CN105205835A - Colored moving target recognizing and tracking method based on Fourier descriptor - Google Patents
Colored moving target recognizing and tracking method based on Fourier descriptor Download PDFInfo
- Publication number
- CN105205835A CN105205835A CN201510592435.4A CN201510592435A CN105205835A CN 105205835 A CN105205835 A CN 105205835A CN 201510592435 A CN201510592435 A CN 201510592435A CN 105205835 A CN105205835 A CN 105205835A
- Authority
- CN
- China
- Prior art keywords
- target
- color
- fourier descriptors
- tracking
- fourier
- 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
Detecting and tracking of moving objects are the key technology in video processing, and are the research hotspot and difficulty in the field of video processing and computer vision. In actual life, moving objects in videos are frequently shielded from time to time, the sizes and angles of the moving objects are frequency changed, and therefore the moving targets can be quite difficultly tracked. Therefore, a method for conducting target tracking through the combination of the color RGB values of the moving target objects and the Fourier descriptors of the shapes of the moving target objects is put forward. By means of the method, according to the color RGB values of the target objects to be tracked, the objects with the RGB values the same as target color RGB values are independently extracted in all frames, measurement and comparison are conducted on the shape Fourier descriptors of the objects and the shape Fourier descriptors of the target objects, and therefore the aim of recognizing and tracking the assigned targets is achieved.
Description
Technical field
The present invention relates to target recognition and tracking technology, particularly a kind of recognition and tracking method of the color motion target based on Fourier descriptors.The method has a wide range of applications in the fields such as community video monitoring, vehicular traffic monitoring.
Background technology
The detection and tracking of moving object is the gordian technique of Video processing. for the video that background is fixing, the tracking of target comparatively speaking, than being easier to.Target following just seems that difficulty how but be change for background, namely video equipment is when ceaselessly moving, as Vehicular video! When the object followed the tracks of is blocked, many target tracking algorisms are just helpless, as target tracking algorism (Vojir, Tomas based on mean shift; Noskova, Jana; Matas, Jiri.Robustscale-adaptivemean-shiftfortracking [J] .PATTERNRECOGNITIONLETTERS.vol49,2014,250-258), the two-dimensional grid model following algorithm (AlhmbasakY of based target, TckalpAM.Occkysion-adaptivecontent-based2Dmeshdesignandf orwardtracking [J] .IEEETransactionsonImageProcessing, 1997,6 (9): 1270-1280), the algorithm (Peng, the Chen that follow the tracks of of distinguished point based (key point); Chen, Qian; Qian, Wei-xian.Eigenspace-BasedTrackingforFeaturePoints [J] .OPTICALREVIEW, 2014, vol.21 (3), PP:304-312) etc.Many target tracking algorisms are only when tracing object generation translation, could accurately follow the tracks of, when the target of following the tracks of rotates, just can not accurately follow the tracks of, as: the people such as LiuCY are (LiuCY in the text, YungN.Scale-adaptivespatialappearancefraturedensityappro ximationforobjecttracking [J] .IEEETransactonsonIntelligentTransportationSystems, 2011,12 (1): 284-290) algorithm carried.In addition when the yardstick of the object followed the tracks of changes, tracking effect will be deteriorated, as article (WangChangjun, ZhuShanan.MeanShiftbasedtargets ' rotationandtranslationtracking [J] .JournalofImageandGraphics, 2007,12 (8): 1367-1371) algorithm carried.
For above-mentioned algorithm, namely when tracing object is blocked, yardstick changes, when angle changes, the situation that tracking effect is deteriorated, the Fourier descriptors [8,9] that this patent proposes a kind of rgb value based on object color and shape thereof carries out the algorithm followed the tracks of.This algorithm is according to the color rgb value of the target that will follow the tracks of, extract in each frame respectively, the object identical with color of object rgb value, then the shape Fourier descriptors of the destination object after the shape Fourier descriptors of these objects and normalization is compared, thus reach the object of target following.This algorithm effectively can solve object and be blocked, and yardstick in tracing process and Rotation.
Summary of the invention
The object of this invention is to provide a kind of method of moving target recognition and tracking, the method utilizes the color of moving target and this two large feature of shape Fourier descriptors thereof, realize target recognition and tracking.In order to achieve the above object, the present invention adopts following technical scheme: a kind of method of recognition and tracking of the colored moving target based on Fourier descriptor, and the method is divided into two large divisions, and the key step of every part can be described below:
Part I: the Fourier descriptors of the destination object of designated color is extracted
Step 1: the color rgb value (RValue, GValue, BValue) determining the destination object that will follow the tracks of;
Step 2: in the first frame of video, according to formula Image=(DiffR < T) & (DiffG < T) & (DiffB < T), wherein DiffR=|f
r(x, y)-ValueR|, DiffG=|f
g(x, y)-ValueG|, DiffB=|f
b(x, y)-ValueB|, T are threshold value, f
r(x, y), f
g(x, y), f
b(x, y) is respectively the red, green, blue passage of two field picture.So just, can obtain a bianry image, the color of its object comprised is identical with designated color, that is, by all object extraction identical with designated color out;
Step 3: after the cavity of the object that step 2 is extracted is filled, utilize formula ObjectSize < α or ObjectSize > β (α < β), the wherein size of ObjectSize destination object, non-designated target is deleted from image, so just can obtain the bianry image only having intended target object;
Step 4: utilize Boundary extracting algorithm (as Canny algorithm), obtains the boundary curve of intended target, and obtains the last discrete point sequence of intended target object bounds curve { (x further
i, y
i): i=0,1,2 ..., N-1}, and represent with the form of plural number;
Step 5: utilize formula
obtain the Fourier descriptors F (k) of intended target object;
Step 6: utilize formula
obtain the Fourier descriptors NormF (k) of normalized intended target object;
Part II: movable object tracking
Step 1: utilize the step 1 in Part I, the method in 2, has the object extraction of same color out by all in jth (j >=2) frame with destination object;
Step 2: utilize the step 4 in Part I, 5,6, obtain the normalization Fourier descriptors { NormF of each extraction object
h(k) h=1,2,3 ...;
Step 3: utilize formula
normF (k) and each { NormF in jth (j>=2) frame of metric objective object
h(k) h=1,2,3 ... similarity difference { D
h(Object, Object
h): h=1,2,3 ...;
Step 4: with the object that the similarity difference of destination object Fourier is minimum, the destination object of specifying will followed the tracks of exactly.
Advantage of the present invention is as follows:
(1) the present invention can in the moving target group of different colours, the moving target of locking designated color;
(2), when energy target shape of the present invention is similar, accurately identifies and follow the tracks of intended target;
(3) the present invention can be used for identifying and following the tracks of multiple target of specifying.
Accompanying drawing explanation
Fig. 1 original image;
All objects that Fig. 2 is identical with designated color;
Image after Fig. 3 fills;
The intended target object that Fig. 4 extracts;
The shape edges of Fig. 5 intended target object;
The object extracted in different frame in Fig. 6 monotrack;
Fig. 7 is for the monotrack result of different frame;
The object extracted in different frame in Fig. 8 multiple target tracking;
Fig. 9 is for the multiple target tracking result of different frame;
The similarity difference of the Fourier descriptors of each object in Figure 10 destination object (ball) and jth (j >=2) frame;
The similarity difference of the Fourier descriptors of each object in Figure 11 destination object (two right hands) and (j >=2) frame.
Embodiment
Below in conjunction with accompanying drawing, the present invention is further illustrated.
Embodiment one
Based on a recognition and tracking technology for the color motion target of Fourier descriptors, the method is made up of two large divisions, specifically comprises following steps:
Part I: the extraction of the Fourier descriptors of the moving target of designated color
Step 1: the color rgb value (RValue, GValue, BValue) determining the destination object that will follow the tracks of;
Step 2: in the first frame of video (shown in Fig. 1), according to formula Image=(DiffR < T) & (DiffG < T) & (DiffB < T), wherein DiffR=|f
r(x, y)-ValueR|, DiffG=|f
g(x, y)-ValueG|, DiffB=|f
b(x, y)-ValueB|, T are threshold value, f
r(x, y), f
g(x, y), f
b(x, y) is respectively the red, green, blue passage of two field picture.So just, can obtain a bianry image, as shown in Figure 2, the color of its object comprised is identical with designated color, that is, by all object extraction identical with designated color out;
Step 3: after the cavity of the object that step 2 is extracted is filled, as shown in Figure 3, utilize formula ObjectSize < α or ObjectSize > β (α < β), the wherein size of ObjectSize destination object, non-designated target is deleted from image, so just, the bianry image only having intended target object can be obtained, as shown in Figure 4;
Step 4: utilize Boundary extracting algorithm (as Canny algorithm) to obtain the boundary curve of intended target, as shown in Figure 5, and obtains the last discrete point sequence of intended target object bounds curve { (x further
i, y
i): i=0,1,2 ..., N-1}, and represent with the form of plural number;
Step 5: utilize formula
obtain the Fourier descriptors F (k) of intended target object;
Step 6: utilize formula
obtain the Fourier descriptors NormF (k) of normalized intended target object;
Part II: movable object tracking
Step 1: utilize the step 1 in Part I, the method in 2, has the object extraction of same color out by all in jth (j >=2) frame with destination object, as shown in Figure 6 and Figure 8;
Step 2: utilize the step 4 in Part I, 5,6, obtain the normalization Fourier descriptors { NormF of each extraction object
h(k) h=1,2,3 ...;
Step 3: utilize formula
normF (k) and each { NormF in jth (j>=2) frame of metric objective object
h(k) h=1,2,3 ... similarity difference { D
h(Object, Object
h): h=1,2,3 ..., as shown in Figure 10 and Figure 11;
Step 4: with the object that the similarity difference of destination object Fourier is minimum, the destination object of specifying will followed the tracks of exactly, as shown in figures 7 and 9.
Claims (4)
1., based on a recognition and tracking method for the color motion target of Fourier descriptors, it is characterized in that, the method is made up of two large divisions, comprises following steps:
Part I: the Fourier descriptors of the destination object of designated color is extracted
Step 1: the color rgb value (RValue, GValue, BValue) determining the destination object that will follow the tracks of;
Step 2: in the first frame of video, according to formula Image=(DiffR < T) & (DiffG < T) & (DiffB < T) wherein DiffR=|f
r(x, y)-ValueR|, DiffG=|f
g(x, y)-ValueG|, DiffB=|f
b(x, y)-ValueB|, T are threshold value, f
r(x, y), f
g(x, y), f
b(x, y) is respectively the red, green, blue passage of two field picture;
So just, can obtain a bianry image, the color of its object comprised is identical with designated color, that is, by all object extraction identical with designated color out;
Step 3: after the cavity of the object that step 2 is extracted is filled, utilize formula ObjectSize < α or ObjectSize > β (α < β), the wherein size of ObjectSize destination object, non-designated target is deleted from image, so just can obtain the bianry image only having intended target object;
Step 4: utilize Boundary extracting algorithm (as Canny algorithm), obtains the boundary curve of intended target, and obtains the last discrete point sequence of intended target object bounds curve { (x further
i, y
i): i=0,1,2 ..., N-1}, and represent with the form of plural number;
Step 5: utilize formula
obtain the Fourier descriptors F (k) of intended target object;
Step 6: utilize formula
obtain the Fourier descriptors NormF (k) of normalized intended target object;
Part II: movable object tracking
Step 1: utilize the step 1 in Part I, the method in 2, has the object extraction of same color out by all in jth (j >=2) frame with destination object;
Step 2: utilize the step 4 in Part I, 5,6, obtain the normalization Fourier descriptors { NormF of each extraction object
h(k) h=1,2,3 ...;
Step 3: utilize formula
normF (k) and each { NormF in jth (j>=2) frame of metric objective object
h(k) h=1,2,3 ... similarity difference { D
h(Object, Object
h): h=1,2,3 ...;
Step 4: with the object that the similarity difference of destination object Fourier is minimum, the destination object of specifying will followed the tracks of exactly.
2. the recognition and tracking method of a kind of color motion target based on Fourier descriptors according to claims 1, is characterized in that, the step 1 in Part I, step 2, step 3 come according to the rgb value of target, extracts and specifies target color.
3. the recognition and tracking method of a kind of color motion target based on Fourier descriptors according to claims 1, it is characterized in that, Part II is the identification utilizing the Fourier descriptors of target shape to realize moving target.
4. the recognition and tracking method of a kind of color motion target based on Fourier descriptors according to claims 1, is characterized in that, the Fourier descriptors of combining target color and target shape realizes colored moving target recognition and tracking.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510592435.4A CN105205835A (en) | 2015-09-02 | 2015-09-17 | Colored moving target recognizing and tracking method based on Fourier descriptor |
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2015206720472 | 2015-09-02 | ||
| CN201520672047 | 2015-09-02 | ||
| CN201510592435.4A CN105205835A (en) | 2015-09-02 | 2015-09-17 | Colored moving target recognizing and tracking method based on Fourier descriptor |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105205835A true CN105205835A (en) | 2015-12-30 |
Family
ID=54953495
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510592435.4A Pending CN105205835A (en) | 2015-09-02 | 2015-09-17 | Colored moving target recognizing and tracking method based on Fourier descriptor |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105205835A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8367430B2 (en) * | 2009-10-07 | 2013-02-05 | GlobalFoundries, Inc. | Shape characterization with elliptic fourier descriptor for contact or any closed structures on the chip |
| CN104331682A (en) * | 2014-10-11 | 2015-02-04 | 东南大学 | Automatic building identification method based on Fourier descriptor |
| CN104680127A (en) * | 2014-12-18 | 2015-06-03 | 闻泰通讯股份有限公司 | Gesture identification method and gesture identification system |
| CN104751487A (en) * | 2015-03-26 | 2015-07-01 | 太原理工大学 | Method for detecting movement target based on colored RGB three-pane color-change frame difference |
-
2015
- 2015-09-17 CN CN201510592435.4A patent/CN105205835A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8367430B2 (en) * | 2009-10-07 | 2013-02-05 | GlobalFoundries, Inc. | Shape characterization with elliptic fourier descriptor for contact or any closed structures on the chip |
| CN104331682A (en) * | 2014-10-11 | 2015-02-04 | 东南大学 | Automatic building identification method based on Fourier descriptor |
| CN104680127A (en) * | 2014-12-18 | 2015-06-03 | 闻泰通讯股份有限公司 | Gesture identification method and gesture identification system |
| CN104751487A (en) * | 2015-03-26 | 2015-07-01 | 太原理工大学 | Method for detecting movement target based on colored RGB three-pane color-change frame difference |
Non-Patent Citations (2)
| Title |
|---|
| 吴青青 等: "复杂背景下的颜色分离背景差分目标检测方法", 《兵工学报》 * |
| 韩鸿哲 等: "基于傅里叶描述子的步态识别", 《计算机工程》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105528794A (en) | Moving object detection method based on Gaussian mixture model and superpixel segmentation | |
| CN109344842A (en) | A kind of pedestrian's recognition methods again based on semantic region expression | |
| CN103473551A (en) | Station logo recognition method and system based on SIFT operators | |
| CN104268853A (en) | Infrared image and visible image registering method | |
| CN103473785A (en) | Rapid multiple target segmentation method based on three-valued image clustering | |
| CN104851089A (en) | Static scene foreground segmentation method and device based on three-dimensional light field | |
| CN103533332B (en) | A kind of 2D video turns the image processing method of 3D video | |
| Qazi et al. | Automated ball tracking in tennis videos | |
| Li et al. | Simplified histograms of oriented gradient features extraction algorithm for the hardware implementation | |
| CN104751139A (en) | Fast fingerprint recognition method based on feature points of sweat glands and fingerprint images | |
| CN106022236A (en) | Action identification method based on human body contour | |
| CN105528795B (en) | A kind of infrared face dividing method using annular shortest path | |
| CN106603888A (en) | Image color extraction processing structure | |
| CN104574435A (en) | Motion camera foreground segmentation method based on block clustering | |
| CN104866819B (en) | A kind of classification of landform method based on trinocular vision system | |
| CN105205835A (en) | Colored moving target recognizing and tracking method based on Fourier descriptor | |
| Edoardo Maresca et al. | The Matrioska tracking algorithm on LTDT2014 dataset | |
| CN101894379A (en) | Method and device for segmentation of characteristic point motion for large interframe motion video | |
| Shrivastava et al. | A smoke detection algorithm based on K-means segmentation | |
| Lee et al. | Real-time pedestrian and vehicle detection in video using 3d cues | |
| Fard et al. | A block matching based method for moving object detection in active camera | |
| Baek et al. | Cognition system of bolt hole using template matching | |
| Yang et al. | Cascaded superpixel pedestrian object segmentation algorithm | |
| CN104298967B (en) | A kind of vehicle image comparison method of view-based access control model feature | |
| Shen et al. | Plant depth detection image restoration based on k-means and k-nearest neighbor algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20151230 |
|
| WD01 | Invention patent application deemed withdrawn after publication |