CN104751487B - A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference - Google Patents
A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference Download PDFInfo
- Publication number
- CN104751487B CN104751487B CN201510134601.6A CN201510134601A CN104751487B CN 104751487 B CN104751487 B CN 104751487B CN 201510134601 A CN201510134601 A CN 201510134601A CN 104751487 B CN104751487 B CN 104751487B
- Authority
- CN
- China
- Prior art keywords
- plane
- pixel
- planes
- moving target
- image
- 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.)
- Active
Links
Abstract
A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference, methods described is to obtain adjacent two color image frame under scene to be measured, and red in extraction coloured image, green and blue color planes are simultaneously stored;According to six planes extracted, red, green and blue color planes are carried out into difference operation respectively using frame difference method, generate three width error images;The pixel for three width error images is calculated successively again;Binary conversion treatment finally is done to each pixel in the new error image of acquisition, so as to obtain the profile of moving target.This method is calculated by carrying out frame difference respectively to tri- planes of RGB, judge whether same coordinate pixel chromatic aberration is in the same direction, calculating is overlapped to chromatic aberration, so that moving target outliner pixel values form larger contrast with other area pixel values, it is achieved thereby that the accurate extraction to moving target, remains the complete information of moving target profile to greatest extent.
Description
Technical field
It is especially a kind of poor based on the plane discoloration frames of colour RGB tri- the present invention relates to a kind of detection method of moving target
Moving target detecting method.
Background technology
Video technique has vast potential for future development in scientific research and engineer applied field.In Computer Vision process
In, the detection and extraction of moving target are a key technologies.Moving object detection is in order to motion mesh with the purpose extracted
Mark is separated from background image, is the segmentation problem of moving target and background.
The top priority of Computer Vision is the target that motion is detected from sequence of video images, how quick effective
Background and target separate be current research focus and emphasis.Conventional detection algorithm has optical flow method, background at present
Difference method and frame-to-frame differences method.Wherein, optical flow method operand is larger, is not suitable for real-time processing;Background difference needs modeling, and to dynamic
State scene changes are more sensitive, parameter(Such as learning rate)Incorrect setting will directly affect whole structure;Frame-to-frame differences method is
A kind of method that difference is carried out in time series, due to that need not model, therefore real-time is best compared with other method.Example
Such as the paper that A.Lipton is delivered《Moving target classification and tracking from real time
video》Detailed elaboration has been carried out to gray-scale plane frame difference;Patent document《A kind of recognition methods of moving target, device》
(CN103826102A)Based on frame difference method, be combined with Background difference, by present frame model and with next two field picture
Difference operation is carried out to determine motion target area.But, the frame difference method used in above-mentioned document only carries out frame to gray-scale plane
Difference is calculated, because single plane each pixel span is small and is easily disturbed by external environment condition light brightness change so that
Movement destination image is difficult to be accurately distinguished with background image, easily causes the loss of effective information, increased complete extraction motion
The difficulty of objective contour, have impact on the accuracy of later stage Computer Vision.
Moving Object in Video Sequences detection is the difficult point in field of video image processing, in intelligent video monitoring application
In, computer needs to carry out acquired image quick and accurate treatment, and this process proposes higher wanting to software algorithm
Ask.Especially, dark small to monitor area or brightness changes frequently video monitoring scene, the accuracy of existing algorithm
It is difficult to meet actual requirement.
The content of the invention
For above-mentioned prior art exist problem, the present invention provide it is a kind of based on the plane discoloration frames of colored RGB tri- difference
Moving target detecting method.
Realize that the technical scheme of above-mentioned moving target detecting method is as follows.
A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference, it is realized by following steps:
(1)The is obtained according to the video content under scene to be measuredk-1Frame andkFrame scene coloured image,kIt is more than 1
Integer;
(2)To step(1)The for obtainingk-1Color image frame, extracts the red plane of the coloured imagef k-1 (x,y, r), green color planef k-1 (x,y,g) and blue color planesf k-1 (x,y,b), it is stored in memory body,x,yRepresent pixel point coordinates, r,g,bRed plane, green color plane and the blue color planes of coloured image, subscript are represented respectivelyk-1Represent video frame number;
(3)To step(1)The for obtainingkColor image frame, extracts the red plane of the coloured imagef k (x,y,r), it is green
Color planef k (x,y,g) and blue color planesf k (x,y,b), it is stored in memory body, subscriptkRepresent video frame number;
(4)According to step(2)And step(3)Six planes of middle extraction, using frame difference method bykFrame andk-1Frame figure
The red plane of picture, green color plane and blue color planes carry out difference operation, three width differential charts of the generation corresponding to the planes of RGB tri- respectively
Picture, respectivelyD k (x,y,r)、D k (x,y,g)、D k (x,y,b);Operation expression is:
(5)To meetingD k (x,y,r)、D k (x,y,g)、D k (x,y,b) it is simultaneously greater than 0 or the simultaneously pixel less than 0
Point is calculatedD k (x,y);C k (x,y) operation expression be:
In formula, | | expression takes absolute value, and Min { } represents the minimum value in each entry value in { } bracket;
(6)According to whether meetingD k (x,y,r)、D k (x,y,g)、D k (x,y,b) it is simultaneously greater than 0 or simultaneously less than 0
Condition, corresponding is calculated to each pixelD k (x,y);For qualified pixel,D k (x,y) operation expression
For:
In formula, δ for (0,3] between real number;For the pixel for not meeting Rule of judgment,D k (x,y) operation table reach
Formula is:
(7)To step(6)InD k (x,y) value is calculated new error imageE k (x,y);E k (x,y) operation table
It is up to formula:
In formula, floor [] function representation carries out downward rounding operation to the numerical value in bracket [];
(8)To step(7)The new error image for obtainingE k (x,y) in each pixel do binary conversion treatment, so as to obtain
The profile of moving target, binary conversion treatment operation expression is:
In formula,A k (x,y) it is image after binaryzation, T is threshold value.
Realize a kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference that the invention described above is provided
Technical scheme, compared with prior art, the present invention is calculated by the way that tri- planes of RGB are carried out with frame difference respectively, with same pixel
The size of point chromatic aberration determines whether to be foreground moving image;Meanwhile, by judging that same coordinate pixel chromatic aberration is
It is no in the same direction, eliminate external environment condition because brightness change and caused by influence;Finally, calculating is overlapped to chromatic aberration so that
Moving target outliner pixel values and other area pixel values form larger contrast, so as to moving target is carried out it is complete, accurately carry
Take, the complete information of moving target profile is remained to greatest extent.
Specific embodiment
Specific embodiment of the invention is will be apparent from below.
A kind of above-mentioned the provided moving target detecting method based on the plane discoloration frames of colored RGB tri- difference of the present invention is provided
Technical scheme, it be by following steps realize.
Step one, obtain the according to the video content under scene to be measuredk-1Frame andkFrame scene coloured image,kBe more than
1 integer;
Step 2, the obtained to step onek-1Color image frame, extracts the red plane of the coloured imagef k-1 (x,y,r), green color planef k-1 (x,y,g) and blue color planesf k-1 (x,y,b), it is stored in memory body,x,yRepresent that pixel is sat
Mark,r,g,bRed plane, green color plane and the blue color planes of coloured image, subscript are represented respectivelyk-1Represent video frame number;
Step 3, the obtained to step onekColor image frame, extracts the red plane of the coloured imagef k-1 (x,y, r), green color planef k-1 (x,y,g) and blue color planesf k-1 (x,y,b), it is stored in memory body, subscriptkRepresent video frame number;
Step 4, according to six planes extracted in step 2 and step 3, using frame difference method by kth frame and the frame of kth -1
The red plane of coloured image, green color plane and blue color planes carry out difference operation, three width of the generation corresponding to the planes of RGB tri- respectively
Error image, respectivelyD k (x,y,r)、D k (x,y,g)、D k (x,y,b);Operation expression is:
f k Image is coloured image,f k (x,y,r)、f k (x,y,g)Withf k (x,y,b)Forf k The RGB three of image is put down
The value in face, wherein each pixel is the integer more than zero;D k (x,y,r)、D k (x,y,g)、D k (x,y,b)Represent respectively color
The error image of the planes of color RGB tri-, the value of error image each pixel may be for just, it is also possible to is negative value.Below with the 2nd frame
With the 1st two field picture(I.e.kWhen=2)In take 3 coordinates(0,0)(1,1)(2,2)Corresponding pixel points concrete numerical value is calculated and illustrated.
False coordinate(0,0)Corresponding R planes f1(0,0,r)=1、f2(0,0,r)=200, G plane f1(0,0,g)=10、f2(0,0,g)=
160, B plane f1(0,0,b)=8、f2(0,0,b)=100;Coordinate(0,0)Corresponding R planes D2(0,0,r)=f2(0,0,r)-f1
(0,0,r)=200-1=199, G plane D2(0,0,g)=f2(0,0,g)-f1(0,0,g)=160-10=150, B plane D2(0,0,b)
=f2(0,0,b)-f1(0,0,b)=100-8=92.Coordinate(1,1)Corresponding R planes f1(1,1,r)=5、f2(1,1,r)=0, G put down
Face f1(1,1,g)=20、f2(1,1,g)=1, B plane f1(1,1,b)=40、f2(1,1,b)=3;Coordinate(1,1)Corresponding R planes D2
(1,1,r)=f2(1,1,r)-f1(1,1,r)=0-5=-5, G plane D2(1,1,g)=f2(1,1,g)-f1(1,1,g)=1-20=-19,
B planes D2(1,1,b)=f2(1,1,b)-f1(1,1,b)=3-40=-37.Coordinate(2,2)Corresponding R planes f1(2,2,r)=50、f2
(2,2,r)=0, G plane f1(2,2,g)=5、f2(2,2,g)=20, B plane f1(2,2,b)=3、f2(2,2,b)=30;Coordinate(2,
2)Corresponding R planes D2(2,2,r)=f2(2,2,r)-f1(2,2,r)=0-50=-50, G planes D2(2,2,g)=f2(2,2,g)-
f1(2,2,g)=20-5=15, B plane D2(2,2,b)=f2(2,2,b)-f1(2,2,b)=30-3=27.
Step 5, to meetingD k (x,y,r)、D k (x,y,g)、D k (x,y,b) it is simultaneously greater than 0 or simultaneously less than 0
Pixel is calculatedD k (x,y);C k (x,y) operation expression be:
In formula, | | expression takes absolute value, and Min { } represents the minimum value in each entry value in { } bracket;
For coordinate(0,0)Point, D2(0,0,r)=199、D2(0,0,g)=150、D2(0,0,b)=92 are all higher than 0, meet and sentence
Broken strip part, the < 199 of 0 <, 92 < 150, therefore Ck(0,0)Value is 92.For coordinate(1,1)Point, D2(1,1,r)=-5、D2(1,1,
g)=-19、D2(1,1,b)=-37 are respectively less than 0, meet Rule of judgment, -37 < -19 < -5 < 0, therefore Ck(1,1)Value is 5.For
Coordinate(2,2)Point, D2(2,2,r)=-50、D2(2,2,g)=15、D2(2,2,b)=27, do not meet Rule of judgment.
Step 6, according to whether meetingD k (x,y,r)、D k (x,y,g)、D k (x,y,b) it is simultaneously greater than 0 or simultaneously
Condition less than 0, calculates corresponding to each pixelD k (x,y);For qualified pixel,D k (x,y) computing
Expression formula is:
In formula, δ for (0,3] between real number;For the pixel for not meeting Rule of judgment,D k (x,y) operation table reach
Formula is:
For the pixel for meeting Rule of judgment in step 5, it is assumed that δ values are 2, coordinate(0,0)Point, D2(0,0)=199
+150+92-2×92=461-184=257;Coordinate(1,1)Point, D2(1,1)=5+19+37-2×5=51.For not meeting judgement
The pixel of condition, D2(2,2)=50+15+27=92.
Step 7, in step 6D k (x,y) value is calculated new error imageE k (x,y);E k (x,y) fortune
Operator expression formula is:
In formula, floor [] function representation carries out downward rounding operation to the numerical value in bracket [];
For coordinate(0,0)Point, D2(0,0)=257 > 255, therefore E2(0,0)=255;Coordinate(1,1)Point, D2(1,1)=51
< 255, therefore E2(1,1)=51;Coordinate(2,2)Point, D2(2,2)=92 < 255, therefore E2(2,2)=92.
Step 8, the new error image obtained to step 7E k (x,y) in each pixel do binary conversion treatment so that
The profile of moving target is obtained, binary conversion treatment operation expression is:
In formula,A k (x,y) it is image after binaryzation, T is threshold value.
Assuming that T is 100, for coordinate(0,0)Point, E2(0,0)=255 > 100, therefore A2(0,0)=1;Coordinate(1,1)Point, E2
(1,1)=51 < 100, therefore A2(1,1)=0;Coordinate(2,2)Point, E2(2,2)=92 < 100, therefore A2(2,2)=0.By abovementioned steps
Treatment, moving target outliner pixel values and other area pixel values form larger contrast, and threshold value T would be readily ascertainable so that
Ensure that carries out complete, accurate extraction to moving target, and the complete information of moving target profile is remained to greatest extent.
Claims (1)
1. it is a kind of based on the plane discoloration frames of colored RGB tri- difference moving target detecting method, it by following steps realize:
(1)The is obtained according to the video content under scene to be measuredk-1Frame andkFrame scene coloured image,kIt is the integer more than 1;
(2)To step(1)The color image frame of kth -1 of acquisition, extracts the red plane of the coloured imagef k-1 (x,y,r), it is green
Color planef k-1 (x,y,g) and blue color planesf k-1 (x,y,b), it is stored in memory body,x,yRepresent pixel point coordinates, r, g, b
Red plane, green color plane and the blue color planes of coloured image, subscript are represented respectivelyk-1Represent video frame number;
(3)To step(1)The for obtainingkColor image frame, extracts the red plane of the coloured imagef k (x,y,r), green it is flat
Facef k (x,y,g) and blue color planesf k (x,y,b), it is stored in memory body, subscriptkRepresent video frame number;
(4)According to step(2)And step(3)Six planes of middle extraction, using frame difference method bykFrame andk-1Two field picture
Red plane, green color plane and blue color planes carry out difference operation respectively, and generation corresponds to three width error images of the planes of RGB tri-,
RespectivelyD k (x,y,r)、D k (x,y,g)、D k (x,y,b);Operation expression is:
(5)To meetingD k (x,y,r)、D k (x,y,g)、D k (x,y,b) it is simultaneously greater than 0 or simultaneously the pixel meter less than 0
CalculateC k (x,y);C k (x,y) operation expression be:
In formula, | | expression takes absolute value, and Min { } represents the minimum value in each entry value in { } bracket;
(6)According to whether meetingD k (x,y,r)、D k (x,y,g)、D k (x,y,b) it is simultaneously greater than 0 or the simultaneously bar less than 0
Part, calculates corresponding to each pixelD k (x,y);For qualified pixel,D k (x,y) operation expression be:
In formula, δ for (0,3] between real number;For the pixel for not meeting Rule of judgment,D k (x,y) operation expression
For:
(7)To step(6)InD k (x,y) value is calculated new error imageE k (x,y);E k (x,y) operation expression
For:
In formula, floor [] function representation carries out downward rounding operation to the numerical value in bracket [];
(8)To step(7)The new error image for obtainingE k (x,y) in each pixel do binary conversion treatment, so as to be moved
The profile of target, binary conversion treatment operation expression is:
In formula,A k (x,y) it is image after binaryzation, T is threshold value.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510134601.6A CN104751487B (en) | 2015-03-26 | 2015-03-26 | A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510134601.6A CN104751487B (en) | 2015-03-26 | 2015-03-26 | A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104751487A CN104751487A (en) | 2015-07-01 |
CN104751487B true CN104751487B (en) | 2017-06-06 |
Family
ID=53591106
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510134601.6A Active CN104751487B (en) | 2015-03-26 | 2015-03-26 | A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104751487B (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105205835A (en) * | 2015-09-02 | 2015-12-30 | 湖南工业大学 | Colored moving target recognizing and tracking method based on Fourier descriptor |
CN109766831A (en) * | 2019-01-09 | 2019-05-17 | 深圳市三宝创新智能有限公司 | A kind of road colour band recognition methods, device, computer equipment and storage medium |
CN109887045B (en) * | 2019-03-29 | 2022-02-01 | 南京维沃软件技术有限公司 | Image display method and terminal equipment |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004282368A (en) * | 2003-03-14 | 2004-10-07 | Omron Corp | Mobile object monitoring device |
CN101098462A (en) * | 2007-07-12 | 2008-01-02 | 上海交通大学 | Chroma deviation and brightness deviation combined video moving object detection method |
CN102915543A (en) * | 2012-09-12 | 2013-02-06 | 西安电子科技大学 | Figure motion change detecting method based on extracting function and three-channel separation |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8611600B2 (en) * | 2011-10-12 | 2013-12-17 | Raytheon Company | Three-frame difference moving target acquisition system and method for target track identification |
-
2015
- 2015-03-26 CN CN201510134601.6A patent/CN104751487B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2004282368A (en) * | 2003-03-14 | 2004-10-07 | Omron Corp | Mobile object monitoring device |
CN101098462A (en) * | 2007-07-12 | 2008-01-02 | 上海交通大学 | Chroma deviation and brightness deviation combined video moving object detection method |
CN102915543A (en) * | 2012-09-12 | 2013-02-06 | 西安电子科技大学 | Figure motion change detecting method based on extracting function and three-channel separation |
Non-Patent Citations (4)
Title |
---|
Motion detection and tracking using belief indicators for an automatic visual-surveillance system;Cina Motamed;《Performance Evaluation of Tracking and Surveillance》;20061101;第24卷(第11期);全文 * |
Tracking Groups of People;SJ Mckenna et al.;《Computer Vision & Image Understanding》;20001231;第80卷(第1期);全文 * |
基于RGB三通道分离的运动目标检测方法;沈盼盼 等;《视频应用与工程》;20121231;第36卷(第3期);全文 * |
基于帧间差分的彩色视频背景提取;苏礼坤 等;《成都信息工程学院学报》;20100430;第25卷(第2期);全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN104751487A (en) | 2015-07-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107808131B (en) | Dynamic gesture recognition method based on dual-channel deep convolutional neural network | |
CN104392468B (en) | Based on the moving target detecting method for improving visual background extraction | |
CN104848851B (en) | Intelligent Mobile Robot and its method based on Fusion composition | |
CN102509086B (en) | Pedestrian object detection method based on object posture projection and multi-features fusion | |
CN105488811B (en) | A kind of method for tracking target and system based on concentration gradient | |
CN105760835A (en) | Gait segmentation and gait recognition integrated method based on deep learning | |
CN109559320A (en) | Realize that vision SLAM semanteme builds the method and system of figure function based on empty convolution deep neural network | |
CN105975941A (en) | Multidirectional vehicle model detection recognition system based on deep learning | |
TWI640964B (en) | Image-based vehicle counting and classification system | |
CN107016691A (en) | Moving target detecting method based on super-pixel feature | |
CN103345755A (en) | Chessboard angular point sub-pixel extraction method based on Harris operator | |
CN106446862A (en) | Face detection method and system | |
CN107146239A (en) | Satellite video moving target detecting method and system | |
CN105894534B (en) | A kind of improvement moving target detecting method based on ViBe | |
CN106204594A (en) | A kind of direction detection method of dispersivity moving object based on video image | |
CN110796018A (en) | Hand motion recognition method based on depth image and color image | |
CN104751487B (en) | A kind of moving target detecting method based on the plane discoloration frames of colored RGB tri- difference | |
CN111886600A (en) | Device and method for instance level segmentation of image | |
Schorghuber et al. | SLAMANTIC-leveraging semantics to improve VSLAM in dynamic environments | |
CN109740609A (en) | A kind of gauge detection method and device | |
CN109166137A (en) | For shake Moving Object in Video Sequences detection algorithm | |
CN108764338A (en) | A kind of pedestrian tracking algorithm applied to video analysis | |
CN106570888A (en) | Target tracking method based on FAST (Features from Accelerated Segment Test) corner point and pyramid KLT (Kanade-Lucas-Tomasi) | |
CN116029979A (en) | Cloth flaw visual detection method based on improved Yolov4 | |
CN102915543B (en) | Figure motion change detecting method based on extracting function and three-channel separation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |