CN110033474A - Object detection method, device, computer equipment and storage medium - Google Patents
Object detection method, device, computer equipment and storage medium Download PDFInfo
- Publication number
- CN110033474A CN110033474A CN201910091468.9A CN201910091468A CN110033474A CN 110033474 A CN110033474 A CN 110033474A CN 201910091468 A CN201910091468 A CN 201910091468A CN 110033474 A CN110033474 A CN 110033474A
- Authority
- CN
- China
- Prior art keywords
- image
- target
- local
- target image
- region
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/136—Segmentation; Edge detection involving thresholding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/187—Segmentation; Edge detection involving region growing; involving region merging; involving connected component labelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/20—Analysis of motion
- G06T7/215—Motion-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20024—Filtering details
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
This application involves a kind of object detection method, device, computer equipment and storage mediums, by obtaining target image, the target image is handled according to the local maxima difference of the Local Minimum contrast of the target image and the target image, obtains the local contrast enhancing image of the target image;And then the local contrast of target image enhancing image is filtered, determine real goal region.It can reduce the interference of complex scene using this method, improve Dim targets detection rate, reduce false-alarm.
Description
Technical field
This application involves detection technique fields, more particularly to a kind of object detection method, device, computer equipment and deposit
Storage media.
Background technique
With the development of detection technique, there is complex scene detection technique, is mainly inhibited at present using background, space filter
The method of wave, spatio-temporal filtering and frequency-domain transform.
However, current method, it is low there are verification and measurement ratio the problems such as.
Summary of the invention
Based on this, it is necessary in view of the above technical problems, provide a kind of object detection method, device, computer equipment and
Storage medium.
A kind of object detection method, which comprises
Target image is obtained, according to the local maxima of the Local Minimum contrast of the target image and the target image
Difference handles the target image, obtains the local contrast enhancing image of the target image;
The local contrast enhancing image of the target image is filtered, determines real goal region.
The acquisition target image in one of the embodiments, according to the Local Minimum contrast of the target image
The target image is handled with the local maxima difference of the target image, obtains the local contrast of the target image
Degree enhances image
The maximum value of center image block pixel is obtained, and according to the maximum value and multiple fields of the center image block pixel
The average gray value of image block obtains the local maxima difference of the target image, wherein the neighborhood image block and it is described in
Heart image block is adjacent;
According to the pixel grey scale mean value of each field image block in multiple fields image block and the center image block picture
The maximum value of element, obtains the Local Minimum contrast of the target image;
According to the local maxima difference of the Local Minimum contrast of the target image and the target image, obtain described
The local contrast of target image enhances image.
The maximum value for obtaining the center image block pixel in one of the embodiments, and according to the center
The average gray value of the maximum value of image block pixel and the multiple field image block, obtains the local maxima of the target image
Include: before difference
The target image is split, obtains center image block and multiple fields image block respectively;
The pixel grey scale mean value of each field image block in the multiple field image block is obtained, and according to each field
The average gray value of the multiple field image block of the pixel grey scale mean value computation of image block.
The local contrast enhancing image to the target image is filtered in one of the embodiments, really
Determining real goal region includes:
Threshold segmentation is carried out according to local contrast enhancing image of the preset threshold to the target image, obtains binary map
Picture;
Connected domain analysis is carried out to the bianry image, obtain the number of candidate target region in the bianry image and is divided
Cloth;
According to the number and distribution of the candidate target region, candidate target regions multiple in the bianry image are carried out
Time domain association analysis obtains the real goal region in multiple candidate targets.
It is described in one of the embodiments, that image is enhanced according to local contrast of the preset threshold to the target image
Threshold segmentation is carried out, obtaining bianry image includes:
Enhance image according to the local contrast of the target image, obtains the part of the local contrast enhancing image
The mean value and standard deviation of contrast;
Enhance the mean value and standard deviation of the local contrast of image according to the local contrast, determines to the target figure
The local contrast enhancing image of picture carries out the preset threshold of Threshold segmentation.
The number and distribution according to the candidate target region in one of the embodiments, to the binary map
Multiple candidate target regions carry out time domain association analysis as in, and the real goal region obtained in multiple candidate targets includes:
According to the motion continuity of adjacent interframe and data correlation degree, gone from the multiple candidate target region unless mesh
Region is marked, the real goal region is obtained.
The motion continuity and data correlation degree according to adjacent interframe in one of the embodiments, from described more
Nontarget area is removed in a candidate target region, obtaining the real goal region includes:
If continuous three frame only detects a candidate target region, and between two continuous frames candidate target region center
Positional distance is less than presetted pixel threshold value, then confirms that the candidate target is the real goal region;
If first frame detects multiple candidate target regions, each candidate target region that the second frame is detected, point
Each candidate region for not detecting with first frame carries out the calculating of Euclidean distance and local contrast, by Euclidean distance it is minimum and
The most similar candidate region of local contrast is as the real goal region.
It is described in one of the embodiments, that connected domain analysis is carried out to the bianry image, obtain the bianry image
The number of middle candidate target region and distribution include:
Clustering is carried out to the bianry image, obtains the number and mark of candidate target region in the bianry image
Note.
A kind of object detecting device, described device include:
Enhance image collection module, for obtaining target image, according to the Local Minimum contrast of the target image and
The local maxima difference of the target image handles the target image, obtains the local contrast of the target image
Enhance image;
Real goal area determination module is filtered for the local contrast enhancing image to the target image,
Determine real goal region.
A kind of computer equipment, including memory and processor, the memory are stored with computer program, the processing
Device realizes the step of as above any one the method when executing the computer program.
A kind of computer readable storage medium, is stored thereon with computer program, and the computer program is held by processor
The step of as above described in any item methods are realized when row.
Above-mentioned object detection method, device, computer equipment and storage medium, by obtaining target image, according to described
The Local Minimum contrast of target image and the local maxima difference of the target image handle the target image, obtain
Local contrast to the target image enhances image;And then the local contrast of target image enhancing image is carried out
Filtering, determines real goal region.By the above method, the interference of complex scene can be reduced, improves Dim targets detection rate,
Reduce false-alarm.
Detailed description of the invention
Fig. 1 is a kind of applied environment figure of object detection method in one embodiment;
Fig. 2 is a kind of flow diagram of object detection method in one embodiment;
Fig. 3 is the flow diagram of step S1 in another embodiment;
Fig. 4 is the flow diagram of step S11 in another embodiment;
Fig. 5 is the flow diagram of step S2 in another embodiment;
Fig. 6 is the flow diagram of step S21 in another embodiment;
Fig. 7 is the flow diagram of step S231 in another embodiment;
Fig. 8 (a) is the hollow middle scene figure of another embodiment;
Fig. 8 (b) is to enhance image using the aerial scene that existing local contrast method obtains in another embodiment;
Fig. 8 (c) is aerial after improving optimization in another embodiment to local contrast using the application method
Scene enhances image;
Fig. 9 (a) is The great ocean stretches away to meet the sky scene in another embodiment;
Fig. 9 (b) is that the Hai Tian after optimization is improved to local contrast using the application method in another embodiment
The scene that connects enhances image;
Figure 10 (a) is Ocean Scenes in another embodiment;
Figure 10 (b) is that the sea after optimization is improved to local contrast using the application method in another embodiment
Scene enhances image;
Figure 11 is a kind of structural block diagram of object detecting device in one embodiment;
Figure 12 is the internal structure chart of computer equipment in one embodiment.
Specific embodiment
It is with reference to the accompanying drawings and embodiments, right in order to which the objects, technical solutions and advantages of the application are more clearly understood
The application is further elaborated.It should be appreciated that specific embodiment described herein is only used to explain the application, not
For limiting the application.
A kind of object detection method provided by the present application, can be applied in application environment as shown in Figure 1.Wherein, eventually
End 102 is communicated by network with server 104.Terminal 102 obtains target image, and the target image is transmitted to clothes
Business device 104, server 104 is according to the Local Minimum contrast of the target image and the local maxima difference of the target image
The target image is handled, the local contrast enhancing image of the target image is obtained;Later again to the target
The local contrast enhancing image of image is filtered, and determines real goal region.Wherein, terminal 102 can be, but not limited to be
Various personal computers, laptop, smart phone, tablet computer and portable wearable device, server 104 can be used
The server cluster of independent server either multiple servers composition is realized.
In one embodiment, as shown in Fig. 2, providing a kind of object detection method, it is applied in Fig. 1 in this way
It is illustrated for terminal, comprising the following steps:
Step S1: obtaining target image, according to the Local Minimum contrast of the target image and the target image
Local maxima difference handles the target image, obtains the local contrast enhancing image of the target image.
Specifically, target image is complex scene image, including The great ocean stretches away to meet the sky scene, Ocean Scenes and high-altitude scene
Deng.For The great ocean stretches away to meet the sky and Ocean Scenes, by the interference of fish scale light reflection and sea clutter, the local contrast of target image is opposite
It is low, the target image need to be handled by Local Minimum contrast and local maximum difference, to promote target image
Local contrast.Further, since some atural objects have similar radiation intensity in visible light, near-infrared or middle infrared band, when
When atural object in target image with similar radiation intensity compares concentration, the low contrast in target image is also resulted in.
The local contrast enhancing image of target image refer to target image overall contrast is divided into it is multiple continuous small
Block region, the contrast of these pockets are exactly the local contrast of target image, and contrast enhancing is by target image
In range of luminance values stretch or be compressed into the specified brightness display range of display, to improve image all or the comparison of part
Degree.
Step S2: the local contrast enhancing image of the target image is filtered, determines real goal region.
Specifically, really target area refers to the Weak target in target image, wherein Weak target can be infrared small and weak
Target or near-infrared Weak target etc..Example, real goal region are the setting sun in the target image that The great ocean stretches away to meet the sky, in target image
Background area occupy significant portion, and interference is generated to real goal region.
Above-mentioned object detection method, device, computer equipment and storage medium, by obtaining target image, according to described
The Local Minimum contrast of target image and the local maxima difference of the target image handle the target image, obtain
Local contrast to the target image enhances image;And then the local contrast of target image enhancing image is carried out
Filtering, determines real goal region.By the above method, the interference of complex scene can be reduced, improves Dim targets detection rate,
Reduce false-alarm.
In one embodiment, include: in conjunction with Fig. 3, the step S1
Step S11: obtaining the maximum value of center image block pixel, and according to the maximum value of the center image block pixel and
The average gray value of multiple fields image block obtains the local maxima difference of the target image, wherein the neighborhood image block
It is adjacent with the center image block.
Specifically, center image block refers to the minimum image block comprising target image central point.Neighborhood image block feeling the pulse with the finger-tip is marked on a map
Each remaining image block of center image block is removed as in.
If center block pixel maximum value is Ln, all neighborhood blocks mean value be mI, target image local maxima difference ZI,
Specific calculating process is as follows:
ZI=Ln-mI。
Step S12: according to the pixel grey scale mean value of each field image block in multiple fields image block and the center
The maximum value of image block pixel obtains the Local Minimum contrast of the target image.
Specifically, the Local Minimum contrast C of target imagewAre as follows:
Wherein, LnFor the maximum value of window center sub-block pixel, miFor the pixel mean value of i-th of neighboring sub-patch, i value is
1、2、…8。
Step S13: according to the local maxima difference of the Local Minimum contrast of the target image and the target image,
Obtain the local contrast enhancing image of the target image.
Specifically, Local Minimum contrast C is improvedwWith local maximum difference ZtSquare product, the part enhanced
Contrast image is denoted as EELCM:
In one embodiment, include: before in conjunction with Fig. 4, the step S11
Step S9: the target image is split, and obtains center image block and multiple fields image block respectively.
Specifically, target image is split and refers to geometry segmentation, i.e., be divided into target image multiple equal or unequal
Image block.Target image is carried out equal part by example, forms N × N image block, sets 3 for N, each tile size is 3*3, i.e.,
Obtain a center image block and 8 neighborhood image blocks.
Step S10: obtaining the pixel grey scale mean value of each field image block in the multiple field image block, and according to
The average gray value of the multiple field image block of the pixel grey scale mean value computation of each field image block.
Specifically, the center image block of above-mentioned acquisition being denoted as " 0 ", remaining neighborhood image block is denoted as " 1 "-" 8 " respectively,
Represent 8 neighborhoods of center image block.
The pixel grey scale mean value m of i-th of neighborhood blockiAre as follows:
NbIndicate the pixel number of each image subblock,For the gray scale of j-th of pixel in i-th of neighborhood block
Value, i value be 1,2 ... 8.
Calculate the average gray value m of 8 neighborhood blocksI:
In one embodiment, include: in conjunction with Fig. 5, the step S2
Step S21: Threshold segmentation is carried out according to local contrast enhancing image of the preset threshold to the target image, is obtained
Obtain bianry image.
Specifically, preset threshold, which refers to, carries out selected by Threshold segmentation the local contrast enhancing image of the target image
Threshold value.In order to adapt to different complex scene variations, preset threshold is adaptively chosen according to the statistical information of contrast value.
Step S22: connected domain analysis is carried out to the bianry image, obtains candidate target region in the bianry image
Number and distribution.
Specifically, the distribution of candidate target region includes region corresponding label in position and its position in bianry image.
Wherein, coordinate can be used in position or numeric form is recorded.
The application carries out connected domain analysis to the bianry image, mainly uses clustering method.In practical engineering applications,
Consider for real-time, since connected domain analysis is marked by traversal entire image, time complexity is higher, is unfavorable for
Embedded platform is realized.Therefore, for speed up processing, the present invention will be made using clustering method apart from nearest white area block
For class object block one by one.Since cluster is only calculated on target area, for carrying out operation in entire image,
Time complexity substantially reduces.
Step S23: according to the number and distribution of the candidate target region, to multiple candidate targets in the bianry image
Region carries out time domain association analysis, obtains the real goal region in multiple candidate targets.
Specifically, time domain association analysis includes closing to the motion continuity and data of the adjacent interframe of multiple candidate target regions
The analysis of connection, to confirm final real goal region.Time domain association is carried out to candidate target regions multiple in bianry image
The purpose of analysis is further to further confirm that candidate target region, to obtain real target area, to improve inspection
The accuracy of survey.
In one embodiment, include: in conjunction with Fig. 6, the step S21
Step S211: enhancing image according to the local contrast of the target image, obtains the local contrast enhancing
The mean value and standard deviation of the local contrast of image.
Specifically, the acquisition modes of the local contrast of image of mean value and standard deviation enhance to(for) local contrast are very
It is more, can between using mean value and standard deviation calculation formula, can also directly be realized using software programming.
Example, here are as follows using the code for calculating mean value and standard deviation using meanStdDev:
Step S212: enhance the mean value and standard deviation of the local contrast of image, determination pair according to the local contrast
The local contrast enhancing image of the target image carries out the preset threshold of Threshold segmentation.
Specifically, the calculation of preset threshold is as follows:
Thr=mu+k*sigma
Wherein mu, sigma are respectively local contrast mean value and variance, and preset threshold range is between 1~3, according to reality
Depending on the scene of border, present specification is set as 2.
In one embodiment, the step S22 includes:
Step S221: clustering is carried out to the bianry image, obtains candidate target region in the bianry image
Number and label.
Specifically, for label is opposite candidate target region position, as the label of candidate target region position is respectively
Number 1,2,3 etc., also using the forms such as text or figure record.
In one embodiment, the step S23 includes:
Step S231: according to the motion continuity of adjacent interframe and data correlation degree, from the multiple candidate target region
Middle removal nontarget area obtains the real goal region.
Specifically, false-alarm (nontarget area) is further removed according to the motion continuity of adjacent interframe and data correlation,
Confirm real goal region.Under normal conditions, the contrast of target area is higher.However, the complicated multiplicity of actual scene, it is possible to
It will appear the case where local contrast of background block is higher than the local contrast of object block, particularly with The great ocean stretches away to meet the sky and sea
Scape, by the interference of fish scale light reflection and sea clutter, the local contrast of target area is likely lower than the contrast of sea clutter, uses
After above-mentioned contrast enhancing, sea clutter is possible to be taken as candidate target region in background, comes to real object tape great
Interference.Therefore, in order to improve the accuracy of detection, it is necessary to further candidate target region be further confirmed that, to obtain
Real target area.
On the other hand, due to the randomness that background occurs, and target has motion continuity between consecutive frame.Therefore,
Consideration further confirms target using interframe continuity and data correlation.
In one embodiment, include: in conjunction with Fig. 7, the step S231
Step S2311: if continuous three frame only detects a candidate target region, and candidate target between two continuous frames
The center distance in region is less than presetted pixel threshold value, then confirms that the candidate target is the real goal region.
Specifically, presetted pixel threshold value refers to the center distance of candidate target region between two continuous frames, presetted pixel
Threshold value can be configured according to the actual situation, and the presetted pixel threshold value in the application is set as 10 pixels.
Step S2312: if first frame detects multiple candidate target regions, the second frame each of is detected into candidate mesh
Region is marked, each candidate region detected respectively with first frame carries out the calculating of Euclidean distance and local contrast, will be European
Distance is minimum and the most similar candidate region of local contrast is as the real goal region.
Specifically, the application is calculated using standard European distance, due to candidate target region be it is two-dimensional, then often
The Euclidean distance of a candidate region may be expressed as:
Wherein, ρ is point (x2, y2) and point (x1, y1) between Euclidean distance;| X | it is point (x2, y2) to origin Euclidean away from
From.
In order to verify present specification method to the validity of practical different scenes small IR targets detection, using reality
Contextual data is tested, including aerial scene, scene that The great ocean stretches away to meet the sky and Ocean Scenes, as a result as seen in figs. 8-10.Fig. 8
It (a) is aerial scene, background has cloud noise, and Fig. 8 (b) is the aerial scene enhancing obtained using existing local contrast method
Image, Fig. 8 (c) are to improve the aerial scene after optimization to local contrast using the application method to enhance result.Fig. 9
It (a) is The great ocean stretches away to meet the sky scene, target is faint, and Fig. 9 (b) is after improving optimization to local contrast using the application method
Scene that The great ocean stretches away to meet the sky enhances result;Figure 10 (a) is Ocean Scenes, and sea clutter interference, Figure 10 (b) is to be played a game using the application method
Portion's contrast improves the Ocean Scenes enhancing result after optimization.
As can be seen that can enhance target using local contrast method from Fig. 8 (b) and 8 (c), but use existing
In the enhancing result that method obtains other than target is enhanced, the higher marginal portion of brightness (the cloud layer side in such as figure in background
Edge) it is same enhanced, more false-alarm is brought to subsequent progress Threshold segmentation and detection;The increasing obtained using Fig. 8 (c) method
Strong background edge (cloud layer edge) is inhibited while target is enhanced in strong result, greatly reduces false-alarm.
In addition, by Fig. 9 (b) and 10 (b) it is found that can be carried out to the sea clutter in Sea background while enhancing target
Inhibit, suitable for extremely faint target.
In order to which the beneficial effect of the application method and the prior art is quantitatively evaluated, using verification and measurement ratio and false alarm rate two indices
To compare the method for the present invention (being denoted as EELCM), existing local contrast calculation method (abbreviation ELCM) and morphological method
The performance of (abbreviation ETH), the results are shown in Table 1.As can be seen from the table, compared with prior art, the method for the present invention is being kept
While compared with high detection rate, false alarm rate is substantially reduced, and improves the adaptability and reliability of product.
Table 1
It should be understood that although each step in the flow chart of Fig. 2 is successively shown according to the instruction of arrow, this
A little steps are not that the inevitable sequence according to arrow instruction successively executes.Unless expressly state otherwise herein, these steps
It executes there is no the limitation of stringent sequence, these steps can execute in other order.Moreover, at least part in Fig. 2
Step may include that perhaps these sub-steps of multiple stages or stage are executed in synchronization to multiple sub-steps
It completes, but can execute at different times, the execution sequence in these sub-steps or stage, which is also not necessarily, successively to be carried out,
But it can be executed in turn or alternately at least part of the sub-step or stage of other steps or other steps.
In one embodiment, as shown in figure 11, a kind of object detecting device is provided, comprising: enhancing image obtains mould
Block and real goal area determination module, in which:
Enhance image collection module 10, for obtaining target image, according to the Local Minimum contrast of the target image
The target image is handled with the local maxima difference of the target image, obtains the local contrast of the target image
Degree enhancing image;
Real goal area determination module 20 was carried out for the local contrast enhancing image to the target image
Filter, determines real goal region.
In one embodiment, the enhancing image collection module 10 includes:
Maximum difference obtains module 11, for obtaining the maximum value of center image block pixel, and according to the center image
The maximum value of block pixel and the average gray value of multiple fields image block, obtain the local maxima difference of the target image,
In, the neighborhood image block is adjacent with the center image block;
Local Minimum contrast obtains module 12, for the picture according to each field image block in multiple fields image block
The maximum value of plain gray average and the center image block pixel obtains the Local Minimum contrast of the target image;
Local contrast enhance image collection module 13, for according to the target image Local Minimum contrast and institute
The local maxima difference for stating target image obtains the local contrast enhancing image of the target image.
In one embodiment, include: before the maximum difference obtains module 11
Target image is split module 14, for the target image to be split, obtains center image block respectively
With multiple fields image block;
Average gray value computing module 15, for obtaining the picture of each field image block in the multiple field image block
Plain gray average, and according to the average gray of the multiple field image block of the pixel grey scale mean value computation of each field image block
Value.
In one embodiment, the real goal area determination module 20 includes:
Bianry image obtains module 21, for enhancing image according to local contrast of the preset threshold to the target image
Threshold segmentation is carried out, bianry image is obtained;
Connected domain analysis module 22 obtains in the bianry image for carrying out connected domain analysis to the bianry image
The number and distribution of candidate target region;
Time domain association analysis module 23, for the number and distribution according to the candidate target region, to the binary map
Multiple candidate target regions carry out time domain association analysis as in, obtain the real goal region in multiple candidate targets.
In one embodiment, the bianry image acquisition module 21 includes:
Mean value and standard deviation obtain module 211, for enhancing image according to the local contrast of the target image, obtain
The mean value and standard deviation of the local contrast of the local contrast enhancing image;
Threshold segmentation module 212, the mean value and mark of the local contrast for enhancing image according to the local contrast
It is quasi- poor, determine the preset threshold that Threshold segmentation is carried out to the local contrast enhancing image of the target image.
In one embodiment, the time domain association analysis module 23 includes:
Real goal region obtains module 231, for the motion continuity and data correlation degree according to adjacent interframe, from institute
It states in multiple candidate target regions and removes nontarget area, obtain the real goal region.
In one embodiment, the real goal region acquisition module 231 includes:
First detection module 2311, if only detecting a candidate target region, and two continuous frames for continuous three frame
Between candidate target region center distance be less than presetted pixel threshold value, then confirm the candidate target be the true mesh
Mark region;
Second detection module 2312 detects the second frame if detecting multiple candidate target regions for first frame
Each candidate target region, each candidate region for detecting respectively with first frame carries out Euclidean distance and local contrast
It calculates, using Euclidean distance minimum and the most similar candidate region of local contrast is as the real goal region.
In one embodiment, the connected domain analysis module 22 includes:
Number and label obtain module 221, for carrying out clustering to the bianry image, obtain the bianry image
The number and label of middle candidate target region.
Specific about a kind of object detecting device limits the limit that may refer to above for a kind of object detection method
Fixed, details are not described herein.Modules in a kind of above-mentioned object detecting device can fully or partially through software, hardware and its
Combination is to realize.Above-mentioned each module can be embedded in the form of hardware or independently of in the processor in computer equipment, can also be with
It is stored in the memory in computer equipment in a software form, in order to which processor calls the above modules of execution corresponding
Operation.
In one embodiment, a kind of computer equipment is provided, which can be server, internal junction
Composition is shown in Fig.12.The computer equipment include by system bus connect processor, memory, network interface and
Database.Wherein, the processor of the computer equipment is for providing calculating and control ability.The memory packet of the computer equipment
Include non-volatile memory medium, built-in storage.The non-volatile memory medium is stored with operating system, computer program and data
Library.The built-in storage provides environment for the operation of operating system and computer program in non-volatile memory medium.The calculating
The database of machine equipment is for storing target detection data.The network interface of the computer equipment is used to pass through with external terminal
Network connection communication.To realize a kind of object detection method when the computer program is executed by processor.
It will be understood by those skilled in the art that structure shown in Figure 12, only part relevant to application scheme
The block diagram of structure, does not constitute the restriction for the computer equipment being applied thereon to application scheme, and specific computer is set
Standby may include perhaps combining certain components or with different component layouts than more or fewer components as shown in the figure.
In one embodiment, a kind of computer equipment, including memory and processor are provided, is stored in memory
Computer program, the processor realize step described in any one method as described above when executing computer program.
In one embodiment, a kind of computer readable storage medium is provided, computer program is stored thereon with, is calculated
Step described in any one method as described above is realized when machine program is executed by processor.
Those of ordinary skill in the art will appreciate that realizing all or part of the process in above-described embodiment method, being can be with
Relevant hardware is instructed to complete by computer program, the computer program can be stored in a non-volatile computer
In read/write memory medium, the computer program is when being executed, it may include such as the process of the embodiment of above-mentioned each method.Wherein,
To any reference of memory, storage, database or other media used in each embodiment provided herein,
Including non-volatile and/or volatile memory.Nonvolatile memory may include read-only memory (ROM), programming ROM
(PROM), electrically programmable ROM (EPROM), electrically erasable ROM (EEPROM) or flash memory.Volatile memory may include
Random access memory (RAM) or external cache.By way of illustration and not limitation, RAM is available in many forms,
Such as static state RAM (SRAM), dynamic ram (DRAM), synchronous dram (SDRAM), double data rate sdram (DDRSDRAM), enhancing
Type SDRAM (ESDRAM), synchronization link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM
(RDRAM), direct memory bus dynamic ram (DRDRAM) and memory bus dynamic ram (RDRAM) etc..
Each technical characteristic of above embodiments can be combined arbitrarily, for simplicity of description, not to above-described embodiment
In each technical characteristic it is all possible combination be all described, as long as however, the combination of these technical characteristics be not present lance
Shield all should be considered as described in this specification.
The several embodiments of the application above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the concept of this application, various modifications and improvements can be made, these belong to the protection of the application
Range.Therefore, the scope of protection shall be subject to the appended claims for the application patent.
Claims (11)
1. a kind of object detection method, which is characterized in that the described method includes:
Target image is obtained, according to the local maxima difference of the Local Minimum contrast of the target image and the target image
The target image is handled, the local contrast enhancing image of the target image is obtained;
The local contrast enhancing image of the target image is filtered, determines real goal region.
2. the method according to claim 1, wherein the acquisition target image, according to the target image
Local Minimum contrast and the local maxima difference of the target image handle the target image, obtain the target
The local contrast of image enhances image
The maximum value of center image block pixel is obtained, and according to the maximum value and multiple fields image of the center image block pixel
The average gray value of block obtains the local maxima difference of the target image, wherein the neighborhood image block and the Centered Graphs
As block is adjacent;
According to the pixel grey scale mean value of each field image block in multiple fields image block and the center image block pixel
Maximum value obtains the Local Minimum contrast of the target image;
According to the local maxima difference of the Local Minimum contrast of the target image and the target image, the target is obtained
The local contrast of image enhances image.
3. according to the method described in claim 2, it is characterized in that, the maximum value for obtaining the center image block pixel,
And according to the average gray value of the maximum value of the center image block pixel and the multiple field image block, the target is obtained
Include: before the local maxima difference of image
The target image is split, obtains center image block and multiple fields image block respectively;
The pixel grey scale mean value of each field image block in the multiple field image block is obtained, and according to each field image
The average gray value of the multiple field image block of the pixel grey scale mean value computation of block.
4. according to the method described in claim 3, it is characterized in that, the local contrast to the target image enhances figure
As being filtered, determine that real goal region includes:
Threshold segmentation is carried out according to local contrast enhancing image of the preset threshold to the target image, obtains bianry image;
Connected domain analysis is carried out to the bianry image, obtains the number and distribution of candidate target region in the bianry image;
According to the number and distribution of the candidate target region, time domain is carried out to candidate target regions multiple in the bianry image
Association analysis obtains the real goal region in multiple candidate targets.
5. according to the method described in claim 4, it is characterized in that, it is described according to preset threshold to the part of the target image
Contrast enhances image and carries out Threshold segmentation, obtains bianry image and includes:
Enhance image according to the local contrast of the target image, obtains the local contrast of the local contrast enhancing image
The mean value and standard deviation of degree;
Enhance the mean value and standard deviation of the local contrast of image according to the local contrast, determines to the target image
Local contrast enhances the preset threshold that image carries out Threshold segmentation.
6. according to the method described in claim 4, it is characterized in that, the number according to the candidate target region and point
Cloth carries out time domain association analysis to candidate target regions multiple in the bianry image, obtains true in multiple candidate targets
Target area includes:
According to the motion continuity of adjacent interframe and data correlation degree, non-target area is removed from the multiple candidate target region
Domain obtains the real goal region.
7. according to the method described in claim 6, it is characterized in that, described close according to the motion continuity and data of adjacent interframe
Connection degree removes nontarget area from the multiple candidate target region, obtains the real goal region and includes:
If continuous three frame only detects a candidate target region, and between two continuous frames candidate target region center
Distance is less than presetted pixel threshold value, then confirms that the candidate target is the real goal region;
If first frame detects multiple candidate target regions, each candidate target region that the second frame is detected, respectively with
Each candidate region that first frame detects carries out the calculating of Euclidean distance and local contrast, and Euclidean distance is minimum and local
The most similar candidate region of contrast is as the real goal region.
8. according to the method described in claim 4, it is characterized in that, it is described to the bianry image carry out connected domain analysis, obtain
Into the bianry image, the number of candidate target region and distribution include:
Clustering is carried out to the bianry image, obtains the number and label of candidate target region in the bianry image.
9. a kind of object detecting device, which is characterized in that described device includes:
Enhance image collection module, for obtaining target image, according to the Local Minimum contrast of the target image and described
The local maxima difference of target image handles the target image, obtains the local contrast enhancing of the target image
Image;
Real goal area determination module is filtered for the local contrast enhancing image to the target image, determines
Real goal region.
10. a kind of computer equipment, including memory and processor, the memory are stored with computer program, feature exists
In the step of processor realizes any one of claims 1 to 8 the method when executing the computer program.
11. a kind of computer readable storage medium, is stored thereon with computer program, which is characterized in that the computer program
The step of method described in any item of the claim 1 to 8 is realized when being executed by processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910091468.9A CN110033474B (en) | 2019-01-30 | 2019-01-30 | Target detection method, target detection device, computer equipment and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910091468.9A CN110033474B (en) | 2019-01-30 | 2019-01-30 | Target detection method, target detection device, computer equipment and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN110033474A true CN110033474A (en) | 2019-07-19 |
CN110033474B CN110033474B (en) | 2021-09-03 |
Family
ID=67235513
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910091468.9A Active CN110033474B (en) | 2019-01-30 | 2019-01-30 | Target detection method, target detection device, computer equipment and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110033474B (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113629574A (en) * | 2021-08-18 | 2021-11-09 | 国网湖北省电力有限公司襄阳供电公司 | Strong wind sand area transmission conductor galloping early warning system based on video monitoring technology |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101620728A (en) * | 2009-06-19 | 2010-01-06 | 北京航空航天大学 | New infrared background inhibiting method based on self-adaption background forecast |
CN101893580A (en) * | 2010-06-10 | 2010-11-24 | 北京交通大学 | Digital image based detection method of surface flaw of steel rail |
CN102682296A (en) * | 2012-03-21 | 2012-09-19 | 北京航空航天大学 | Self-adaption estimating method of size of infrared small dim target under complicated background condition |
CN103996209A (en) * | 2014-05-21 | 2014-08-20 | 北京航空航天大学 | Infrared vessel object segmentation method based on salient region detection |
CN104834915A (en) * | 2015-05-15 | 2015-08-12 | 中国科学院武汉物理与数学研究所 | Small infrared object detection method in complex cloud sky background |
US20150344942A1 (en) * | 2012-10-17 | 2015-12-03 | Spatial Transcriptomics Ab | Methods and product for optimising localised or spatial detection of gene expression in a tissue sample |
CN106056115A (en) * | 2016-05-25 | 2016-10-26 | 西安科技大学 | Infrared small target detection method under non-uniform background |
CN107590496A (en) * | 2017-09-18 | 2018-01-16 | 南昌航空大学 | The association detection method of infrared small target under complex background |
CN107992873A (en) * | 2017-10-12 | 2018-05-04 | 西安天和防务技术股份有限公司 | Object detection method and device, storage medium, electronic equipment |
CN109191395A (en) * | 2018-08-21 | 2019-01-11 | 深圳创维-Rgb电子有限公司 | Method for enhancing picture contrast, device, equipment and storage medium |
-
2019
- 2019-01-30 CN CN201910091468.9A patent/CN110033474B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101620728A (en) * | 2009-06-19 | 2010-01-06 | 北京航空航天大学 | New infrared background inhibiting method based on self-adaption background forecast |
CN101893580A (en) * | 2010-06-10 | 2010-11-24 | 北京交通大学 | Digital image based detection method of surface flaw of steel rail |
CN102682296A (en) * | 2012-03-21 | 2012-09-19 | 北京航空航天大学 | Self-adaption estimating method of size of infrared small dim target under complicated background condition |
US20150344942A1 (en) * | 2012-10-17 | 2015-12-03 | Spatial Transcriptomics Ab | Methods and product for optimising localised or spatial detection of gene expression in a tissue sample |
CN103996209A (en) * | 2014-05-21 | 2014-08-20 | 北京航空航天大学 | Infrared vessel object segmentation method based on salient region detection |
CN104834915A (en) * | 2015-05-15 | 2015-08-12 | 中国科学院武汉物理与数学研究所 | Small infrared object detection method in complex cloud sky background |
CN106056115A (en) * | 2016-05-25 | 2016-10-26 | 西安科技大学 | Infrared small target detection method under non-uniform background |
CN107590496A (en) * | 2017-09-18 | 2018-01-16 | 南昌航空大学 | The association detection method of infrared small target under complex background |
CN107992873A (en) * | 2017-10-12 | 2018-05-04 | 西安天和防务技术股份有限公司 | Object detection method and device, storage medium, electronic equipment |
CN109191395A (en) * | 2018-08-21 | 2019-01-11 | 深圳创维-Rgb电子有限公司 | Method for enhancing picture contrast, device, equipment and storage medium |
Non-Patent Citations (3)
Title |
---|
FATEMEH NOURI ET.AL: "Salient object detection using local, global and high contrast graphs", 《SIGNAL, IMAGE AND VIDEO PROCESSING》 * |
姚朝霞 等: "基于局部对比度测量的红外弱小目标恒虚警检测", 《红外技术》 * |
孟博: "空中红外小目标检测及硬件加速研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113629574A (en) * | 2021-08-18 | 2021-11-09 | 国网湖北省电力有限公司襄阳供电公司 | Strong wind sand area transmission conductor galloping early warning system based on video monitoring technology |
Also Published As
Publication number | Publication date |
---|---|
CN110033474B (en) | 2021-09-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhong et al. | Detecting camouflaged object in frequency domain | |
CN109035304B (en) | Target tracking method, medium, computing device and apparatus | |
Walia et al. | Recent advances on multicue object tracking: a survey | |
Seo et al. | Nonparametric bottom-up saliency detection by self-resemblance | |
Liu et al. | Aggregation signature for small object tracking | |
CN109903272A (en) | Object detection method, device, equipment, computer equipment and storage medium | |
CN108765460B (en) | Hyperspectral image-based space-time joint anomaly detection method and electronic equipment | |
CN111445496B (en) | Underwater image recognition tracking system and method | |
Mo et al. | PVDet: Towards pedestrian and vehicle detection on gigapixel-level images | |
Ma et al. | Layn: Lightweight multi-scale attention yolov8 network for small object detection | |
Xu et al. | Feature extraction algorithm of basketball trajectory based on the background difference method | |
Zhao et al. | ST-YOLOA: a Swin-transformer-based YOLO model with an attention mechanism for SAR ship detection under complex background | |
Huang et al. | Robust simultaneous localization and mapping in low‐light environment | |
CN112686122B (en) | Human body and shadow detection method and device, electronic equipment and storage medium | |
Zhang et al. | Target tracking for mobile robot platforms via object matching and background anti-matching | |
CN110033474A (en) | Object detection method, device, computer equipment and storage medium | |
US20140376822A1 (en) | Method for Computing the Similarity of Image Sequences | |
Li et al. | [Retracted] Image Processing and Recognition Algorithm Design in Intelligent Imaging Device System | |
CN116805327A (en) | Infrared small target tracking method based on dynamic convolution kernel | |
Liu et al. | [Retracted] Mean Shift Fusion Color Histogram Algorithm for Nonrigid Complex Target Tracking in Sports Video | |
Gu et al. | Infrared-visible synthetic data from game engine for image fusion improvement | |
Patil et al. | Detection and tracking of moving object: A survey | |
Du et al. | Saliency Computation for Virtual Cinematography in 360 Videos | |
CN112580475A (en) | Face detection method and system | |
Liu et al. | Distractor-aware tracking with multi-task and dynamic feature learning |
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 |