CN106951898A - Recommend method and system, electronic equipment in a kind of vehicle candidate region - Google Patents
Recommend method and system, electronic equipment in a kind of vehicle candidate region Download PDFInfo
- Publication number
- CN106951898A CN106951898A CN201710153509.3A CN201710153509A CN106951898A CN 106951898 A CN106951898 A CN 106951898A CN 201710153509 A CN201710153509 A CN 201710153509A CN 106951898 A CN106951898 A CN 106951898A
- Authority
- CN
- China
- Prior art keywords
- pixel
- image
- candidate region
- sum
- input picture
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/255—Detecting or recognising potential candidate objects based on visual cues, e.g. shapes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/46—Descriptors for shape, contour or point-related descriptors, e.g. scale invariant feature transform [SIFT] or bags of words [BoW]; Salient regional features
- G06V10/462—Salient features, e.g. scale invariant feature transforms [SIFT]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
- G06V20/54—Surveillance or monitoring of activities, e.g. for recognising suspicious objects of traffic, e.g. cars on the road, trains or boats
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
- G06V20/584—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads of vehicle lights or traffic lights
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
- Traffic Control Systems (AREA)
Abstract
The present invention provides a kind of vehicle candidate region and recommends method and system, electronic equipment, including input picture is pre-processed based on significance analysis, obtains including the target image of vehicle image；Binary conversion treatment is carried out to the target image, binary image is obtained；Boundary mergence based on gradient is carried out to the binary image, recommended vehicle candidate region is obtained.Vehicle candidate region can be more precisely located in the vehicle candidate region recommendation method and system of the present invention, electronic equipment.
Description
Technical field
The present invention relates to the technical field of image procossing, more particularly to a kind of vehicle candidate region recommends method and is
System, electronic equipment.
Background technology
Vehicle testing techniques are the important components of intelligent transportation system.Traffic intelligence management needs to examine by vehicle
Survey technology gathers objective, effective Traffic Information, obtains the magnitude of traffic flow, speed, roadway occupancy, following distance, vehicle class
The basic datas such as type, so as to purposefully realize monitoring, control, analysis, decision-making, dispatch and the intellectuality means such as dredge.
Specifically, vehicle testing techniques refer to enter line search and judgement to the vehicle in image using image sensing means,
With the process for many attribute (such as position, speed, shape, outward appearance) for obtaining vehicle in image.It is field of automotive active safety,
Especially realize one of key technology of knock into the back early warning and automatic emergency brake function.Current vehicle testing techniques extensive use
In intelligent transportation system and senior drive assist system (Advanced Driver Assistant Systems, ADAS).In city
In the intelligentized traffic system in city, many traffic block ports are fitted with video sensor, can all produce daily thousands of
Video data.And traffic density is big in urban transportation, traffic congestion is serious, and diversity is presented in each road user, is handed over from city
Lead to detection in complicated background and obtain vehicle, it is most important to urban transportation and municipal public safety.Auxiliary system is driven senior
In system, vehicle testing techniques are mainly used in frontal collisions early warning system (Forward Collision Warning, FCW), lead to
Vehicle testing techniques are crossed, the distance between this car and front truck, orientation and relative velocity is judged, and when there is potential risk of collision
Driver is alerted.
At present, the vehicle checking method of main flow is all to use first to screen candidate region, reuses machine learning means or depth
The method that degree learning wayses are further accurately confirmed, to obtain the positional information of vehicle in the picture.Generally, candidate is screened
Region uses the method that candidate region is recommended.Therefore, the method positioning that candidate region is recommended is more accurate, and grader is helped to get over
Greatly.Because positioning is more accurate, the fraction that grader is returned can be higher, so as to suppress false-alarm to a certain extent.
In the prior art, the method that candidate region is recommended can substantially be divided into two categories below：
(1)grouping method
Specifically, first picture is smashed, a kind of method then polymerizeing again.Such as selective search algorithms.
(2)window scoring method。
Specifically, a large amount of window are generated and are given a mark, a kind of low point of method is then filtered out.Such as objectness is calculated
Method.
(3) method between Jie's above two algorithm, such as multibox.
But, existing candidate region recommends requirement of the method to weather condition, road conditions and background environment all higher, very
Hardly possible is obtained and is accurately positioned, and is difficult unified applicable to different types of vehicle, false-alarm easily occurs.Therefore, how more
Vehicle candidate region, which is precisely located, becomes those skilled in the art institute urgent problem to be solved.
The content of the invention
The shortcoming of prior art in view of the above, it is an object of the invention to provide a kind of vehicle candidate region recommendation side
Method and system, electronic equipment, by carrying out the pretreatment based on significance analysis to input picture, contain mesh after being screened
Mark the candidate region of vehicle；The boundary mergence based on gradient is carried out to the candidate target region containing target vehicle again, waited
The recommendation region of choosing, so as to realize being accurately positioned for vehicle candidate region.
In order to achieve the above objects and other related objects, the present invention provides a kind of vehicle candidate region recommendation method, including
Following steps：Input picture is pre-processed based on significance analysis, obtains including the target image of vehicle image；To institute
State target image and carry out binary conversion treatment, obtain binary image；The border based on gradient is carried out to the binary image to close
And, obtain recommended vehicle candidate region.
In one embodiment of the invention, the input picture is the Y channel image information Jing Guo sampling processing.
It is described that input picture is pre-processed based on significance analysis in one embodiment of the invention, included
The target image of vehicle image comprises the following steps：
The input picture is traveled through, the pixel value of each pixel in the input picture is obtained；
Each pixel is calculated to other each pixels apart from sum, and records ultimate range sum and minimum range
Sum；
The difference of the ultimate range sum and the minimum range sum is calculated, and according to the difference to each pixel
Point is corresponding to be normalized apart from sum；
The input picture is normalized, obtains stretching image；It is each after the input picture normalized
The excursion of the pixel value of individual pixel is identical with each excursion after sum normalized；
Pixel value based on each pixel in the stretching image and apart from sum, the conspicuousness for calculating each pixel is special
Value indicative, obtains significance analysis image；
The significance analysis image is subtracted into the stretching image, target image is obtained.
In one embodiment of the invention, binary conversion treatment is carried out to the target image by following steps：
By the pixel of each pixel on the target image and horizontal direction upper left side pixel and subtract and horizontal direction
The pixel of upper right side pixel and, obtain the first difference；
By the pixel of the pixel of top two on each pixel on the target image and vertical direction and subtract and vertically
The pixel of two pixels of direction upper and lower and, obtain the second difference；
When at least one in first difference and second difference is more than predetermined threshold value, by the picture of corresponding pixel points
Plain value is set to 1；Otherwise, the pixel value of corresponding pixel points is set to 0.
In one embodiment of the invention, the boundary mergence based on gradient is carried out to the binary image, obtains being recommended
Vehicle candidate region comprise the following steps：
Record the line segment of each horizontal direction in the binary image；
The line segment of adjacent or alternate horizontal direction is merged, until merging into a line segment；The merging refers to will be upper
The line segment of the horizontal direction of side is moved downward to the line segment of the horizontal direction of lower section, and the length of superposition two lines section；
By the line segment base candidate line obtained after merging, choose using base candidate line as base, the length of base candidate line
The vehicle candidate region recommended is used as the square area of the length of side.
Meanwhile, the present invention provides a kind of vehicle candidate region commending system, including significance analysis module, binarization block
With boundary mergence module；
The significance analysis module is used to pre-process input picture based on significance analysis, obtains including car
The target image of image；
The binary conversion treatment module is used to carry out binary conversion treatment to the target image, obtains binary image；
The boundary mergence module is used to carry out the boundary mergence based on gradient to the binary image, obtains being recommended
Vehicle candidate region.
In one embodiment of the invention, the input picture is the Y channel image information Jing Guo sampling processing.
In one embodiment of the invention, the significance analysis module performs following operate：
The input picture is traveled through, the pixel value of each pixel in the input picture is obtained；
Each pixel is calculated to other each pixels apart from sum, and records ultimate range sum and minimum range
Sum；
The difference of the ultimate range sum and the minimum range sum is calculated, and according to the difference to each pixel
Point is corresponding to be normalized apart from sum；
The input picture is normalized, obtains stretching image；It is each after the input picture normalized
The excursion of the pixel value of individual pixel is identical with each excursion after sum normalized；
Pixel value based on each pixel in the stretching image and apart from sum, the conspicuousness for calculating each pixel is special
Value indicative, obtains significance analysis image；
The significance analysis image is subtracted into the stretching image, target image is obtained.
In one embodiment of the invention, the binary conversion treatment module performs following operate：
By the pixel of each pixel on the target image and horizontal direction upper left side pixel and subtract and horizontal direction
The pixel of upper right side pixel and, obtain the first difference；
By the pixel of the pixel of top two on each pixel on the target image and vertical direction and subtract and vertically
The pixel of two pixels of direction upper and lower and, obtain the second difference；
When at least one in first difference and second difference is more than predetermined threshold value, by the picture of corresponding pixel points
Plain value is set to 1；Otherwise, the pixel value of corresponding pixel points is set to 0.
In one embodiment of the invention, the boundary mergence module performs following operate：
Record the line segment of each horizontal direction in the binary image；
The line segment of adjacent or alternate horizontal direction is merged, until merging into a line segment；The merging refers to will be upper
The line segment of the horizontal direction of side is moved downward to the line segment of the horizontal direction of lower section, and the length of superposition two lines section；
By the line segment base candidate line obtained after merging, choose using base candidate line as base, the length of base candidate line
The vehicle candidate region recommended is used as the square area of the length of side.
In addition, the present invention also provides a kind of electronic equipment, including any of the above-described described vehicle candidate region commending system.
As described above, method and system, electronic equipment are recommended in the vehicle candidate region of the present invention, with following beneficial effect
Really：
(1) target image of information relative abundance is obtained using significance analysis, by asking for level and vertical direction
Unequal gradient determines that candidate base is carried to carry out level and with having tendentiousness border enhancing on vertical direction to be follow-up
For tendentious edge material；
(2) candidate's base screening is carried out using the binary image for enhancing edge, is further enhanced using line of fall mechanism
The characteristic range of base candidate region, so as to vehicle candidate region is more precisely located.
Brief description of the drawings
Fig. 1 is shown as the flow chart of the vehicle candidate region recommendation method of the present invention；
Fig. 2 is shown as the flow chart pre-processed based on significance analysis to input picture of the present invention；
Fig. 3 is shown as the flow chart of the image binaryzation processing of the present invention；
Fig. 4 is shown as the flow chart that the boundary mergence based on gradient is carried out to binaryzation target image of the present invention；
Fig. 5 is shown as the schematic diagram that the line segment of the present invention merges；
Fig. 6 is shown as the structural representation of the vehicle candidate region commending system of the present invention；
Fig. 7 is shown as the structural representation of the electronic equipment of the present invention.
Component label instructions
1 significance analysis module
2 binarization blocks
3 boundary mergence modules
Embodiment
Illustrate embodiments of the present invention below by way of specific instantiation, those skilled in the art can be by this specification
Disclosed content understands other advantages and effect of the present invention easily.The present invention can also pass through specific realities different in addition
The mode of applying is embodied or practiced, the various details in this specification can also based on different viewpoints with application, without departing from
Various modifications or alterations are carried out under the spirit of the present invention.
It should be noted that the diagram provided in the present embodiment only illustrates the basic conception of the present invention in a schematic way,
Then only display is painted with relevant component in the present invention rather than according to component count, shape and the size during actual implementation in schema
System, it is actual when implementing, and kenel, quantity and the ratio of each component can be a kind of random change, and its assembly layout kenel also may be used
Can be increasingly complex.
Method and system, electronic equipment is recommended to be based on significance analysis and gradient boundaries conjunction in the vehicle candidate region of the present invention
And method be more accurately determined the vehicle candidate region in image, giving grader by vehicle candidate region is carried out more
Plus accurate judgement, so as to recognize target vehicle image, improve the degree of accuracy of vehicle detection.
Method is recommended to comprise the following steps in reference picture 1, vehicle candidate region of the invention：
Step S1, based on significance analysis input picture is pre-processed, obtain including the target figure of vehicle image
Picture.
In the present invention, input picture is gathered by image capture device, using yuv format.Wherein, YUV is European
A kind of colour coding method that television system is used, wherein " Y " represents lightness (Luminance or Luma), that is, ash
Rank is worth；And that " U " and " V " expression is then colourity (Chrominance or Chroma), effect is description colors of image and saturation
Degree, for specified pixel color its.That is, the monochrome information of Y channel image information table diagram pictures, U channel images letter
The colour information of breath and V channel image information table diagram pictures, by Y channel images information, U channel images information and V channel images
Information combines, it becomes possible to form colour picture.In the present invention, input picture is the Y channel images Jing Guo sampling processing
Information.
As shown in Fig. 2 being pre-processed based on significance analysis to input picture, obtain including the target of vehicle image
Image comprises the following steps：
Step S11, traversal input picture, obtain the pixel value of each pixel in input picture, and record max pixel value
And minimum pixel value.
Step S12, calculate each pixel to other each pixels apart from sum, and record ultimate range sum with
Minimum range sum.
Preferably, the distance of each pixel other each pixels into input picture is Euclidean distance, but is not limited to
Euclidean distance.Euclidean distance is a distance definition generally used, refers to the actual distance between two points in m-dimensional space,
Or the natural length (i.e. distance of the point to origin) of vector.Euclidean distance in two and three dimensions space be exactly 2 points it
Between actual range.
In the present invention, each pixel is used as the measurement pixel contrast to other each pixels apart from sum
A kind of measurement.
Step S13, the difference for calculating ultimate range sum and minimum range sum, and according to the difference to each pixel
It is corresponding to be normalized apart from sum.
Specifically, it is according to ultimate range sum and the difference of minimum range sum, each pixel is corresponding apart from it
In the range of Linear Mapping to 0-255.
The difference of foundation ultimate range sum and minimum range sum is corresponding to each pixel to return apart from sum
After one change processing, the data obtained after normalization are deployed line by line, the array for representing each pixel value efficiency is obtained.Namely
Say, in accordance with the order from top to bottom, it is 256 that the data obtained after each row is normalized, which are stitched together obtain a size successively,
Array.The index of the array is 0-255.
Step S14, input picture is normalized, obtains stretching image；It is each after input picture normalized
The excursion of the pixel value of individual pixel is identical with each excursion after sum normalized.
Specifically, when input picture being normalized, by the pixel value Linear Mapping of each pixel to 0-255
In the range of, so as to obtain stretching image.
Step S15, the pixel value based on each pixel in stretching image and apart from sum, calculate the notable of each pixel
Property characteristic value, obtains significance analysis image.
Specifically, it regard the pixel value for stretching each pixel in image as the array for representing each pixel value efficiency
Index, finds respective value from the array for representing each pixel value efficiency, is used as the significant characteristics value of each pixel.Respectively
The significant characteristics value of pixel constitutes significance analysis image.
Step S16, significance analysis image subtracted into stretching image, obtain target image.
Significance analysis image and tensile diagram seem in identical Value space, and with mapping relations.Moreover, significantly
Property analysis image carried out enhancing processing to the strong region relatively of the contrast in input picture, and tensile diagram seem it is approximate
The image of weighing apparatus.The difference of significance analysis image and stretching image is asked for, then obtains target image.Target image eliminates input
Enhanced region is relatively free of in image, reservation enhances the strong region of contrast, that is, enhances the bag of contrast strongly
The region of vehicle image is included, other background areas are eliminated, so as to obtain the mesh of prominent saliency object, i.e. vehicle image
Logo image.
Step S2, to target image carry out binary conversion treatment, obtain binary image.
In order that the contrast for obtaining target image is stronger, binary conversion treatment is carried out to target image.Binary image
Obvious black and white effect is showed, the positioning of vehicle candidate region is easily facilitated.
Specifically, as shown in figure 3, carrying out binary conversion treatment to target image by following steps：
Step S21, by the pixel of each pixel on target image and horizontal direction upper left side pixel and subtract and level
On direction right side pixel pixel and, obtain the first difference.
It should be noted that when the pixel of two pixels and during more than 255, then by pixel and 255 are modified to.
Step S22, by the pixel of two pixels in top on each pixel on target image and vertical direction and subtract with
The pixel of two pixels of vertical direction upper and lower and, obtain the second difference.
It should be noted that when the pixel of three pixels and during more than 255, then by pixel and 255 are modified to.
Step S23, when at least one in the first difference and the second difference be more than predetermined threshold value when, by the picture of corresponding pixel points
Plain value is set to 1；Otherwise, the pixel value of corresponding pixel points is set to 0.
So far, it just can obtain binary image.
Step S3, to binary image carry out the boundary mergence based on gradient, obtain recommended vehicle candidate region.
As shown in figure 4, step S3 comprises the following steps：
Step S31, each horizontal direction recorded in binary image line segment.
Specifically, the line segment of horizontal direction is obtained from binary image, and records each water in binary image
Square to line segment vertical direction line number and starting point and ending point in the horizontal direction.
Step S32, the line segment merging by adjacent or alternate horizontal direction, until merging into a line segment；Merging refers to
The line segment of the horizontal direction of top is moved downward to the line segment of the horizontal direction of lower section, and the length of superposition two lines section.
Specifically, in the vertical direction, searches the relation between adjacent straightway from low to high.By the adjacent poor a line in top
Or the straightway of multirow is moved downward to the straightway of lower section, two such straightway is possible to overlap.When two straight up and down
When line segment merges, length is the superposition of latter two mobile length of straigh line.In addition, the intersegmental alternate line number root of two straight lines up and down
Depending on concrete condition.
For example, as shown in figure 5, first merge line segment 2 with line segment 3, line segment length L1 is obtained, then by the line segment after merging
Merge with line segment 1, obtain final line segment, its length is L2.
Step S33, by the line segment base candidate line obtained after merging, choose using base candidate line as base, base is waited
The length of route selection is used as the vehicle candidate region recommended as the square area of the length of side.
After the vehicle candidate region is obtained, export to classifier modules, accurately to determine the position of vehicle in the picture
Information.
Reference picture 6, vehicle candidate region commending system of the invention includes significance analysis module 1, the and of binarization block 2
Boundary mergence module 3.
Significance analysis module 1 is used to pre-process input picture based on significance analysis, obtains including vehicle
The target image of image.
In the present invention, input picture is gathered by image capture device, using yuv format.Wherein, YUV is European
A kind of colour coding method that television system is used, wherein " Y " represents lightness (Luminance or Luma), that is, ash
Rank is worth；And that " U " and " V " expression is then colourity (Chrominance or Chroma), effect is description colors of image and saturation
Degree, for specified pixel color its.That is, the monochrome information of Y channel image information table diagram pictures, U channel images letter
The colour information of breath and V channel image information table diagram pictures, by Y channel images information, U channel images information and V channel images
Information combines, it becomes possible to form colour picture.In the present invention, input picture is the Y channel images Jing Guo sampling processing
Information.
As shown in Fig. 2 significance analysis module 1 performs following operation：
Step S11, traversal input picture, obtain the pixel value of each pixel in input picture, and record max pixel value
And minimum pixel value.
Step S12, calculate each pixel to other each pixels apart from sum, and record ultimate range sum with
Minimum range sum.
Preferably, the distance of each pixel other each pixels into input picture is Euclidean distance, but is not limited to
Euclidean distance.Euclidean distance is a distance definition generally used, refers to the actual distance between two points in m-dimensional space,
Or the natural length (i.e. distance of the point to origin) of vector.Euclidean distance in two and three dimensions space be exactly 2 points it
Between actual range.
In the present invention, each pixel is used as the measurement pixel contrast to other each pixels apart from sum
A kind of measurement.
Step S13, the difference for calculating ultimate range sum and minimum range sum, and according to the difference to each pixel
It is corresponding to be normalized apart from sum.
Specifically, it is according to ultimate range sum and the difference of minimum range sum, each pixel is corresponding apart from it
In the range of Linear Mapping to 0-255.
The difference of foundation ultimate range sum and minimum range sum is corresponding to each pixel to return apart from sum
After one change processing, the data obtained after normalization are deployed line by line, the array for representing each pixel value efficiency is obtained.Namely
Say, in accordance with the order from top to bottom, it is 256 that the data obtained after each row is normalized, which are stitched together obtain a size successively,
Array.The index of the array is 0-255.
Step S14, input picture is normalized, obtains stretching image；It is each after input picture normalized
The excursion of the pixel value of individual pixel is identical with each excursion after sum normalized.
Specifically, when input picture being normalized, by the pixel value Linear Mapping of each pixel to 0-255
In the range of, so as to obtain stretching image.
Step S15, the pixel value based on each pixel in stretching image and apart from sum, calculate the notable of each pixel
Property characteristic value, obtains significance analysis image.
Specifically, it regard the pixel value for stretching each pixel in image as the array for representing each pixel value efficiency
Index, finds respective value from the array for representing each pixel value efficiency, is used as the significant characteristics value of each pixel.Respectively
The significant characteristics value of pixel constitutes significance analysis image.
Step S16, significance analysis image subtracted into stretching image, obtain target image.
Significance analysis image and tensile diagram seem in identical Value space, and with mapping relations.Moreover, significantly
Property analysis image carried out enhancing processing to the strong region relatively of the contrast in input picture, and tensile diagram seem it is approximate
The image of weighing apparatus.The difference of significance analysis image and stretching image is asked for, then obtains target image.Target image eliminates input
Enhanced region is relatively free of in image, reservation enhances the strong region of contrast, that is, enhances the bag of contrast strongly
The region of vehicle image is included, other background areas are eliminated, so as to obtain the mesh of prominent saliency object, i.e. vehicle image
Logo image.
Binary conversion treatment module 2 is connected with significance analysis module 1, for carrying out binary conversion treatment to target image, obtains
To binary image.
In order that the contrast for obtaining target image is stronger, binary conversion treatment is carried out to target image.Binary image
Obvious black and white effect is showed, the positioning of vehicle candidate region is easily facilitated.
Specifically, as shown in figure 3, binary conversion treatment module 2 carries out binary conversion treatment by following steps to target image：
Step S21, by the pixel of each pixel on target image and horizontal direction upper left side pixel and subtract and level
On direction right side pixel pixel and, obtain the first difference.
It should be noted that when the pixel of two pixels and during more than 255, then by pixel and 255 are modified to.
Step S22, by the pixel of two pixels in top on each pixel on target image and vertical direction and subtract with
The pixel of two pixels of vertical direction upper and lower and, obtain the second difference.
It should be noted that when the pixel of three pixels and during more than 255, then by pixel and 255 are modified to.
Step S23, when at least one in the first difference and the second difference be more than predetermined threshold value when, by the picture of corresponding pixel points
Plain value is set to 1；Otherwise, the pixel value of corresponding pixel points is set to 0.
So far, it just can obtain binary image.
Boundary mergence module 3 is connected with binary conversion treatment module 2, for carrying out the border based on gradient to binary image
Merge, obtain recommended vehicle candidate region.
As shown in figure 4, boundary mergence module 3 performs following operation：
Step S31, each horizontal direction recorded in binary image line segment.
Specifically, the line segment of horizontal direction is obtained from binary image, and records each water in binary image
Square to line segment vertical direction line number and starting point and ending point in the horizontal direction.
Step S32, the line segment merging by adjacent or alternate horizontal direction, until merging into a line segment；Merging refers to
The line segment of the horizontal direction of top is moved downward to the line segment of the horizontal direction of lower section, and the length of superposition two lines section.
Specifically, in the vertical direction, searches the relation between adjacent straightway from low to high.By the adjacent poor a line in top
Or the straightway of multirow is moved downward to the straightway of lower section, two such straightway is possible to overlap.When two straight up and down
When line segment merges, length is the superposition of latter two mobile length of straigh line.In addition, the intersegmental alternate line number root of two straight lines up and down
Depending on concrete condition.
For example, as shown in figure 5, first merge line segment 2 with line segment 3, line segment length L1 is obtained, then by the line segment after merging
Merge with line segment 1, obtain final line segment, its length is L2.
Step S33, by the line segment base candidate line obtained after merging, choose using base candidate line as base, base is waited
The length of route selection is used as the vehicle candidate region recommended as the square area of the length of side.
After the vehicle candidate region is obtained, export to classifier modules, accurately to determine the position of vehicle in the picture
Information.
As shown in fig. 7, the present invention also provides a kind of server, including above-mentioned vehicle candidate region commending system, its is specific
Structure is as described above, therefore will not be repeated here.
In summary, vehicle candidate region of the invention recommends method and system, electronic equipment to be obtained using significance analysis
To the target image of information relative abundance, level is carried out by asking for the unequal gradient of level and vertical direction and vertical
Border strengthens with having tendentiousness on direction, and then determines that candidate base provides tendentious edge material to be follow-up；Utilize enhancing
The binary image at edge carries out candidate base screening, and the feature of base candidate region is further enhancing using line of fall mechanism
Scope, so as to which vehicle candidate region is more precisely located.So, the present invention effectively overcomes of the prior art a variety of
Shortcoming and have high industrial utilization.
The above-described embodiments merely illustrate the principles and effects of the present invention, not for the limitation present invention.It is any ripe
Know the personage of this technology all can carry out modifications and changes under the spirit and scope without prejudice to the present invention to above-described embodiment.Cause
This, those of ordinary skill in the art is complete without departing from disclosed spirit and institute under technological thought such as
Into all equivalent modifications or change, should by the present invention claim be covered.
Claims (11)
1. method is recommended in a kind of vehicle candidate region, it is characterised in that：Comprise the following steps：
Input picture is pre-processed based on significance analysis, obtains including the target image of vehicle image；
Binary conversion treatment is carried out to the target image, binary image is obtained；
Boundary mergence based on gradient is carried out to the binary image, recommended vehicle candidate region is obtained.
2. method is recommended in vehicle candidate region according to claim 1, it is characterised in that：The input picture is by adopting
The Y channel image information of sample processing.
3. method is recommended in vehicle candidate region according to claim 1, it is characterised in that：It is described to be based on significance analysis pair
Input picture is pre-processed, and the target image for obtaining including vehicle image comprises the following steps：
The input picture is traveled through, the pixel value of each pixel in the input picture is obtained；
Each pixel is calculated to other each pixels apart from sum, and record ultimate range sum and minimum range it
With；
The difference of the ultimate range sum and the minimum range sum is calculated, and according to the difference to each pixel pair
That answers is normalized apart from sum；
The input picture is normalized, obtains stretching image；Each picture after the input picture normalized
The excursion of the pixel value of vegetarian refreshments is identical with each excursion after sum normalized；
Pixel value based on each pixel in the stretching image and apart from sum, calculates the significant characteristics of each pixel
Value, obtains significance analysis image；
The significance analysis image is subtracted into the stretching image, target image is obtained.
4. method is recommended in vehicle candidate region according to claim 1, it is characterised in that：By following steps to the mesh
Logo image carries out binary conversion treatment：
By the pixel of each pixel on the target image and horizontal direction upper left side pixel and subtract upper right with horizontal direction
The pixel of side pixel and, obtain the first difference；
By the pixel of the pixel of top two on each pixel on the target image and vertical direction and subtract and vertical direction
The pixel of two pixels of upper and lower and, obtain the second difference；
When at least one in first difference and second difference is more than predetermined threshold value, by the pixel value of corresponding pixel points
It is set to 1；Otherwise, the pixel value of corresponding pixel points is set to 0.
5. method is recommended in vehicle candidate region according to claim 1, it is characterised in that：The binary image is carried out
Boundary mergence based on gradient, obtains recommended vehicle candidate region and comprises the following steps：
Record the line segment of each horizontal direction in the binary image；
The line segment of adjacent or alternate horizontal direction is merged, until merging into a line segment；The merging refers to top
The line segment of horizontal direction is moved downward to the line segment of the horizontal direction of lower section, and the length of superposition two lines section；
By the line segment base candidate line obtained after merging, choose using base candidate line as base, the long conduct of base candidate line
The square area of the length of side is used as the vehicle candidate region recommended.
6. a kind of vehicle candidate region commending system, it is characterised in that：Including significance analysis module, binarization block and border
Merging module；
The significance analysis module is used to pre-process input picture based on significance analysis, obtains including vehicle figure
The target image of picture；
The binary conversion treatment module is used to carry out binary conversion treatment to the target image, obtains binary image；
The boundary mergence module is used to carry out the boundary mergence based on gradient to the binary image, obtains recommended car
Candidate region.
7. vehicle candidate region according to claim 6 commending system, it is characterised in that：The input picture is by adopting
The Y channel image information of sample processing.
8. vehicle candidate region according to claim 6 commending system, it is characterised in that：The significance analysis module is held
Row is following to be operated：
The input picture is traveled through, the pixel value of each pixel in the input picture is obtained；
Each pixel is calculated to other each pixels apart from sum, and record ultimate range sum and minimum range it
With；
The difference of the ultimate range sum and the minimum range sum is calculated, and according to the difference to each pixel pair
That answers is normalized apart from sum；
The input picture is normalized, obtains stretching image；Each picture after the input picture normalized
The excursion of the pixel value of vegetarian refreshments is identical with each excursion after sum normalized；
Pixel value based on each pixel in the stretching image and apart from sum, calculates the significant characteristics of each pixel
Value, obtains significance analysis image；
The significance analysis image is subtracted into the stretching image, target image is obtained.
9. vehicle candidate region according to claim 6 commending system, it is characterised in that：The binary conversion treatment module is held
Row is following to be operated：
By the pixel of each pixel on the target image and horizontal direction upper left side pixel and subtract upper right with horizontal direction
The pixel of side pixel and, obtain the first difference；
By the pixel of the pixel of top two on each pixel on the target image and vertical direction and subtract and vertical direction
The pixel of two pixels of upper and lower and, obtain the second difference；
When at least one in first difference and second difference is more than predetermined threshold value, by the pixel value of corresponding pixel points
It is set to 1；Otherwise, the pixel value of corresponding pixel points is set to 0.
10. vehicle candidate region according to claim 6 commending system, it is characterised in that：The boundary mergence module is held
Row is following to be operated：
Record the line segment of each horizontal direction in the binary image；
The line segment of adjacent or alternate horizontal direction is merged, until merging into a line segment；The merging refers to top
The line segment of horizontal direction is moved downward to the line segment of the horizontal direction of lower section, and the length of superposition two lines section；
By the line segment base candidate line obtained after merging, choose using base candidate line as base, the long conduct of base candidate line
The square area of the length of side is used as the vehicle candidate region recommended.
11. a kind of electronic equipment, it is characterised in that：Recommend system including the vehicle candidate region described in one of claim 6-10
System.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710153509.3A CN106951898B (en) | 2017-03-15 | 2017-03-15 | Vehicle candidate area recommendation method and system and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710153509.3A CN106951898B (en) | 2017-03-15 | 2017-03-15 | Vehicle candidate area recommendation method and system and electronic equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106951898A true CN106951898A (en) | 2017-07-14 |
| CN106951898B CN106951898B (en) | 2021-06-04 |
Family
ID=59471978
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710153509.3A Active CN106951898B (en) | 2017-03-15 | 2017-03-15 | Vehicle candidate area recommendation method and system and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106951898B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109960984A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on contrast and significance analysis |
| CN109960981A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Left and right vehicle wheel boundary alignment system and device based on gradient and picture contrast |
| CN109960978A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle detecting system and device based on image layered technology |
| CN109961420A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on more subgraphs fusion and significance analysis |
| CN109960979A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on image layered technology |
| CN109960982A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle detecting system and device based on contrast and significance analysis |
| CN109960977A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Based on image layered conspicuousness preprocess method |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102087652A (en) * | 2009-12-08 | 2011-06-08 | 百度在线网络技术（北京）有限公司 | Method for screening images and system thereof |
| CN102156882A (en) * | 2011-04-14 | 2011-08-17 | 西北工业大学 | Method for detecting airport target based on high-resolution remote sensing image |
| US20120148151A1 (en) * | 2010-12-10 | 2012-06-14 | Casio Computer Co., Ltd. | Image processing apparatus, image processing method, and storage medium |
| US20130207920A1 (en) * | 2010-08-20 | 2013-08-15 | Eric McCann | Hand and finger registration for control applications |
| CN104504684A (en) * | 2014-12-03 | 2015-04-08 | 小米科技有限责任公司 | Edge extraction method and device |
| CN104778713A (en) * | 2015-04-27 | 2015-07-15 | 清华大学深圳研究生院 | Image processing method |
| CN105118084A (en) * | 2015-09-10 | 2015-12-02 | 天津大学 | Depth perception enhancement method based on significance |
| CN106203267A (en) * | 2016-06-28 | 2016-12-07 | 成都之达科技有限公司 | Vehicle collision avoidance method based on machine vision |
| CN106295636A (en) * | 2016-07-21 | 2017-01-04 | 重庆大学 | Passageway for fire apparatus based on multiple features fusion cascade classifier vehicle checking method |
-
2017
- 2017-03-15 CN CN201710153509.3A patent/CN106951898B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102087652A (en) * | 2009-12-08 | 2011-06-08 | 百度在线网络技术（北京）有限公司 | Method for screening images and system thereof |
| US20130207920A1 (en) * | 2010-08-20 | 2013-08-15 | Eric McCann | Hand and finger registration for control applications |
| US20120148151A1 (en) * | 2010-12-10 | 2012-06-14 | Casio Computer Co., Ltd. | Image processing apparatus, image processing method, and storage medium |
| CN102156882A (en) * | 2011-04-14 | 2011-08-17 | 西北工业大学 | Method for detecting airport target based on high-resolution remote sensing image |
| CN104504684A (en) * | 2014-12-03 | 2015-04-08 | 小米科技有限责任公司 | Edge extraction method and device |
| CN104778713A (en) * | 2015-04-27 | 2015-07-15 | 清华大学深圳研究生院 | Image processing method |
| CN105118084A (en) * | 2015-09-10 | 2015-12-02 | 天津大学 | Depth perception enhancement method based on significance |
| CN106203267A (en) * | 2016-06-28 | 2016-12-07 | 成都之达科技有限公司 | Vehicle collision avoidance method based on machine vision |
| CN106295636A (en) * | 2016-07-21 | 2017-01-04 | 重庆大学 | Passageway for fire apparatus based on multiple features fusion cascade classifier vehicle checking method |
Non-Patent Citations (3)
| Title |
|---|
| VIJAY GAIKWAD等: "Lane Departure Identification for Advanced Driver Assistance", 《 IEEE TRANSACTIONS ON INTELLIGENT TRANSPORTATION SYSTEMS》 * |
| 彭玉青等: "基于动态模板匹配的移动机器人目标识别", 《传感技术学报》 * |
| 王鑫等: "基于图像显著性区域的遥感图像机场检测", 《计算机辅助设计与图形学学报》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109960984A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on contrast and significance analysis |
| CN109960981A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Left and right vehicle wheel boundary alignment system and device based on gradient and picture contrast |
| CN109960978A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle detecting system and device based on image layered technology |
| CN109961420A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on more subgraphs fusion and significance analysis |
| CN109960979A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle checking method based on image layered technology |
| CN109960982A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Vehicle detecting system and device based on contrast and significance analysis |
| CN109960977A (en) * | 2017-12-25 | 2019-07-02 | 大连楼兰科技股份有限公司 | Based on image layered conspicuousness preprocess method |
| CN109960977B (en) * | 2017-12-25 | 2023-11-17 | 大连楼兰科技股份有限公司 | Saliency preprocessing method based on image layering |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106951898B (en) | 2021-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106951898A (en) | Recommend method and system, electronic equipment in a kind of vehicle candidate region | |
| CN109977812B (en) | Vehicle-mounted video target detection method based on deep learning | |
| CN106599792B (en) | Method for detecting hand driving violation behavior | |
| CN104766071B (en) | A kind of traffic lights fast algorithm of detecting applied to pilotless automobile | |
| CN103400150B (en) | A kind of method and device that road edge identification is carried out based on mobile platform | |
| CN104021378B (en) | Traffic lights real-time identification method based on space time correlation Yu priori | |
| CN103824081B (en) | Method for detecting rapid robustness traffic signs on outdoor bad illumination condition | |
| CN106980813A (en) | Generation is watched in machine learning attentively | |
| CN103530640B (en) | Unlicensed vehicle checking method based on AdaBoost Yu SVM | |
| CN111767878B (en) | Deep learning-based traffic sign detection method and system in embedded device | |
| CN105608455A (en) | License plate tilt correction method and apparatus | |
| CN102163278B (en) | Illegal vehicle intruding detection method for bus lane | |
| Yang et al. | A vehicle license plate recognition system based on fixed color collocation | |
| CN103544480A (en) | Vehicle color recognition method | |
| CN111967396A (en) | Processing method, device and equipment for obstacle detection and storage medium | |
| CN109791607A (en) | It is detected from a series of images of video camera by homography matrix and identifying object | |
| CN103186905A (en) | Color detector for vehicle | |
| CN102142090B (en) | Vehicle detection method and system | |
| CN106339445B (en) | Vehicle retrieval method and device based on big data | |
| CN107145824A (en) | A kind of lane line dividing method and system, car-mounted terminal based on significance analysis | |
| CN102169583A (en) | Vehicle shielding detection and segmentation method based on vehicle window positioning | |
| CN110249366B (en) | Image feature quantity output device, image recognition device, and storage medium | |
| CN112053407B (en) | Automatic lane line detection method based on AI technology in traffic law enforcement image | |
| CN105654140A (en) | Complex industrial environment-oriented wagon number positioning and identifying method for railway tank wagon | |
| CN108268866A (en) | A kind of vehicle checking method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |