidate's number of people windows that the little feature of Haar that extracts and gray average feature obtain search and detect and filter.

Because all candidate's number of people windows are after gray scale normalization is handled, might exist candidate's number of people window of some non-number of people similar for the intensity profile of candidate's number of people window of the number of people with reality, distinguish comparatively difficulty, therefore, this step does not carry out the gray scale normalization processing and detects the candidate's number of people window that filters some above-mentioned non-number of people by the first order excluding to all candidate's number of people windows earlier, to reduce follow-up is the processing of distinguishing candidate's number of people window of some above-mentioned non-number of people, thereby can improve the efficient that the number of people detects.

Need to prove that the kind quantity of the little feature of Harr that extracts from each candidate's number of people window in this step can be set arbitrarily; The length and width of black region or white portion can be selected arbitrarily in the little feature of Harr, and the size that only need be no more than candidate's number of people window gets final product; Position in the little feature of Harr also can be selected arbitrarily.

Step 102c, the first order detect is filtered the remaining candidate's number of people window in back carry out gray scale normalization and handle.

The second level sorter that step 102d, utilization obtain by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows after handling from gray scale normalization extract the little feature of Haar respectively, and all the candidate's number of people windows after according to the little feature of Haar that extracts gray scale normalization being handled carry out the second level and detect and filter.

Though all candidate's number of people windows are after gray scale normalization is handled, might exist candidate's number of people window of some non-number of people similar for the intensity profile of candidate's number of people window of the number of people, distinguish comparatively difficulty with reality, but, candidate's number of people window of some above-mentioned non-number of people is excluded because having detected when filtering for the first time, therefore, the subsequent step that begins from this step has all been avoided the processing to candidate's number of people window of some above-mentioned non-number of people, thereby has improved the efficient that the number of people detects.

Need to prove that the kind quantity of the little feature of Harr that extracts from each candidate's number of people window in this step can be set arbitrarily; The length and width of black region or white portion can be selected arbitrarily in the little feature of Harr, and the size that only need be no more than candidate's number of people window gets final product; Position in the little feature of Harr also can be selected arbitrarily.

Step 102e, the second level detect is filtered in remaining all the candidate's number of people windows in back, adjacent a plurality of candidate's number of people windows merge.

Described adjacent can being meant of this step: size difference each other less than pre-set dimension difference threshold value and/or position difference less than predeterminated position difference threshold value and/or overlapping area greater than default overlapping area threshold value.

Because some neighboring candidate number of people window that search obtains from input picture, in fact may be corresponding be the same number of people in this input picture, therefore, a plurality of neighboring candidate number of people windows for fear of the same number of people of correspondence are identified as the different numbers of people respectively, by this step adjacent a plurality of candidate's number of people windows are merged into one and only handled at the candidate's number of people window after merging by subsequent step, with the accuracy of raising number of people detection, thus the precision of raising demographics; And, the appearance of false-alarm is often more isolated owing to the just possible corresponding a plurality of candidate's number of people windows of the real number of people, therefore, if subsequent step is only handled at the candidate's number of people window after merging, then can avoid the false-alarm flase drop in the image is surveyed is the number of people, thereby can improve the accuracy that the number of people detects again further, and then can further improve the precision of demographics.

Certainly, because the effect of this step mainly is to improve the accuracy that the number of people detects, only be to reduce the accuracy that the number of people detects and the realization that can not hinder the number of people to detect if do not carry out this step, so this step is optional step.

All the candidate's number of people windows that step 102f, calculating merging obtain and the similarity of default number of people feature rule.

In this step, default number of people feature rule comprises default normal man boundary's feature in front.

Correspondingly, the concrete processing procedure in this step can comprise: extract the mode of point in detecting according to existing people's face earlier, extract each point in candidate's number of people window; Utilize each point in the existing Sobel operator calculated candidate number of people window at the boundary value edge_x of x direction and at the boundary value edge_y of y direction then; After this, utilize formula again

The border amplitude of each point in the calculated candidate number of people window, utilize formula The boundary direction of each point in the calculated candidate number of people window; At last, the border amplitude of each point and boundary direction and the normal man border amplitude of each point on boundary and the similarity of boundary direction in front in the calculated candidate number of people window obtains the similarity of candidate's number of people window and default number of people feature rule.

Need to prove, because step 102e is optional step, therefore, when behind the execution of step 102d, execution in step 102e and when directly carrying out this step, this step can be calculated the second level in the manner described above and detect the similarity of filtering remaining all candidate's number of people windows in back and default number of people feature rule not.

Step 102g, similarity is defined as the number of people greater than candidate's number of people window of preset first threshold value.

So far, the flow process of number of people testing process as shown in Figure 4 finishes.

3) about step 103:

Present embodiment is for the concrete processing procedure of step 103, having proposed a kind of pixel matching mode comprises: based on block of pixels, each number of people in each number of people in the former frame image and the present image is carried out pixel matching, and, estimate the translation vector speed of each number of people in the former frame image according to the alternate position spike of the number of people in former frame image and present image of pixel matching.

4) about step 104:

Present embodiment is for the concrete processing procedure of step 104, provide a kind of mode of position-based coupling to comprise: the predicting tracing position of determining each number of people in the former frame image according to the translation vector speed of each number of people in the former frame image that estimates, and the physical location of each number of people in the predicting tracing position of each number of people in the former frame image and the present image mated, to determine the corresponding respectively number of people and newly appear at the number of people in the present image in present image of each number of people in the former frame image.Wherein, if any number of people in the former frame image, the number of people of location coupling in present image then can be determined the number of people of this number of people in the former frame image corresponding and its location matches in present image; If any number of people in the former frame image, the number of people of location coupling not in present image determines that then this number of people in the former frame image temporarily disappears; If the number of people in the former frame image that any number of people in the present image does not match determines that then this number of people in the present image is the number of people that newly appears in the present image.

For example, the predicting tracing position of each number of people is provided with corresponding prediction rectangle frame in the former frame image, and the physical location place of each number of people is provided with the relevant detection rectangle frame in present image; Calculate respectively then and predict that respectively rectangle frame and each detect the overlapping area of rectangle frame, overlapping area is big more, then the corresponding detection rectangle frame position of expression might be more this prediction rectangle frame the position of the corresponding number of people in present image, therefore, to predict that one of rectangle frame overlapping area maximum is detected rectangle frame with each, be defined as the detection rectangle frame of this prediction rectangle frame location matches respectively, according to the prediction rectangle frame and the detection rectangle frame of location matches, determine each number of people corresponding number of people of difference in present image in the former frame image then.That is, will with the number of people in the pairing present image of detection rectangle frame of the prediction rectangle frame overlapping area maximum of each number of people in the former frame image, be defined as the corresponding respectively number of people in present image of the number of people in the former frame image respectively; In present image, in the former frame image, do not find the pairing number of people of detection rectangle frame of overlapping prediction rectangle frame, be defined as the emerging number of people in present image.

In addition, each number of people in the former frame image usually can only corresponding one detects rectangle frame, and one is detected the number of people of rectangle frame in usually also can only corresponding former frame image.So as certain number of people in the former frame image corresponding any one detect rectangle frame, i.e. the prediction rectangle frame and the equal zero lap of all monitoring rectangle frames in the present image of this number of people, then think this number of people disappearance temporarily in present image in the former frame image.But, do not delete this number of people immediately in the present embodiment but still follow the tracks of this number of people in the former frame image, when follow-up every two field picture is carried out this step as present image, continue to upgrade the prediction rectangle frame of this number of people according to the point-to-point speed of this number of people, all do not have overlapping detection rectangle frame as this prediction rectangle in the continuous P two field picture, P is the positive integer greater than 1, determines that again this number of people disappears, otherwise thinks that this number of people reappears.

5) about step 105:

How to determine number in the present image in step 105, can set arbitrarily according to the actual conditions and the needs of monitoring scene.

For example, suppose the quantity of each number of people in the former frame image, quantity greater than the corresponding number of people of the difference in present image of each number of people in the former frame image, represent that then at least one individual in the former frame image is blocked or disappears in present image, the monitoring scene that can comprise whole closed room scene so for image, because the no one leaves from closed room, thereby the number that step 105 can be defined as the quantity of each number of people in the former frame image in the present image gets final product; And move very frequent monitoring scene for people such as subway gateways, since the people normally fast folk prescription seldom can be blocked to moving owing to stop, thereby the number that step 105 can be defined as the quantity of the corresponding respectively number of people in present image of each number of people in the former frame image in the present image get final product, or the while considers that more multiple other conditions determine.

Again for example, for the comparatively frequent monitoring scenes such as gate of coming in and going out as the stream of people, the translation vector speed that step 105 also can further obtain according to step 103, determine in the present image number respectively in the different motion direction, promptly, if certain number of people has striden across the line of entering, and the direction of point-to-point speed is consistent with the direction of entering, and the number of people number of then entering adds 1; Striden across the line of going out as certain number of people, and the direction of point-to-point speed is consistent with the direction of going out, the number of people number of then going out adds 1.

This shows, determine in this step that the concrete mode of number in the present image can only and need be set according to the actual conditions of monitoring scene, therefore, can't give unnecessary details one by one at this.

Need to prove, in the step 105 why during the number in determining present image, only consider according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people in the former frame image and/or the former frame image, and do not consider newly to appear at number of people quantity in the present image, this is because the number of people that newly appears in the present image only continues to occur in follow-up at least one two field picture, just might not be empty scape.

Thus, which kind of no matter adopts specifically establish rules then really, preferably, present embodiment number in the determined present image in step 105 only comprises: the quantity of the number of people that all occurs in the N continuous two field picture, wherein, N is the positive integer more than or equal to 2.In addition, as previously mentioned, alternatively, the user can be provided for expression according to the actual conditions of monitoring scene arbitrarily only to the effective counting of the pre-interior number of people counting in this district subregion, so at this moment, number in the determined present image of step 105 only is the position of default counting subregion, the number of people quantity in the size and dimension.

Certainly, for number of people detection mode as shown in Figure 4, preferably, number in step 105 in the determined present image only comprises: in the N continuous two field picture, all occur and the sum total of the resulting similarity of step 102f in the N continuous two field picture greater than the quantity of the number of people of default second threshold value, wherein, N is the positive integer more than or equal to 2.Promptly not judging with the quantity that this number of people occurs, is basis for estimation but whether have higher similarity with this number of people in multiple image.

More than be in the present embodiment based on the detailed description of the demographic method of video monitoring.Below, again the passenger number statistical system capable based on video monitoring in the present embodiment is described.

Fig. 5 is the exemplary block diagram of passenger number statistical system capable in the embodiment of the invention.As shown in Figure 5, the passenger number statistical system capable in the present embodiment comprises: foreground detection module 501, number of people detection module 502, image memory module 503, motion estimation module 504, predicting tracing module 505 and quantity determination module 506.

Foreground detection module 501 is used for utilizing the background area of former frame image according to existing any foreground detection mode, detects the foreground area that comprises moving object from present image.

Number of people detection module 502 is used for carrying out the number of people in the foreground area of present image and detects, and determines each number of people in the present image.Wherein, number of people detection module 502 can be realized the number of people based on the principle of work of existing any number of people detection mode, also can realize that the number of people detects based on the principle of work based on this mode of two-stage classification device that is proposed according to method part in the present embodiment.

Need to prove, foreground detection module 501 is optional module, for the situation that comprises foreground detection module 501, number of people detection module 502 need not to carry out in the background area that the number of people can not occur the number of people and detects, thereby can get rid of unnecessary testing process, to improve the efficient of demographics; And for the situation that does not comprise foreground detection module 501, number of people detection module 502 directly carries out the number of people and detects in whole frame present image.Further alternatively, the user can be according to the actual conditions of monitoring scene, only in number of people detection module 502, be provided for arbitrarily representing to this district pre-in number of people counting effectively count subregion and/or the size of the number of people that is used to count; At this moment, 502 of number of people detection modules can only be carried out the number of people and detect, and/or only detect the number of people that meets default people's area of bed when the execution number of people detects according to position, the size and dimension of default counting subregion in the part foreground area of present image.Wherein, above-mentioned any setting is meant that the counting subregion that counting subregion and setting are set at an arbitrary position can have any shape.

Image memory module 503, the number of people testing result that is used for storing former frame image and each number of people of expression former frame image.Wherein, in order to save the hardware resource of realizing storage, image memory module 503 can only be stored a two field picture, the number of people testing result of representing each number of people in this two field picture, and other relevant informations in this two field picture, promptly after system as shown in Figure 5 finishes processing to present image, present image, the number of people testing result of each number of people in the expression present image, and other relevant informations in the current frame image all can be stored to image memory module 503, with the former frame image in the overlay image memory module 503, the number of people testing result of each number of people in the expression former frame image, and other relevant informations in the last whole two field picture.Like this, for the next frame image, present image is promptly as the former frame image of this next frame image.

Motion estimation module 504 is used for utilizing the position of present image and each number of people of present image, estimates the translation vector speed of each number of people in the former frame image.Wherein, motion estimation module 504 can realize estimation based on the principle of work of existing any one estimation mode; Certainly, the principle of work of the motion estimation module 504 pixel matching mode that also can be provided based on present embodiment method part realizes estimation.

Predicting tracing module 505, be used for translation vector speed according to each number of people of former frame image, each number of people in the former frame image is carried out predicting tracing, determine the corresponding respectively number of people in present image of each number of people in the former frame image, also determine newly to appear at the number of people in the present image simultaneously, use when handling the next frame images for described velocity estimation module 504 and described predicting tracing module 505.Wherein, predicting tracing module 505 can realize predicting tracing based on the principle of work of existing any predicting tracing mode; The principle of work of the position-based matching way that predicting tracing module 505 also can be provided based on method part in the present embodiment realizes predicting tracing.

Quantity determination module 506 is used for determining the number in the present image according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people of former frame image and/or the former frame image.Wherein, further alternatively, the user can be according to the actual conditions of monitoring scene, also can in quantity determination module 506, be provided for arbitrarily expression only to this district pre-in the effective subregion of counting of number of people counting, at this moment, 506 of quantity determination modules can only be determined the interior number of counting subregion of present image.

Below, each module in the said system is elaborated.

Fig. 6 is the exemplary block diagram of the foreground detection module of passenger number statistical system capable in the embodiment of the invention.As shown in Figure 6, comprise foreground extraction submodule 511 in the foreground detection module 501, be used for utilizing the background area of former frame image, from present image, detect the foreground area that comprises moving object according to existing any foreground detection mode; Background storage submodule 510, the background area that is used to store the former frame image.Wherein, during as present image, entire image is foreground area at first two field picture of video monitoring; And to follow-up other two field pictures except that first two field picture during as present image, only some then is the background area for the remaining another part of foreground area usually.

Like this, because every two field picture that will be except that first two field picture is during as present image, foreground extraction submodule 511 all needs to utilize the background area of the former frame image of this two field picture, and therefore, the foreground detection module 501 in the present embodiment can also comprise:

Estimation submodule 512, be used for mode based on block of pixels, pixel matching is carried out in each moving object in each moving object in the former frame image and the present image, and, estimate the translation vector speed of each moving object in the former frame image according to the alternate position spike of moving object in former frame image and present image of pixel matching;

Clustering processing submodule 513, clustering processing is carried out in each moving object that is used for foreground area that foreground extraction submodule 511 is obtained; Clustering processing submodule 514 is optional submodule;

Predicting tracing submodule 514, be used for translation vector speed according to each moving object of former frame image that estimates, directly determine or determine the predicting tracing position of each moving object in the former frame image according to the clustering processing result of each moving object in the former frame image, and the physical location of each moving object in the predicting tracing position of each moving object in the former frame image and the present image mated, to determine the corresponding respectively moving object and newly appear at moving object in the present image in present image of each moving object in the former frame image;

Context update submodule 515, be used for background that moving object that present image all moves is set to present image, M for more than or equal to 1 positive integer in preceding M two field picture, and be updated in the background storage submodule 510, use when from the next frame image, detecting the foreground area that comprises moving object for described foreground extraction submodule 511.

Thus, by detected foreground area is carried out estimation and predicting tracing, to identify the stationary object that in multiple image, occurs continuously and to upgrade the background area, can improve the precision of demographics.

Fig. 7 is the exemplary block diagram of the foreground detection module of passenger number statistical system capable in the embodiment of the invention.As shown in Figure 7, the principle of work of the pixel matching mode that is provided based on present embodiment method part, number of people detection module 502 comprises:

Window search submodule 521 is used for obtaining candidate's number of people window in the foreground area search of present image; Preferably, in order to guarantee that as far as possible all possible candidate's number of people window can not omitted in the input picture, the processing procedure of window search submodule 521 can specifically comprise: earlier to the image of input carry out mirror image, for example 1.05 multiple lengths cun amplify or 0.95 multiple length such as cun dwindles at the convergent-divergent of preset ratio, for example ± 10 rotation of predetermined angle such as degree; Then the input image and carry out in described convergent-divergent, the described postrotational image, obtain some candidate's number of people windows of different size with exhaustive mode search; At last, the some candidate's number of people windows with different size carry out the size normalized again, obtain some candidate's number of people windows of preset standard size; In addition, for as mentioned before, the user can be provided for representing only the pre-interior number of people in this district to be counted the size of the number of people of effectively counting subregion and/or being used to count according to the actual conditions of monitoring scene arbitrarily, at this moment, window search submodule 521 then can only be carried out search according to position, the size and dimension of default counting subregion suddenly in the part foreground area of present image, and/or only searches for the candidate's number of people window that meets default people's area of bed when the execution search;

Utilize the first order sorter 522 that obtains by positive sample of some numbers of people and anti-sample training in advance, be used for extracting little feature of Haar and gray average feature respectively, and carry out the first order according to all candidate's number of people windows that the little feature of Haar that extracts and gray average feature obtain search and detect and filter from all candidate's number of people windows that search obtains; Wherein, first order sorter 522 can be set arbitrarily from the kind quantity of the little feature of Harr of each candidate's number of people window extraction; The length and width of black region or white portion can be selected arbitrarily in the little feature of Harr, and the size that only need be no more than candidate's number of people window gets final product; Position in the little feature of Harr also can be selected arbitrarily;

Gray scale normalization submodule 523 is used for that the first order is detected the remaining candidate's number of people window in filtration back and carries out the gray scale normalization processing;

Utilize the second level sorter 524 that obtains by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows after being used for handling from gray scale normalization extract the little feature of Haar respectively, and all the candidate's number of people windows after according to the little feature of Haar that extracts gray scale normalization being handled carry out the second level and detect and filter; Wherein, second level sorter 524 can be set arbitrarily from the kind quantity of the little feature of Harr of each candidate's number of people window extraction; The length and width of black region or white portion can be selected arbitrarily in the little feature of Harr, and the size that only need be no more than candidate's number of people window gets final product; Position in the little feature of Harr also can be selected arbitrarily;

Window merges submodule 525, is used for remaining all the candidate's number of people windows in filtration back are detected in the second level, and adjacent a plurality of candidate's number of people windows merge; Wherein, adjacent can being meant described here: size difference each other less than pre-set dimension difference threshold value and/or position difference less than predeterminated position difference threshold value and/or overlapping area greater than default overlapping area threshold value; And it is optional that window merges submodule 525;

Similarity calculating sub module 526 is used to calculate and merges all the candidate's number of people windows that obtain and the similarity of default number of people feature rule; Wherein, for not comprising the situation that window merges submodule 525, similarity calculating sub module 526 can be calculated the second level and detect the similarity of filtering remaining all candidate's number of people windows in back and default number of people feature rule; And the concrete processing procedure that similarity calculating sub module 526 is calculated similarity can comprise: extract the mode of point in detecting according to existing people's face earlier, extract each point in candidate's number of people window; Utilize each point in the existing Sobel operator calculated candidate number of people window at the boundary value edge_x of x direction and at the boundary value edge_y of y direction then; After this, utilize formula again

The border amplitude of each point in the calculated candidate number of people window, utilize formula The boundary direction of each point in the calculated candidate number of people window; At last, the border amplitude of each point and boundary direction and the normal man border amplitude of each point on boundary and the similarity of boundary direction in front in the calculated candidate number of people window obtains the similarity of candidate's number of people window and default number of people feature rule;

Decision sub-module 527 as a result, are used for similarity is defined as the number of people greater than candidate's number of people window of preset first threshold value.

For the motion estimation module 504 in the system as shown in Figure 5, if its principle of work based on the pixel matching mode that present embodiment method part is provided realizes estimation, then this motion estimation module 504 is carried out pixel matching with each number of people in each number of people in the former frame image and the present image, and, estimate the translation vector speed of each number of people in the former frame image according to the alternate position spike of the number of people in former frame image and present image of pixel matching.

For the predicting tracing module 505 in the system as shown in Figure 5, if its principle of work based on the position-based matching way that method part in the present embodiment is provided realizes predicting tracing, then this predicting tracing module 505 needs to determine according to the translation vector speed of each number of people in the former frame image that estimates the predicting tracing position of each number of people in the former frame image, and the physical location of each number of people in the predicting tracing position of each number of people in the former frame image and the present image mated, to determine the corresponding respectively number of people in present image of each number of people in the former frame image, and newly appear at the number of people in the present image.Wherein, if any number of people in the former frame image, the number of people of location coupling in present image then can be determined the number of people of this number of people in the former frame image corresponding and its location matches in present image; If any number of people in the former frame image, the number of people of location coupling not in present image determines that then this number of people in the former frame image temporarily disappears; If the number of people in the former frame image that any number of people in the present image does not match determines that then this number of people in the present image is the number of people that newly appears in the present image.

For example, predicting tracing module 505 predicting tracing position of each number of people in the former frame image is provided with corresponding prediction rectangle frame, and the physical location place of each number of people is provided with the relevant detection rectangle frame in present image; Calculate respectively then and predict that respectively rectangle frame and each detect the overlapping area of rectangle frame, overlapping area is big more, then the corresponding detection rectangle frame position of expression might be more this prediction rectangle frame the position of the corresponding number of people in present image, therefore, to predict that one of rectangle frame overlapping area maximum is detected rectangle frame with each, be defined as the detection rectangle frame of this prediction rectangle frame location matches respectively, according to the prediction rectangle frame and the detection rectangle frame of location matches, determine each number of people corresponding number of people of difference in present image in the former frame image then.That is, predicting tracing module 505 will with the number of people in the pairing present image of detection rectangle frame of the prediction rectangle frame overlapping area maximum of each number of people in the former frame image, be defined as the corresponding respectively number of people in present image of the number of people in the former frame image respectively; Predicting tracing module 505 does not find the pairing number of people of detection rectangle frame of overlapping prediction rectangle frame with in the present image in the former frame image, be defined as the emerging number of people in present image.

In addition, each number of people in the former frame image usually can only corresponding one detects rectangle frame, and one is detected the number of people of rectangle frame in usually also can only corresponding former frame image.So as certain number of people in the former frame image corresponding any one detect rectangle frame, i.e. the prediction rectangle frame and the equal zero lap of all monitoring rectangle frames in the present image of this number of people, then predicting tracing module 505 is thought this number of people disappearance temporarily in present image in the former frame image.But, do not delete this number of people immediately in the present embodiment but still follow the tracks of this number of people in the former frame image, when follow-up every two field picture is carried out this step as present image, continue to upgrade the prediction rectangle frame of this number of people according to the point-to-point speed of this number of people, all do not have overlapping detection rectangle frame as this prediction rectangle in the continuous P two field picture, P is the positive integer greater than 1, determines that again this number of people disappears, otherwise thinks that this number of people reappears.

For the quantity determination module 506 in the system as shown in Figure 5, how to determine the number in the present image, can set arbitrarily according to the actual conditions and the needs of monitoring scene.Specifically see also the example that the method part is lifted, the quantity determination module 506 translation vector speed that can further obtain for example according to described motion estimation module, in the number of different motion direction, other all possible examples do not repeat them here in definite respectively present image.

Need to prove that preferably, for fear of empty scape, the number in the present image that quantity determination module 506 is determined only comprises: the quantity of the number of people that in the N continuous two field picture, all occurs, wherein, N is the positive integer more than or equal to 2; And, alternatively, the user can be provided for expression according to the actual conditions of monitoring scene arbitrarily only to the effective counting of the pre-interior number of people counting in this district subregion, so this moment, the number in the quantity determination module 506 determined present images only is the position of default counting subregion, the number of people quantity in the size and dimension.

Certainly, for based on the number of people detection module 502 of structure as shown in Figure 7, number in the quantity determination module 506 determined present images only comprises: in the N continuous two field picture, all occur and the described similarity sum total in the N continuous two field picture greater than the quantity of the number of people of default second threshold value, wherein, N is the positive integer more than or equal to 2.

The above is preferred embodiment of the present invention only, is not to be used to limit protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of being done, be equal to and replace and improvement etc., all should be included within protection scope of the present invention.

Claims (20)

1, a kind of demographic method based on video monitoring is characterized in that, this method comprises:

A1, in present image, carry out the number of people and detect, determine each number of people in the present image;

A2, utilize the position of each number of people in present image and the present image, estimate the translation vector speed of each number of people in the former frame image;

A3, according to the translation vector speed of each number of people in the former frame image, each number of people in the former frame image is carried out predicting tracing, determine the corresponding respectively number of people in present image of each number of people in the former frame image, also determine newly to appear at the number of people in the present image simultaneously, use when the next frame image is carried out described step a2 and a3;

A4, according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people in the former frame image and/or the former frame image, determine the number in the present image.

2, the method for claim 1 is characterized in that, before the described step a1, this method further comprises: a0, utilize the background area of former frame image, detect the foreground area that comprises moving object from present image;

And, only in the foreground area of present image, detect the number of people among the described step a1.

3, method as claimed in claim 2 is characterized in that, detects to comprise after the foreground area of moving object from present image, and described step a0 further comprises:

A01, pixel matching is carried out in each moving object in each moving object in the former frame image and the present image, and, estimate the translation vector speed of each moving object in the former frame image according to the alternate position spike of moving object in former frame image and present image of pixel matching;

The translation vector speed of each moving object is determined the predicting tracing position of each moving object in the former frame image in a02, the basis former frame image that estimates, and the physical location of each moving object in the predicting tracing position of each moving object in the former frame image and the present image mated, to determine the corresponding respectively moving object and newly appear at moving object in the present image in present image of each moving object in the former frame image;

The moving object of all moving in former two field pictures in a03, the present image is set to the background of present image, uses when detecting the foreground area that comprises moving object from the next frame image.

4, method as claimed in claim 2 is characterized in that, described step a1 comprises:

A11, search obtains candidate's number of people window in the foreground area of present image;

The first order sorter that a12, utilization obtain by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows that obtain from search extract little feature of Haar and gray average feature respectively, and carry out the first order according to all candidate's number of people windows that the little feature of Haar that extracts and gray average feature obtain search and detect and filter;

A13, the first order detect is filtered the remaining candidate's number of people window in back carry out gray scale normalization and handle;

The second level sorter that a14, utilization obtain by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows after handling from gray scale normalization extract the little feature of Haar respectively, and all the candidate's number of people windows after according to the little feature of Haar that extracts gray scale normalization being handled carry out the second level and detect and filter;

A15, the second level detect is filtered in remaining all the candidate's number of people windows in back, adjacent a plurality of candidate's number of people windows merge;

All the candidate's number of people windows that a16, calculating merging obtain and the similarity of default number of people feature rule;

A17, similarity is defined as the number of people greater than candidate's number of people window of preset first threshold value.

5, method as claimed in claim 4, it is characterized in that, among the described step a11, position, size and dimension according to default counting subregion are only carried out described search in the part foreground area of present image, and/or only search for candidate's number of people window of default people's area of bed when carrying out described search.

6, as each described method in the claim 1 to 5, it is characterized in that, described step a2 comprises: each number of people in each number of people in the former frame image and the present image is carried out pixel matching, and, estimate the translation vector speed of each number of people in the former frame image according to the alternate position spike of the number of people in former frame image and present image of pixel matching.

7, as each described method in the claim 1 to 5, it is characterized in that, described step a3 comprises: the predicting tracing position of determining each number of people in the former frame image according to the translation vector speed of each number of people in the former frame image that estimates, and the physical location of each number of people in the predicting tracing position of each number of people in the former frame image and the present image mated, to determine the corresponding respectively number of people and newly appear at the number of people in the present image in present image of each number of people in the former frame image.

As each described method in the claim 1 to 5, it is characterized in that 8, in described step a4, the number in the determined present image only comprises: the quantity of the number of people that in the N continuous two field picture, all occurs, wherein, N is the positive integer more than or equal to 2;

And/or the number in the determined present image only is the position of default counting subregion, the number in the size and dimension.

9, method as claimed in claim 4, it is characterized in that, in described step a4, number in the determined present image only comprises: in the N continuous two field picture, all occur and the described similarity sum total in the N continuous two field picture greater than the quantity of the number of people of default second threshold value, wherein, N is the positive integer more than or equal to 2.

As each described method in the claim 1 to 5, it is characterized in that 10, in described step a4, the further translation vector speed that obtains according to described step a2 is determined in the present image number in the different motion direction respectively.

11, a kind of passenger number statistical system capable based on video monitoring is characterized in that, comprising:

Number of people detection module is used for carrying out the number of people at present image and detects, and determines each number of people in the present image;

The image memory module, the number of people testing result that is used for storing former frame image and each number of people of expression former frame image;

Motion estimation module is used for utilizing the position of present image and each number of people of present image, estimates the translation vector speed of each number of people in the former frame image;

The predicting tracing module, be used for translation vector speed according to each number of people of former frame image, each number of people in the former frame image is carried out predicting tracing, determine the corresponding respectively number of people in present image of each number of people in the former frame image, also determine newly to appear at the number of people in the present image simultaneously, use during for described velocity estimation module and described predicting tracing resume module next frame image;

The quantity determination module is used for determining the number in the present image according to the quantity of the corresponding number of people respectively in present image of each number of people in the quantity of each number of people of former frame image and/or the former frame image.

12, system as claimed in claim 11 is characterized in that, this system further comprises: the foreground detection module, be used to utilize the background area of former frame image, and from present image, detect the foreground area that comprises moving object;

And described number of people detection module only detects the number of people in the foreground area of present image.

13, system as claimed in claim 12 is characterized in that, described foreground detection module comprises: the foreground extraction submodule is used for detecting the foreground area that comprises moving object from present image;

And described foreground detection module further comprises:

The estimation submodule, be used for pixel matching is carried out in each moving object in each moving object of former frame image and the present image, and, estimate the translation vector speed of each moving object in the former frame image according to the alternate position spike of moving object in former frame image and present image of pixel matching;

The predicting tracing submodule, be used for determining the predicting tracing position of each moving object in the former frame image according to the translation vector speed of each moving object of former frame image that estimates, and the physical location of each moving object in the predicting tracing position of each moving object in the former frame image and the present image mated, to determine the corresponding respectively moving object and newly appear at moving object in the present image in present image of each moving object in the former frame image;

The context update submodule is used for the background that present image all not mobile moving object in former two field pictures is set to present image, uses when detecting the foreground area that comprises moving object for described foreground extraction submodule from the next frame image.

14, system as claimed in claim 12 is characterized in that, described number of people detection module comprises:

The window search submodule is used for obtaining candidate's number of people window in the foreground area search of present image;

Utilize the first order sorter that obtains by positive sample of some numbers of people and anti-sample training in advance, be used for extracting little feature of Haar and gray average feature respectively, and carry out the first order according to all candidate's number of people windows that the little feature of Haar that extracts and gray average feature obtain search and detect and filter from all candidate's number of people windows that search obtains;

The gray scale normalization submodule is used for that the first order is detected the remaining candidate's number of people window in filtration back and carries out the gray scale normalization processing;

Utilize the second level sorter that obtains by positive sample of some numbers of people and anti-sample training in advance, all candidate's number of people windows after being used for handling from gray scale normalization extract the little feature of Haar respectively, and all the candidate's number of people windows after according to the little feature of Haar that extracts gray scale normalization being handled carry out the second level and detect and filter;

Window merges submodule, is used for remaining all the candidate's number of people windows in filtration back are detected in the second level, and adjacent a plurality of candidate's number of people windows merge;

The similarity calculating sub module is used to calculate and merges all the candidate's number of people windows that obtain and the similarity of default number of people feature rule;

Decision sub-module as a result is used for similarity is defined as the number of people greater than candidate's number of people window of preset first threshold value.

15, system as claimed in claim 14, it is characterized in that, described window search submodule is only carried out described search according to position, the size and dimension of default counting subregion in the part foreground area of present image, and/or only searches for candidate's number of people window of default people's area of bed when carrying out described search.

16, as each described system in the claim 11 to 15, it is characterized in that, described motion estimation module is carried out pixel matching with each number of people in each number of people in the former frame image and the present image, and, estimate the translation vector speed of each number of people in the former frame image according to the alternate position spike of the number of people in former frame image and present image of pixel matching.

17, as each described system in the claim 11 to 15, it is characterized in that, described predicting tracing module is determined the predicting tracing position of each number of people in the former frame image according to the translation vector speed of each number of people in the former frame image that estimates, and the physical location of each number of people in the predicting tracing position of each number of people in the former frame image and the present image mated, to determine the corresponding respectively number of people and newly appear at the number of people in the present image in present image of each number of people in the former frame image.

As each described system in the claim 11 to 15, it is characterized in that 18, the number in the determined present image of described quantity determination module only comprises: the quantity of the number of people that in the N continuous two field picture, all occurs, wherein, N is the positive integer more than or equal to 2;

And/or the number in the determined present image only is the position of default counting subregion, the number in the size and dimension.

19, system as claimed in claim 14, it is characterized in that, number in the determined present image of described quantity determination module only comprises: in the N continuous two field picture, all occur and the described similarity sum total in the N continuous two field picture greater than the quantity of the number of people of default second threshold value, wherein, N is the positive integer more than or equal to 2.

As each described system in the claim 11 to 15, it is characterized in that 20, the translation vector speed that described quantity determination module further obtains according to described motion estimation module is determined in the present image number in the different motion direction respectively.

Images