CN101667250A - Device and method for identifying hand on basis of Camshift algorithm and hand-shape shade - Google Patents
Device and method for identifying hand on basis of Camshift algorithm and hand-shape shade Download PDFInfo
- Publication number
- CN101667250A CN101667250A CN200910196986A CN200910196986A CN101667250A CN 101667250 A CN101667250 A CN 101667250A CN 200910196986 A CN200910196986 A CN 200910196986A CN 200910196986 A CN200910196986 A CN 200910196986A CN 101667250 A CN101667250 A CN 101667250A
- Authority
- CN
- China
- Prior art keywords
- hand
- algorithm
- module
- hand shape
- camshift
- 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
Images
Abstract
The invention discloses a device and a method for identifying hand on basis of an Camshift algorithm and a hand-shape shade, comprising an interference area removing module, a hand part area initial tracking module, an image processing module and an identification module based on a hand-shaped template; the method comprises the following steps: 1) adopting an Haar algorithm to identify a face areaand adding removal; 2) adopting the Camshift algorithm to carry out initial identification on the hand part area of the image; 3) carrying out binarization, corrosion and expansion processing on thecolor probability distribution chart generated by the Camshift algorithm; 4) segmenting the communicated area meeting specified size in the processed chart and comparing the communicated area with each hand-shaped shade respectively so as to carry out similarity contrast; and 5) judging whether the maximum value in the similarity value is more than the threshold or not; if so, determining the handand marking the communicated area; and if not, inputting next frame and executing the step 1). The method integrates the advantages of a plurality of algorithms, carries out continuous tracking on the areas of the hand in the video, and has high execution efficiency and high tracking accuracy.
Description
Technical field
The present invention relates to virtual reality and computer graphics field, particularly a kind of hand recognition device and method thereof based on CamShift algorithm and hand shape shade of the staff position in the video being carried out continuous recognition and tracking location.
Background technology
Hand identification is important contents in field such as man-machine interaction, robot, computer vision and Digital Image Processing.Adversary's position is accurately measured and is followed the tracks of, and in man-machine interaction, aspects such as gesture identification are significant.
The method of hand recognition and tracking is a lot, and existing method comprises based on colour of skin information, based on movable information, based on motion model etc.At present, people such as the Gu Li of Chinese University of Science and Technology, Zhuan Zhenquan utilize illumination compensation and extract based on the local threshold hand shape extracting method adversary shape of searching classifiably, and extraction effect is better, but fails to be implemented in the tracking in the continuous videos.The Zhang Liangguo of Harbin Institute of Technology etc. has proposed based on the hand recognition system of hausdorff distance and has utilized it to carry out gesture identification, but responsive to factors such as illumination and complex backgrounds.The mode that the Sun Chao of Harbin Institute of Technology, Jiang Li utilize the FSR sensor to add support vector machine detects, and effect is better but higher to hardware requirement.
Existing hand recognition technology or technology are complicated, certain defective is arranged on real-time, or mainly apply to the still image of single frames, can not carry out higher tracking of accuracy and identification to continuous videos.Native system proposes the hand recognition methods that a kind of Camshift of utilization algorithm combines with hand shape shade.This method is at continuous videos, based on color analysis based on CamShift, in conjunction with the shape analysis technology of hand shape template, and utilize Haar algorithm etc. that the interference component of video is handled and removed, can carry out continuously and the higher recognition and tracking of accuracy position, hand place in the video.
CamShift wherein is the abbreviation of " Continuously Adaptive Mean-SHIFT ", and promptly continuous adaptive Mean-SHIFT algorithm is a kind of based on the Mean-SHIFT algorithm, is applied to the track algorithm of continuous videos.CamShift is based on the random color probability model, and is irrelevant with the concrete shape of tracked object.The CamShift algorithm has adopted the HSV color space, is subjected to such environmental effects such as illumination less, and the staff colour of skin distributes distinct at HSV space chromatic value.CamShift is divided into back projection (Back Projection), MeanShift algorithm and CamShift algorithm three parts.
List of references [1] and [2] disclose the Haar algorithm.
[1]Paul?Viola，Michael?J.Jones.Robust?Real-Time?Face?Detection[J].International?Journal?of?Computer?Vision，May?2004，57：137-154.
[2]Friedman，J.H.，Hastie，T.and?Tibshirani，R.Additive?Logistic?Regression：a?Statistical?View?of?Boosting.Technical?Report[R]，Dept.of?Statistics，StanfordUniversity，1998.
In view of this, this area inventor has researched and developed a kind of hand recognition device and method thereof based on CamShift algorithm and hand shape shade of the staff position in the video being carried out continuous recognition and tracking location at the problems referred to above.
Summary of the invention
The objective of the invention is to, a kind of hand recognition device and method thereof based on CamShift algorithm and hand shape shade is provided, overcome the difficulty of prior art, reach the purpose of the staff position in the video being carried out continuous recognition and tracking location.
The present invention adopts following technical scheme:
A kind of hand recognition device based on CamShift algorithm and hand shape shade of the present invention comprises that interference region is removed module, the preliminary tracing module of hand region, image processing module and based on the identification module of hand shape template;
Described interference region is removed module by the Haar algorithm identified and remove the interference position that people's face etc. comprises the colour of skin;
The preliminary tracing module of described hand region is carried out preliminary identification and is followed the trail of continuously the hand zone in the video by the CamShift algorithm;
Described image processing module is to handling each frame that obtains through CamShift, and the binaryzation, the burn into that carry out image expand, mark off operations such as each connected region, remove the image noise;
Described identification module based on hand shape template utilizes hand shape template that image is handled, and judges that whether each connected region is the hand zone, determines final recognition and tracking zone.
Further, described interference region removal module comprises Haar algorithm training module and Haar algorithm identified module at least.
Further, the preliminary tracing module of described hand region comprises processing module and the identification module based on the CamShift algorithm at least.
Further, describedly be divided into a left side based on hand shape shade in the identification module of hand shape template and open, a left side merges, and open on the right side, rightly merges four kinds.
Further, four kinds of described hand shape shades are further divided into rotation 0 degree, positive and negative 45 degree, five kinds of positive and negative 90 degree, totally 20 kinds of shades.
Above-mentioned hand recognition device based on CamShift algorithm and hand shape shade is adopted in a kind of hand recognition methods based on CamShift algorithm and hand shape shade of the present invention, may further comprise the steps:
1) adopts Haar algorithm identified face area and be added on removal;
2) adopt the CamShift algorithm that the hand region of image is carried out preliminary identification;
3) the color probability distribution graph that the CamShift algorithm is generated is carried out binaryzation, burn into expansion process;
4) connected region that satisfies a certain size among the figure after will handling splits, and carries out the similarity contrast with each hand shape shade respectively, and formula (1) is for calculating at certain formula of the similarity value of shape shade on the other hand:
p
r＝(S/S
1+S/S
2)/2 (1)
Wherein, p
rBe similarity value, S
1Be certain connected region area, S
2Be hand shade area, S is both overlapping areas,
Each hand shape shade is calculated corresponding p
rValue is got all p
rMiddle maximal value p
RmaxBe final similarity value;
5) judge described final similarity value p
RmaxWhether greater than threshold value, if, then be judged to be hand, mark this connected region; If not, then import next frame, execution in step 1).
Further, also comprise predetermined threshold value before the described step 1).
Further, the hand shape shade in the described step 4) is divided into a left side and opens, and a left side merges, and open on the right side, four kinds of right merging.
Further, four kinds of described hand shape shades are further divided into rotation 0 degree, positive and negative 45 degree, five kinds of positive and negative 90 degree, totally 20 kinds of shades.
Further, when being judged to be hand, described connected region is marked with square frame in the described step 5), finish the tracking of hand position.
With respect to existing additive method, the present invention is creatively with the Haar algorithm, the CamShift algorithm, computer vision such as hand shape shade processing method and Digital Image Processing algorithm combine, and concentrate the advantage of several algorithms, earlier image being carried out interference region removes, to combine based on the recognition and tracking of color and recognition and tracking algorithm again, can carry out Continuous Tracking, evidence the hand region in the video based on shape, it is higher that this method is carried out efficient, and the tracking accuracy is higher.Also have compared to existing technology and can carry out higher tracking of accuracy and identification continuous videos, low to factor susceptibilitys such as illumination and complex backgrounds, the advantage of reduction hardware requirement.
Further specify the present invention below in conjunction with drawings and Examples.
Description of drawings
Fig. 1 among the present invention based on the structural representation of the hand recognition device of CamShift algorithm and hand shape shade;
Fig. 2 among the present invention based on the process flow diagram of the hand recognition methods of CamShift algorithm and hand shape shade;
Fig. 3 is by being used part hand shape template figure among the embodiment 1.
Embodiment
Below by Fig. 1 to 3, introduce a kind of specific embodiment of the present invention.
As shown in Figure 1, a kind of hand recognition device based on CamShift algorithm and hand shape shade of the present invention comprises by data line linking to each other and the interference region of swap data is removed module, the preliminary tracing module of hand region, image processing module and based on the identification module of hand shape template.
Wherein, described interference region is removed module by the Haar algorithm identified and remove the interference position that people's face etc. comprises the colour of skin, and described interference region is removed module and comprised Haar algorithm training module and Haar algorithm identified module at least.
The preliminary tracing module of described hand region is carried out preliminary identification and is followed the trail of continuously the hand zone in the video by the CamShift algorithm, and the preliminary tracing module of described hand region comprises processing module and the identification module based on the CamShift algorithm at least.
Described image processing module is to handling each frame that obtains through CamShift, and the binaryzation, the burn into that carry out image expand, mark off operations such as each connected region, remove the image noise;
Described identification module based on hand shape template utilizes hand shape template that image is handled, and judges that whether each connected region is the hand zone, determines final recognition and tracking zone.Describedly be divided into a left side based on hand shape shade in the identification module of hand shape template and open, a left side merges, and open on the right side, rightly merges four kinds, and these four kinds described hand shape shades are further divided into rotation 0 degree, positive and negative 45 degree, five kinds of positive and negative 90 degree, totally 20 kinds of shades.
Referring to Fig. 3, enumerated wherein 8 kinds:
(left side merges, 0 degree), (the right merging, 0 degree);
(open on a left side, 0 degree), (open on the right side, 0 degree);
(left side merges, 90 degree), (the right merging, 90 degree);
(left side merges, 45 degree), (the right merging, 45 degree).
As shown in Figure 2, above-mentioned hand recognition device based on CamShift algorithm and hand shape shade is adopted in a kind of hand recognition methods based on CamShift algorithm and hand shape shade of the present invention, may further comprise the steps:
Predetermined threshold value.
1) adopts Haar algorithm identified face area and be added on removal;
2) adopt the CamShift algorithm that the hand region of image is carried out preliminary identification;
3) the color probability distribution graph that the CamShift algorithm is generated is carried out binaryzation, burn into expansion process;
4) connected region that satisfies a certain size among the figure after will handling splits, and carries out the similarity contrast with each hand shape shade respectively, and formula (1) is for calculating at certain formula of the similarity value of shape shade on the other hand:
p
r＝(S/S
1+S/S
2)/2 (1)
Wherein, p
rBe similarity value, S
1Be certain connected region area, S
2Be hand shade area, S is both overlapping areas,
Each hand shape shade is calculated corresponding p
rValue is got all p
rMiddle maximal value p
RmaxBe final similarity value;
Hand shape shade is divided into a left side and opens, and a left side merges, and open on the right side, four kinds of right merging, and these four kinds described hand shape shades are further divided into rotation 0 degree, positive and negative 45 degree, five kinds of positive and negative 90 degree, totally 20 kinds of shades.
Referring to Fig. 3, enumerated wherein 8 kinds:
(left side merges, 0 degree), (the right merging, 0 degree);
(open on a left side, 0 degree), (open on the right side, 0 degree);
(left side merges, 90 degree), (the right merging, 90 degree);
(left side merges, 45 degree), (the right merging, 45 degree).
5) judge described final similarity value p
RmaxWhether greater than threshold value, if, then be judged to be hand, described connected region is marked with square frame, finish the tracking of hand position; If not, then import next frame, execution in step 1).
In summary, the present invention is creatively with the Haar algorithm, the CamShift algorithm, computer vision such as hand shape shade processing method and Digital Image Processing algorithm combine, and concentrate the advantage of several algorithms, earlier image being carried out interference region removes, to combine based on the recognition and tracking of color and recognition and tracking algorithm again, can carry out Continuous Tracking, evidence the hand region in the video based on shape, it is higher that this method is carried out efficient, and the tracking accuracy is higher.
Above-described embodiment only is used to illustrate technological thought of the present invention and characteristics, its purpose is to make those skilled in the art can understand content of the present invention and implements according to this, can not only limit claim of the present invention with present embodiment, be all equal variation or modifications of doing according to disclosed spirit, still drop in the claim of the present invention.
Claims (10)
1, a kind of hand recognition device based on CamShift algorithm and hand shape shade is characterized in that: comprise that interference region is removed module, the preliminary tracing module of hand region, image processing module and based on the identification module of hand shape template;
Described interference region is removed module by the Haar algorithm identified and remove the interference position that people's face etc. comprises the colour of skin;
The preliminary tracing module of described hand region is carried out preliminary identification and is followed the trail of continuously the hand zone in the video by the CamShift algorithm;
Described image processing module is to handling each frame that obtains through CamShift, and the binaryzation, the burn into that carry out image expand, mark off operations such as each connected region, remove the image noise;
Described identification module based on hand shape template utilizes hand shape template that image is handled, and judges that whether each connected region is the hand zone, determines final recognition and tracking zone.
2, the hand recognition device based on CamShift algorithm and hand shape shade according to claim 1 is characterized in that: described interference region is removed module and is comprised Haar algorithm training module and Haar algorithm identified module at least.
3, the hand recognition device based on CamShift algorithm and hand shape shade according to claim 1, it is characterized in that: the preliminary tracing module of described hand region comprises processing module and the identification module based on the CamShift algorithm at least.
4, the hand recognition device based on CamShift algorithm and hand shape shade according to claim 1 is characterized in that: describedly be divided into a left side based on hand shape shade in the identification module of hand shape template and open, a left side merges, and open on the right side, rightly merges four kinds.
5, the hand recognition device based on CamShift algorithm and hand shape shade according to claim 4 is characterized in that: four kinds of described hand shape shades are further divided into rotation 0 degree, positive and negative 45 degree, five kinds of positive and negative 90 degree, totally 20 kinds of shades.
6, any hand recognition device based on CamShift algorithm and hand shape shade in the claim 1 to 5 is as described adopted in a kind of hand recognition methods based on CamShift algorithm and hand shape shade, it is characterized in that may further comprise the steps:
1) adopts Haar algorithm identified face area and be added on removal;
2) adopt the CamShift algorithm that the hand region of image is carried out preliminary identification;
3) the color probability distribution graph that the CamShift algorithm is generated is carried out binaryzation, burn into expansion process;
4) connected region that satisfies a certain size among the figure after will handling splits, and carries out the similarity contrast with each hand shape shade respectively, and formula (1) is for calculating at certain formula of the similarity value of shape shade on the other hand:
p
r＝(S/S
1+S/S
2)/2(1)
Wherein, p
rBe similarity value, S
1Be certain connected region area, S
2Be hand shade area, S is both overlapping areas,
Each hand shape shade is calculated corresponding p
rValue is got all p
rMiddle maximal value p
RmaxBe final similarity value;
5) judge described final similarity value p
RmaxWhether greater than threshold value, if, then be judged to be hand, mark this connected region; If not, then import next frame, execution in step 1).
7, the hand recognition methods based on CamShift algorithm and hand shape shade according to claim 6 is characterized in that: also comprise predetermined threshold value before the described step 1).
8, the hand recognition methods based on CamShift algorithm and hand shape shade according to claim 6, it is characterized in that: the hand shape shade in the described step 4) is divided into a left side and opens, and a left side merges, and open on the right side, four kinds of right merging.
9, the hand recognition methods based on CamShift algorithm and hand shape shade according to claim 8 is characterized in that: four kinds of described hand shape shades are further divided into rotation 0 degree, positive and negative 45 degree, five kinds of positive and negative 90 degree, totally 20 kinds of shades.
10, the hand recognition methods based on CamShift algorithm and hand shape shade according to claim 6 is characterized in that: when being judged to be hand, described connected region is marked with square frame in the described step 5), finish the tracking of hand position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910196986A CN101667250A (en) | 2009-10-10 | 2009-10-10 | Device and method for identifying hand on basis of Camshift algorithm and hand-shape shade |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910196986A CN101667250A (en) | 2009-10-10 | 2009-10-10 | Device and method for identifying hand on basis of Camshift algorithm and hand-shape shade |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101667250A true CN101667250A (en) | 2010-03-10 |
Family
ID=41803866
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200910196986A Pending CN101667250A (en) | 2009-10-10 | 2009-10-10 | Device and method for identifying hand on basis of Camshift algorithm and hand-shape shade |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101667250A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894378A (en) * | 2010-06-13 | 2010-11-24 | 南京航空航天大学 | Moving target visual tracking method and system based on double ROI (Region of Interest) |
| WO2012139241A1 (en) * | 2011-04-11 | 2012-10-18 | Intel Corporation | Hand gesture recognition system |
| CN108333150A (en) * | 2018-01-09 | 2018-07-27 | 北京航空航天大学 | A kind of two-dimensional surface plasma most preferably excites the Direct Recognition method of Angle Position |
| CN110609617A (en) * | 2013-08-04 | 2019-12-24 | 艾斯适配有限公司 | Apparatus, system and method for virtual mirrors |
-
2009
- 2009-10-10 CN CN200910196986A patent/CN101667250A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101894378A (en) * | 2010-06-13 | 2010-11-24 | 南京航空航天大学 | Moving target visual tracking method and system based on double ROI (Region of Interest) |
| WO2012139241A1 (en) * | 2011-04-11 | 2012-10-18 | Intel Corporation | Hand gesture recognition system |
| US8781221B2 (en) | 2011-04-11 | 2014-07-15 | Intel Corporation | Hand gesture recognition system |
| CN110609617A (en) * | 2013-08-04 | 2019-12-24 | 艾斯适配有限公司 | Apparatus, system and method for virtual mirrors |
| CN110609617B (en) * | 2013-08-04 | 2023-09-26 | 艾斯适配有限公司 | Apparatus, system and method for virtual mirror |
| CN108333150A (en) * | 2018-01-09 | 2018-07-27 | 北京航空航天大学 | A kind of two-dimensional surface plasma most preferably excites the Direct Recognition method of Angle Position |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9916012B2 (en) | Image processing apparatus, image processing method, and program | |
| JP6608465B2 (en) | Gesture detection and identification method and system | |
| Keskin et al. | Randomized decision forests for static and dynamic hand shape classification | |
| CN107679512A (en) | A kind of dynamic gesture identification method based on gesture key point | |
| CN109145742A (en) | A kind of pedestrian recognition method and system | |
| CN103218605A (en) | Quick eye locating method based on integral projection and edge detection | |
| TW200719871A (en) | A real-time face detection under complex backgrounds | |
| Yang et al. | Real-time traffic sign detection via color probability model and integral channel features | |
| CN102194108A (en) | Smiley face expression recognition method based on clustering linear discriminant analysis of feature selection | |
| CN103106409A (en) | Composite character extraction method aiming at head shoulder detection | |
| Uchiyama et al. | Deformable random dot markers | |
| CN102610104B (en) | Onboard front vehicle detection method | |
| CN103425985B (en) | A kind of face wrinkles on one's forehead detection method | |
| CN109086772A (en) | A kind of recognition methods and system distorting adhesion character picture validation code | |
| CN105913034B (en) | Vehicle identification method and device and vehicle | |
| CN107563331A (en) | A kind of road sign line detecting method and system based on geometrical relationship | |
| CN101667250A (en) | Device and method for identifying hand on basis of Camshift algorithm and hand-shape shade | |
| Lee et al. | An effective method for detecting facial features and face in human–robot interaction | |
| KR101407070B1 (en) | Method of extracting feature for human detection and apparatus for performing the same | |
| CN110807348A (en) | Method for removing interference lines in document image based on greedy algorithm | |
| CN110781769A (en) | Method for rapidly detecting and tracking pedestrians | |
| CN105184244A (en) | Video face detection method and apparatus | |
| Cai et al. | SVM-KNN algorithm for image classification based on enhanced HOG feature | |
| KR100729265B1 (en) | A face detection method using difference image and color information | |
| Niazi et al. | Hybrid face detection with HSV Color method and HAAR classifier |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20100310 |