CN105184809A - Moving object detection method and moving object detection device - Google Patents
Moving object detection method and moving object detection device Download PDFInfo
- Publication number
- CN105184809A CN105184809A CN201410226777.XA CN201410226777A CN105184809A CN 105184809 A CN105184809 A CN 105184809A CN 201410226777 A CN201410226777 A CN 201410226777A CN 105184809 A CN105184809 A CN 105184809A
- Authority
- CN
- China
- Prior art keywords
- subimage block
- image
- pending
- prospect profile
- block
- 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.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention provides a moving object detection method and a moving object detection device. The method comprises the steps: obtaining a foreground outline of an outline image of a current frame image, and generating a foreground outline image, wherein the foreground outline image is a differential image between the outline image of the current frame image and a background outline image corresponding to a last frame image; selecting sub-image blocks one by one in the foreground outline image according to a preset interval; judging whether the sub-image blocks comprise foreground outline pixels or not; enabling the sub-image blocks comprising the foreground outline pixels to serve as to-be-processed sub-image blocks according to judgment results, or else skipping the sub-image blocks; calculating the directional gradient histogram characteristics of each to-be-processed sub-image block; and employing a classifier to detect the calculated directional gradient histogram characteristics, so as to determine a moving object. The method provided by the invention can improve the detection speed and accuracy.
Description
Technical field
The present invention relates to technical field of image processing, in particular to Moving Objects detection method and Moving Objects pick-up unit.
Background technology
Along with the fast development of intelligent transportation system (ITS), the moving vehicle detection algorithm of view-based access control model becomes a basic technology of computer vision research.The application of intelligent transportation system has a lot, such as automatic incident detection (AID), and vehicle flow is added up, collecting vehicle information, vehicle tracking, and moving vehicle detection is crucial.
There is the moving vehicle detection method of some view-based access control model at present, as background subtraction, optical flow method and frame difference method:
Background subtraction uses gauss hybrid models, and Gauss model, Kalman filter or other statistical model extract the background image of scene, and extract moving region according to the difference image of background image and present frame.But time in the process extracting background image and in the scene of vehicle at this background image, this method can bring the problem of duplicate detection.Meanwhile, the brightness change of scene is another serious problem.
Optical flow method to use the Optic flow information of moving region to detect moving region, and its calculated amount is very large.
Frame difference method extracts moving region based on the difference of present frame and former frame or a few frame.This method can the environment of promptly Adaptive change, and the region of extracting may not be so accurate, particularly for the target of movement fast.
Feature extraction and sorting technique, by feature extraction and classification to detect vehicle.This method even can detect vehicle in single image.But calculated amount is excessive, and detection perform is determined by the quality of data set.
Therefore, need a kind of new Moving Objects detection method, to solve the large and problem that Detection accuracy is low of calculated amount existing for current detection method.
Summary of the invention
In view of this, the present invention proposes a kind of new Moving Objects detection technique, with the problem that the accuracy of detection at least solving existing target detection technique is low.
According to an aspect of the present invention, propose a kind of Moving Objects detection method, comprise: obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, described prospect profile image is the difference image of the background profile image that the contour images of described current frame image is corresponding with its previous frame image; In described prospect profile image, subimage block is chosen one by one according to predetermined interval; Judge whether described subimage block comprises prospect profile pixel; According to judged result, will the subimage block of described prospect profile pixel be comprised as pending subimage block, otherwise skip described subimage block; Calculate the histograms of oriented gradients feature of each pending subimage block; Sorter is adopted to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
According to a further aspect in the invention, additionally provide a kind of Moving Objects pick-up unit, comprise: prospect profile image generation unit, obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, described prospect profile image is the difference image of the background profile image that the contour images of described current frame image is corresponding with its previous frame image; Choose unit, in described prospect profile image, choose subimage block according to predetermined interval one by one; Pending subimage block determining unit, judges whether described subimage block comprises prospect profile pixel, according to judged result, will comprise the subimage block of described prospect profile pixel as pending subimage block, otherwise skips described subimage block; Computing unit, calculates the histograms of oriented gradients feature of each pending subimage block; Detecting unit, adopts sorter to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
According to a further aspect of the invention, additionally provide a kind of electronic equipment, this electronic equipment comprises Moving Objects pick-up unit as above.
According to a further aspect of the invention, additionally provide a kind of program product storing the instruction code of machine-readable, said procedure product can make above-mentioned machine perform Moving Objects detection method as above when performing.
In addition, according to other aspects of the invention, additionally provide a kind of computer-readable recording medium, it stores program product as above.
The above-mentioned Moving Objects pick-up unit according to the embodiment of the present invention, Moving Objects detection method and electronic equipment, the subimage block comprising prospect profile pixel in Utilization prospects contour images is to detect Moving Objects, can at least realize one of following beneficial effect: subimage block comprises prospect profile pixel, the precision of detection can be made higher; Decrease the check processing process of irrelevant subimage block in prospect profile image, calculated amount is less; And power consumption is lower.
By below in conjunction with the detailed description of accompanying drawing to most preferred embodiment of the present invention, these and other advantage of the present invention will be more obvious.
Accompanying drawing explanation
The present invention can be better understood by reference to hereinafter given by reference to the accompanying drawings description, wherein employs same or analogous Reference numeral in all of the figs to represent identical or similar parts.Described accompanying drawing comprises in this manual together with detailed description below and forms the part of this instructions, and is used for illustrating the preferred embodiments of the present invention further and explaining principle and advantage of the present invention.In the accompanying drawings:
Fig. 1 shows the schematic flow sheet of Moving Objects detection method according to an embodiment of the invention;
Fig. 2 shows the schematic flow sheet of extraction prospect profile according to another embodiment of the invention;
Fig. 3 shows contours extract schematic diagram according to an embodiment of the invention;
Fig. 4 shows picture size according to an embodiment of the invention adjustment schematic diagram;
Fig. 5 shows the schematic diagram of histograms of oriented gradients characteristic image according to an embodiment of the invention;
Fig. 6 shows the treatment scheme schematic diagram of the pending subimage block of selection in correlation technique;
Fig. 7 shows the treatment scheme schematic diagram of the pending subimage block of selection according to an embodiment of the invention;
Fig. 8 shows the partition structure schematic diagram of pending subimage according to an embodiment of the invention;
Fig. 9 shows the block diagram of Moving Objects pick-up unit according to an embodiment of the invention.
Embodiment
In order to more clearly understand above-mentioned purpose of the present invention, feature and advantage, below in conjunction with the drawings and specific embodiments, the present invention is further described in detail.It should be noted that, when not conflicting, the feature in the embodiment of the application and embodiment can combine mutually.
Set forth a lot of detail in the following description so that fully understand the present invention, but the present invention can also adopt other to be different from other modes described here and implement, and therefore, the present invention is not limited to the restriction of following public specific embodiment.
Fig. 1 shows the process flow diagram of Moving Objects detection method according to an embodiment of the invention.
As shown in Figure 1, Moving Objects detection method mainly comprises following two large steps according to an embodiment of the invention:
Step 102, obtains the prospect profile in the contour images of current frame image, and generate prospect profile image, wherein, prospect profile image is the difference image of the background profile image that the contour images of current frame image is corresponding with its previous frame image.
How the contour images of background extraction contour images and current frame image will elaborate hereinafter.
Step 104, the Moving Objects based on sorter detects.
At step 104, first adjusted size is carried out, to meet related request to the prospect profile image obtained.Then feature preextraction is carried out.Finally originally be the key component of embodiment: in prospect profile image, choose subimage block according to predetermined interval one by one.Generally, this predetermined interval is the distance between adjacent two described subimage block centers.Certainly, also can suitable predetermined interval be set according to actual needs.
Judge whether subimage block comprises prospect profile pixel.According to judged result, will the subimage block of prospect profile pixel be comprised as pending subimage block, otherwise skip subimage block.In the present embodiment, only check processing is carried out to the subimage block including prospect profile pixel, the subimage block not comprising prospect profile pixel is then ignored, so just can avoid the process to the subimage block of Moving Objects can not be had to process, can calculated amount be reduced on the one hand, improve detection speed, on the other hand, Detection accuracy can be improved, because handled subimage block all includes foreground pixel.
Calculate the histograms of oriented gradients feature of each pending subimage block.Sorter is adopted to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
A specific implementation process of above-mentioned steps 102 is described in detail below in conjunction with Fig. 2.
As shown in Figure 2, step 202, obtains the multiple image contour images separately of catching for region to be detected.Above-mentioned multiple image can be obtained in advance by first-class photography of making a video recording, picture pick-up device.Region to be detected can be certain room, certain place, or can be certain road, etc.In one example in which, region to be detected can be certain predetermined road, the maybe segment path of this predetermined road.In actual applications, such as, the image pickup scope of first-class photography of such as making a video recording, picture pick-up device can be set to above-mentioned region to be detected accordingly, so that catch the image in above-mentioned region to be detected.
In a kind of implementation of Moving Objects detection method according to an embodiment of the invention, the contour images of every two field picture in above-mentioned multiple image can be obtained by gradient algorithm.
In one example in which, for each pixel in the every two field picture in above-mentioned multiple image, the profile of this two field picture can be obtained by the contour detecting device of such as 5 × 5, complete contouring process.
For certain pixel p (x, y) in certain two field picture in above-mentioned multiple image, Fig. 3 A-3D gives the example for the 4 kinds of contour detecting devices calculating pixel p (x, y).
In Fig. 3 A-3D, the pixel of black is p (x, y), and for calculating profile value, Fig. 3 A-3D respectively illustrates for differently contoured type of detection 20 pixels (pixel of white and the pixel of grey) around it.Use often kind of contour detecting device to calculate the difference of the absolute average of white pixel and gray pixels, wherein, the profile value of what difference was maximum be selected as p (x, y).For each pixel in the every two field picture in above-mentioned multiple image all can adopt as above method to calculate the profile value of this pixel.Like this, after obtaining the profile value of all pixels of every two field picture, represent this pixel by the profile value of each pixel, can obtain the contour images of this two field picture, its gray-scale value is 0 ~ 225.
Step 204, after completing contours extract, performs background profile information updating.If present frame is the first frame, the contour images then extracted will be saved using as initial background contour images, otherwise, the intensity of the intensity of each pixel in the contour images of this two field picture except prospect profile to that pixel corresponding with this location of pixels in the environment contour images corresponding to the previous frame image of this two field picture can be utilized to upgrade, and utilize the intensity of the intensity of each pixel in the prospect profile in the contour images of this two field picture to that pixel corresponding with this location of pixels in the environment contour images corresponding to the previous frame image of this two field picture to upgrade.In this implementation, update process can be completed according to formula one.
Formula one:
CB(x,y,t
k)＝CB(x,y,t
k-1)+(b×(1-Seg(x,y,t
k))+a×Seg(x,y,t
k))×(CC(x,y,t
k)-CB(x,y,t
k-1))
Wherein, a is default foreground pixel turnover rate, CB (x, y, t
k-1) expression capture time is t
k-1environment contour images corresponding to image in the intensity of pixel (x, y); CC (x, y, t
k) expression capture time is t
kimage contour images in the intensity of pixel (x, y); X, y represent the position of pixel (x, y) in corresponding image respectively; B is default, corresponding with the pixel in contour images except prospect profile turnover rate (referred to as background pixel turnover rate).In addition, Seg (x, y, t
k) can with reference to the definition of formula four.It should be noted that capture time is t
k-1image be capture time be t
kthe previous frame image of image, and i>=2.Wherein, capture time is t
1environment contour images corresponding to image in intensity CB (x, y, the t of pixel (x, y)
1) to equal capture time be t
1image contour images in intensity CC (x, y, the t of pixel (x, y)
1).
Step 206, after background profile upgrades, obtains the prospect profile image in the contour images of current frame image.Difference image, this difference image and the prospect profile image of the contour images of the i-th two field picture environment contour images corresponding with the i-th-1 two field picture can be obtained according to formula two.
Wherein, CB (x, y, t
k-1) expression capture time is t
k-1environment contour images corresponding to image in the intensity of pixel (x, y), CC (x, y, t
k) expression capture time is t
kimage contour images in the intensity of pixel (x, y).
In the difference image of the contour images of this two field picture background profile image corresponding with its previous frame image, intensity is defined as foreground pixel higher than those pixels of dynamic threshold corresponding to this two field picture, and the profile formed by determined all foreground pixels is as the prospect profile in the contour images of this two field picture.
Like this, by the setting of dynamic threshold, pixel that can be lower by those intensity in difference image excludes from prospect profile, thus makes the pixel in the prospect profile determined (i.e. foreground pixel) can get rid of the interference of neighbourhood noise better.
Wherein, for the every two field picture in above-mentioned multiple image, dynamic threshold corresponding to this two field picture such as can be determined according to the difference image of the contour images of this two field picture background profile image corresponding with its previous frame image.
In one example in which, following formula three can be adopted to calculate above-mentioned dynamic threshold:
Thre(t
k)＝mean(CD(t
k))+Wegt×(max(CD(t
k))-mean(CD(t
k))+std(CD(t
k)))
Wherein, Thre (t
k) expression capture time is t
kthe dynamic threshold corresponding to image.Mean (CD (x, y, t
k)) expression capture time is t
kimage in corresponding CD (x, y, the t of all pixels
k) mean value, max (CD (x, y, t
k)) expression capture time is t
iimage in corresponding CD (x, y, the t of all pixels
k) in maximal value, std (CD (x, y, t
k)) expression capture time is t
iimage in corresponding CD (x, y, the t of all pixels
k) standard deviation, and Wegt is default weighted value.
Step 208, in a kind of implementation of Moving Objects detection method according to an embodiment of the invention, the prospect profile that can be obtained according to formula four splits from the contour images of corresponding two field picture.
Formula four:
Wherein, Seg (x, y, t
kit is t that)=1 corresponds to capture time
kimage (such as the contour images of kth two field picture) contour images in prospect profile in pixel (i.e. foreground pixel), Seg (x, y, t
kit is t that)=0 corresponds to capture time
kimage (such as the contour images of kth two field picture) contour images in pixel except prospect profile.
Following detailed description detects based on the Moving Objects of sorter.
Adjusted size is carried out to the prospect profile image (CMI image) of the image of catching (CF image) and acquisition, its size is reduced.Such as, if CF image is of a size of M × N, then the height after adjustment is M × z, and width is N × z, wherein, and 0≤z≤0.Before and after the adjusted size of CF image and CF image, comparison diagram is respectively see Fig. 4 A and Fig. 4 B.
After the size that have adjusted CF image and CMI image, carry out the preextraction of feature.
In the present embodiment, use histograms of oriented gradients feature in feature extraction, this part is the preparation process of feature extraction.In this step, absolute gradient figure and the angle figure of the prospect profile image after adjusted size is calculated as follows:
Vertical gradient value and horizontal gradient value is calculated respectively according to formula five and formula six.
Formula five:
grad
v(x,y)＝pixel(x,y+1)-pixel(x,y-1)
Formula six:
grad
h(x,y)＝pixel(x+1,y)-pixel(x-1,y)
Wherein, grad
v(x, y) and grad
h(x, y) is vertical gradient value and the horizontal gradient value of pixel (x, y) respectively.
Calculate absolute gradient figure according to formula seven and calculate angular divisions figure according to formula eight, Fig. 5 shows the absolute gradient schematic diagram calculated.
Formula seven:
Formula eight:
Wherein, grad (x, y) is the Grad of pixel (x, y), angle (x, y) be the angular divisions value of pixel (x, y), angel (x, y) ∈ [1, BinNum], BinNum is angular divisions threshold value, in the present embodiment BinNum=9.
Next illustrate how selection process is carried out to prospect profile image, obtain multiple pending subimage block in conjunction with example.
In traditional detection of classifier processing procedure, step is as shown in Figure 6 adopted to detect, each subimage block (image block that detected frame is chosen) has default size, in the present embodiment, subimage block is of a size of Bh × Bw, move this detection block according to predetermined space, sorter detects each subimage block that detection block is chosen.As can be seen from Figure 6, some unnecessary subimage blocks are selected equally, such as third step, the subimage block chosen in n-th step and m step, the central area of these subimage blocks does not comprise prospect profile pixel,, uncared-for subimage block does not comprise vehicle ' s contour substantially as can be seen from figure also, therefore these subimage blocks can be left in the basket.Therefore, this disposal route, except adding unnecessary calculated amount, also can cause detecting mistake.
In order to solve the above-mentioned technical matters in traditional detection processing procedure, have employed a kind of check processing method (be the image of catching see Fig. 7 A to Fig. 7 B, Fig. 7 A, Fig. 7 B is prospect profile image) of improvement in this example.
In fig. 7, be pending subimage block by the subimage block that the first choice box 30 is chosen, this pending subimage block is used for the detection of sorter in the future, will be left in the basket by the subimage block that the second choice box 32 is chosen.As shown in Figure 7 B, if comprise at least one prospect profile pixel in the predeterminable area in the subimage block in prospect profile image, then this subimage block can be selected and in subsequent detection process, otherwise skip this subimage block, move choice box with predetermined interval, continue to judge whether next son image block comprises prospect profile pixel.
In one example in which, this predeterminable area is the central area of subimage block, the size of this central area can be that the ratio of arbitrary dimension and shared subimage block is not unique yet, in this example, this central area is preferably dimensioned to be 3 × 3 pixels, and the center pixel in this central area is (0.67 × Bh, the 0.67 × Bw) pixel in subimage block.
Therefore the selection processing procedure of improvement provided by the invention have selected the subimage block including prospect profile pixel, ignore the subimage block not having prospect profile pixel, therefore, relative to classic method, this selection disposal route at least eliminates the calculation processes for the subimage block not having prospect profile pixel, decreases calculated amount, and comprises prospect profile pixel due to the subimage block chosen, therefore can improve Detection accuracy, reduce detection error rate.
After selecting multiple pending subimage block, calculate the histograms of oriented gradients feature of each pending subimage block, in this step the absolute gradient figure using previous calculations to go out and angular divisions figure is carried out calculated direction histogram of gradients feature.
First, adjust the size of the subimage block that each selects according to the size of the training sample of sorter, such as, if the size of training sample is 64 × 64, so the size of subimage block is also applied as 64 × 64, and namely Bh and Bw is the length of 64 pixels.
Suppose in this example, the subimage block selected is of a size of 64 × 64, and each subimage block is broken down into multiple cellular, and as shown in Figure 8, then each cellular is of a size of 8 pixel × 8 pixels.In order to calculate HOG feature, N number of cellular is combined into a characteristic block, and M characteristic block is combined into pending subimage block; Calculate the absolute gradient histogram of each cellular, the absolute gradient histogram of all cellulars in characteristic block is merged, form a subcharacter; The subcharacter of all characteristic blocks in pending subimage block is cascaded, obtains histograms of oriented gradients feature and use feature.In this example, N is 4, M is 4.
Wherein, the histogram feature of the absolute gradient of each cellular is calculated according to formula nine.
Wherein, CelHis is the absolute gradient histogram of cellular.
Fig. 9 shows the block diagram of Moving Objects pick-up unit according to an embodiment of the invention.
As shown in Figure 9, Moving Objects pick-up unit 900 according to an embodiment of the invention, can comprise:
Prospect profile image generation unit 902, obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, prospect profile image is the difference image of the background profile image that the contour images of current frame image is corresponding with its previous frame image;
Choose unit 904, in prospect profile image, choose subimage block according to predetermined interval one by one;
Pending subimage block determining unit 906, judges whether subimage block comprises prospect profile pixel, according to judged result, will comprise the subimage block of prospect profile pixel as pending subimage block, otherwise skips subimage block;
Computing unit 908, calculates the histograms of oriented gradients feature of each pending subimage block;
Detecting unit 910, adopts sorter to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
Wherein, pending subimage block determining unit 906 comprises:
Judging unit 9062, judge whether the predeterminable area place in the subimage block chosen comprises at least one prospect profile pixel, if predeterminable area place comprises at least one prospect profile pixel, then using subimage block as pending subimage block, otherwise, skip subimage block.
Predetermined interval is the distance between adjacent two subimage block centers.The size of the pending subimage block chosen is determined according to the size of the training sample of sorter.
Moving Objects pick-up unit 900 can also comprise: resize unit 912, before selecting multiple pending subimage block, according to the full-size of motion object outline in prospect profile image and the size of pending subimage block, adjusted size is carried out to prospect profile image.
Wherein, computing unit 908 comprises: division unit 9082, and each pending subimage block is divided into multiple cellular, and N number of cellular is combined into a characteristic block, and M characteristic block is combined into pending subimage block; Fusion Features unit 9084, calculate the absolute gradient histogram of each cellular, the absolute gradient histogram of all cellulars in characteristic block is merged, form a subcharacter, and the subcharacter of all characteristic blocks in pending subimage block is cascaded, obtain histograms of oriented gradients feature.
Utilization prospects contour images of the present invention, and propose choose the method that subimage block carries out Moving Objects detection in prospect profile image, filter out the subimage block with prospect profile pixel, have ignored the process that the subimage block not comprising prospect profile pixel is processed, thus minimizing calculated amount, improve computing velocity, and all comprise prospect profile pixel due to the subimage block chosen, therefore improve the accuracy rate detecting Moving Objects.
In addition, embodiments of the invention additionally provide a kind of electronic equipment, and this electronic equipment comprises Moving Objects pick-up unit as above.In the specific implementation of above-mentioned according to an embodiment of the invention electronic equipment, above-mentioned electronic equipment can be any one equipment in following equipment: computing machine; Panel computer; Personal digital assistant; Multimedia play equipment; Mobile phone and electric paper book etc.Wherein, this electronic equipment has the above-mentioned various function for Moving Objects pick-up unit and technique effect, repeats no more here.
Each component units, subelement, module etc. in the above-mentioned pick-up unit of Moving Objects according to an embodiment of the invention can be configured by the mode of software, firmware, hardware or its combination in any.When being realized by software or firmware, to the machine with specialized hardware structure, the program forming this software or firmware can be installed from storage medium or network, this machine, when being provided with various program, can perform the various functions of above-mentioned each component units, subelement.
In addition, the invention allows for a kind of program product storing the instruction code of machine-readable.When above-mentioned instruction code is read by machine and performs, the above-mentioned detection method of Moving Objects according to an embodiment of the invention can be performed.Correspondingly, the various storage mediums for the such as disk, CD, magneto-optic disk, semiconductor memory etc. that carry this program product are also included within of the present invention disclosing.
In addition, the method for various embodiments of the present invention is not limited to describe the to specifications or time sequencing shown in accompanying drawing performs, also can according to other time sequencing, perform concurrently or independently.Therefore, the execution sequence of the method described in this instructions is not construed as limiting technical scope of the present invention.
In addition, obviously, also can realize in the mode being stored in the computer executable program in various machine-readable storage medium according to each operating process of said method of the present invention.
And, object of the present invention also can be realized by following manner: the storage medium storing above-mentioned executable program code is supplied to system or equipment directly or indirectly, and computing machine in this system or equipment or CPU (central processing unit) (CPU) read and perform said procedure code.
Now, as long as this system or equipment have the function of executive routine, then embodiments of the present invention are not limited to program, and this program also can be arbitrary form, such as, the program that performs of target program, interpreter or be supplied to the shell script etc. of operating system.
These machinable mediums above-mentioned include but not limited to: various storer and storage unit, semiconductor equipment, and disc unit is light, magnetic and magneto-optic disk such as, and other is suitable for the medium etc. of storage information.
The foregoing is only the preferred embodiments of the present invention, be not limited to the present invention, for a person skilled in the art, the present invention can have various modifications and variations.Within the spirit and principles in the present invention all, any amendment done, equivalent replacement, improvement etc., all should be included within protection scope of the present invention.
To sum up, in an embodiment according to the present invention, the invention provides following scheme but be not limited thereto:
Remarks 1. 1 kinds of Moving Objects detection methods, comprising:
Obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, described prospect profile image is the difference image of the background profile image that the contour images of described current frame image is corresponding with its previous frame image;
In described prospect profile image, subimage block is chosen one by one according to predetermined interval;
Judge whether described subimage block comprises prospect profile pixel;
According to judged result, will the subimage block of described prospect profile pixel be comprised as pending subimage block, otherwise skip described subimage block;
Calculate the histograms of oriented gradients feature of each pending subimage block;
Sorter is adopted to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
The Moving Objects detection method of remarks 2. according to remarks 1, wherein, describedly judges whether described subimage block comprises prospect profile pixel, comprising:
Judge whether the predeterminable area place in the subimage block chosen comprises at least one prospect profile pixel;
If described predeterminable area place comprises at least one prospect profile pixel, then using described subimage block as described pending subimage block, otherwise, skip described subimage block.
The Moving Objects detection method of remarks 3. according to remarks 1, described predetermined interval is the distance between adjacent two described subimage block centers.
The Moving Objects detection method of remarks 4. according to remarks 1, determines the size of the pending subimage block chosen according to the size of the training sample of described sorter.
The Moving Objects detection method of remarks 5. according to remarks 1, before selecting multiple described pending subimage block, also comprises:
According to the full-size of motion object outline in described prospect profile image and the size of described pending subimage block, adjusted size is carried out to described prospect profile image.
The Moving Objects detection method of remarks 6. according to any one of remarks 1 to 5, the histograms of oriented gradients feature of each pending subimage block of described calculating, comprising:
Subimage block pending described in each is divided into multiple cellular, and N number of cellular is combined into a characteristic block, and M described characteristic block is combined into described pending subimage block;
Calculate the absolute gradient histogram of each cellular, the absolute gradient histogram of all cellulars in described characteristic block is merged, form a subcharacter;
The subcharacter of all characteristic blocks in described pending subimage block is cascaded, obtains described histograms of oriented gradients feature.
Remarks 7. 1 kinds of Moving Objects pick-up units, comprising:
Prospect profile image generation unit, obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, described prospect profile image is the difference image of the background profile image that the contour images of described current frame image is corresponding with its previous frame image;
Choose unit, in described prospect profile image, choose subimage block according to predetermined interval one by one;
Pending subimage block determining unit, judges whether described subimage block comprises prospect profile pixel, according to judged result, will comprise the subimage block of described prospect profile pixel as pending subimage block, otherwise skips described subimage block;
Computing unit, calculates the histograms of oriented gradients feature of each pending subimage block;
Detecting unit, adopts sorter to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
The Moving Objects pick-up unit of remarks 8. according to remarks 7, wherein, described pending subimage block determining unit comprises:
Judging unit, judge whether the predeterminable area place in the subimage block chosen comprises at least one prospect profile pixel, if described predeterminable area place comprises at least one prospect profile pixel, then using described subimage block as described pending subimage block, otherwise, skip described subimage block.
The Moving Objects pick-up unit of remarks 9. according to remarks 7, described predetermined interval is the distance between adjacent two described subimage block centers.
The Moving Objects pick-up unit of remarks 10. according to remarks 7, determines the size of the pending subimage block chosen according to the size of the training sample of described sorter.
The Moving Objects pick-up unit of remarks 11. according to remarks 7, also comprise: resize unit, before selecting multiple described pending subimage block, according to the full-size of motion object outline in described prospect profile image and the size of described pending subimage block, adjusted size is carried out to described prospect profile image.
The Moving Objects pick-up unit of remarks 12. according to any one of remarks 7 to 11, described computing unit comprises:
Division unit, is divided into multiple cellular by subimage block pending described in each, and N number of cellular is combined into a characteristic block, and M described characteristic block is combined into described pending subimage block;
Fusion Features unit, calculate the absolute gradient histogram of each cellular, the absolute gradient histogram of all cellulars in described characteristic block is merged, form a subcharacter, and the subcharacter of all characteristic blocks in described pending subimage block is cascaded, obtain described histograms of oriented gradients feature.
Remarks 13. 1 kinds of electronic equipments, comprise the Moving Objects pick-up unit according to any one of remarks 7-12.
Claims (10)
1. a Moving Objects detection method, comprising:
Obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, described prospect profile image is the difference image of the background profile image that the contour images of described current frame image is corresponding with its previous frame image;
In described prospect profile image, subimage block is chosen one by one according to predetermined interval;
Judge whether described subimage block comprises prospect profile pixel;
According to judged result, will the subimage block of described prospect profile pixel be comprised as pending subimage block, otherwise skip described subimage block;
Calculate the histograms of oriented gradients feature of each pending subimage block;
Sorter is adopted to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
2. Moving Objects detection method according to claim 1, wherein, describedly judges whether described subimage block comprises prospect profile pixel, comprising:
Judge whether the predeterminable area place in the subimage block chosen comprises at least one prospect profile pixel;
If described predeterminable area place comprises at least one prospect profile pixel, then using described subimage block as described pending subimage block, otherwise, skip described subimage block.
3. Moving Objects detection method according to claim 1, determines the size of the pending subimage block chosen according to the size of the training sample of described sorter.
4. Moving Objects detection method according to claim 1, before selecting multiple described pending subimage block, also comprises:
According to the full-size of motion object outline in described prospect profile image and the size of described pending subimage block, adjusted size is carried out to described prospect profile image.
5. Moving Objects detection method according to any one of claim 1 to 4, the histograms of oriented gradients feature of each pending subimage block of described calculating, comprising:
Subimage block pending described in each is divided into multiple cellular, and N number of cellular is combined into a characteristic block, and M described characteristic block is combined into described pending subimage block;
Calculate the absolute gradient histogram of each cellular, the absolute gradient histogram of all cellulars in described characteristic block is merged, form a subcharacter;
The subcharacter of all characteristic blocks in described pending subimage block is cascaded, obtains described histograms of oriented gradients feature.
6. a Moving Objects pick-up unit, comprising:
Prospect profile image generation unit, obtain the prospect profile in the contour images of current frame image, generate prospect profile image, wherein, described prospect profile image is the difference image of the background profile image that the contour images of described current frame image is corresponding with its previous frame image;
Choose unit, in described prospect profile image, choose subimage block according to predetermined interval one by one;
Pending subimage block determining unit, judges whether described subimage block comprises prospect profile pixel, according to judged result, will comprise the subimage block of described prospect profile pixel as pending subimage block, otherwise skips described subimage block;
Computing unit, calculates the histograms of oriented gradients feature of each pending subimage block;
Detecting unit, adopts sorter to detect the histograms of oriented gradients feature calculated, to determine Moving Objects.
7. Moving Objects pick-up unit according to claim 6, wherein, described pending subimage block determining unit comprises:
Judging unit, judge whether the predeterminable area place in the subimage block chosen comprises at least one prospect profile pixel, if described predeterminable area place comprises at least one prospect profile pixel, then using described subimage block as described pending subimage block, otherwise, skip described subimage block.
8. Moving Objects pick-up unit according to claim 6, determines the size of the pending subimage block chosen according to the size of the training sample of described sorter.
9. Moving Objects pick-up unit according to claim 6, also comprise: resize unit, before selecting multiple described pending subimage block, according to the full-size of motion object outline in described prospect profile image and the size of described pending subimage block, adjusted size is carried out to described prospect profile image.
10. the Moving Objects pick-up unit according to any one of claim 6 to 9, described computing unit comprises:
Division unit, is divided into multiple cellular by subimage block pending described in each, and N number of cellular is combined into a characteristic block, and M described characteristic block is combined into described pending subimage block;
Fusion Features unit, calculate the absolute gradient histogram of each cellular, the absolute gradient histogram of all cellulars in described characteristic block is merged, form a subcharacter, and the subcharacter of all characteristic blocks in described pending subimage block is cascaded, obtain described histograms of oriented gradients feature.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410226777.XA CN105184809A (en) | 2014-05-26 | 2014-05-26 | Moving object detection method and moving object detection device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410226777.XA CN105184809A (en) | 2014-05-26 | 2014-05-26 | Moving object detection method and moving object detection device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105184809A true CN105184809A (en) | 2015-12-23 |
Family
ID=54906862
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410226777.XA Pending CN105184809A (en) | 2014-05-26 | 2014-05-26 | Moving object detection method and moving object detection device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105184809A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169918A (en) * | 2016-03-08 | 2017-09-15 | 株式会社理光 | For determining the mobile mesh calibration method in target zone, device and system |
| CN110598785A (en) * | 2019-09-11 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Training sample image generation method and device |
| US11631768B2 (en) | 2016-03-25 | 2023-04-18 | Taiwan Semiconductor Manufacturing Co., Ltd. | Semiconductor device and method of manufacturing thereof |
| CN116434127A (en) * | 2023-06-14 | 2023-07-14 | 季华实验室 | Human body posture estimation method, device, equipment and storage medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080025568A1 (en) * | 2006-07-20 | 2008-01-31 | Feng Han | System and method for detecting still objects in images |
| JP2011013838A (en) * | 2009-06-30 | 2011-01-20 | Sumitomo Electric Ind Ltd | Mobile object detection device, mobile object detection method and computer program |
| US20120274755A1 (en) * | 2011-04-29 | 2012-11-01 | Tata Consultancy Services Limited | System and method for human detection and counting using background modeling, hog and haar features |
| CN103106409A (en) * | 2013-01-29 | 2013-05-15 | 北京交通大学 | Composite character extraction method aiming at head shoulder detection |
| CN103390167A (en) * | 2013-07-18 | 2013-11-13 | 奇瑞汽车股份有限公司 | Multi-characteristic layered traffic sign identification method |
| CN103473571A (en) * | 2013-09-12 | 2013-12-25 | 天津大学 | Human detection method |
-
2014
- 2014-05-26 CN CN201410226777.XA patent/CN105184809A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080025568A1 (en) * | 2006-07-20 | 2008-01-31 | Feng Han | System and method for detecting still objects in images |
| JP2011013838A (en) * | 2009-06-30 | 2011-01-20 | Sumitomo Electric Ind Ltd | Mobile object detection device, mobile object detection method and computer program |
| US20120274755A1 (en) * | 2011-04-29 | 2012-11-01 | Tata Consultancy Services Limited | System and method for human detection and counting using background modeling, hog and haar features |
| CN103106409A (en) * | 2013-01-29 | 2013-05-15 | 北京交通大学 | Composite character extraction method aiming at head shoulder detection |
| CN103390167A (en) * | 2013-07-18 | 2013-11-13 | 奇瑞汽车股份有限公司 | Multi-characteristic layered traffic sign identification method |
| CN103473571A (en) * | 2013-09-12 | 2013-12-25 | 天津大学 | Human detection method |
Non-Patent Citations (4)
| Title |
|---|
| DALAL NARNEET 等: "Histograms of Oriented Gradients for Human Detection", 《PROCEEDINGS OF IEEE CONFERENCE ON COMPUTER VISION AND PATTERN ＲECOGNITION，SAN DIEGO》 * |
| 吴暾华 等: "平面旋转人脸检测与特征定位方法研究", 《电子学报》 * |
| 贾世杰 等: "基于改进塔式梯度直方图的行人检测算法", 《大连交通大学学报》 * |
| 陈国玉: "序列图像中运动目标检测算法研究", 《中国优秀硕士学位论文全文数据库-信息科技辑》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107169918A (en) * | 2016-03-08 | 2017-09-15 | 株式会社理光 | For determining the mobile mesh calibration method in target zone, device and system |
| CN107169918B (en) * | 2016-03-08 | 2020-08-04 | 株式会社理光 | Method, device and system for determining a moving object in an object range |
| US11631768B2 (en) | 2016-03-25 | 2023-04-18 | Taiwan Semiconductor Manufacturing Co., Ltd. | Semiconductor device and method of manufacturing thereof |
| CN110598785A (en) * | 2019-09-11 | 2019-12-20 | 腾讯科技（深圳）有限公司 | Training sample image generation method and device |
| CN110598785B (en) * | 2019-09-11 | 2021-09-07 | 腾讯科技（深圳）有限公司 | Training sample image generation method and device |
| CN116434127A (en) * | 2023-06-14 | 2023-07-14 | 季华实验室 | Human body posture estimation method, device, equipment and storage medium |
| CN116434127B (en) * | 2023-06-14 | 2023-11-07 | 季华实验室 | Human body posture estimation method, device, equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109753913B (en) | Multi-mode video semantic segmentation method with high calculation efficiency | |
| CN105260749B (en) | Real-time target detection method based on direction gradient binary pattern and soft cascade SVM | |
| CN104517095B (en) | A kind of number of people dividing method based on depth image | |
| CN103208008A (en) | Fast adaptation method for traffic video monitoring target detection based on machine vision | |
| CN109961037A (en) | A kind of examination hall video monitoring abnormal behavior recognition methods | |
| CN113076871A (en) | Fish shoal automatic detection method based on target shielding compensation | |
| CN111091023B (en) | Vehicle detection method and device and electronic equipment | |
| CN110781882A (en) | License plate positioning and identifying method based on YOLO model | |
| CN112464912B (en) | Robot end face detection method based on YOLO-RGGNet | |
| CN111931720B (en) | Method, apparatus, computer device and storage medium for tracking image feature points | |
| CN111161309B (en) | Searching and positioning method for vehicle-mounted video dynamic target | |
| CN111209858A (en) | Real-time license plate detection method based on deep convolutional neural network | |
| CN111008633A (en) | License plate character segmentation method based on attention mechanism | |
| CN111046746A (en) | License plate detection method and device | |
| CN112926461B (en) | Neural network training and driving control method and device | |
| CN105184809A (en) | Moving object detection method and moving object detection device | |
| CN109635667A (en) | A kind of vehicle detecting system based on Guided Faster-RCNN | |
| CN113850136A (en) | Yolov5 and BCNN-based vehicle orientation identification method and system | |
| CN110516527B (en) | Visual SLAM loop detection improvement method based on instance segmentation | |
| CN106558069A (en) | A kind of method for tracking target and system based under video monitoring | |
| CN114419102B (en) | Multi-target tracking detection method based on frame difference time sequence motion information | |
| CN113963333B (en) | Traffic sign board detection method based on improved YOLOF model | |
| CN112101113B (en) | Lightweight unmanned aerial vehicle image small target detection method | |
| CN110717910B (en) | CT image target detection method based on convolutional neural network and CT scanner | |
| CN105335683A (en) | Object detection method and object detection apparatus |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| AD01 | Patent right deemed abandoned |
Effective date of abandoning: 20181019 |
|
| AD01 | Patent right deemed abandoned |