CN101702234A - Method for selecting optimal viewing angle of video human face by multi-lens camera - Google Patents
Method for selecting optimal viewing angle of video human face by multi-lens camera Download PDFInfo
- Publication number
- CN101702234A CN101702234A CN200910198221A CN200910198221A CN101702234A CN 101702234 A CN101702234 A CN 101702234A CN 200910198221 A CN200910198221 A CN 200910198221A CN 200910198221 A CN200910198221 A CN 200910198221A CN 101702234 A CN101702234 A CN 101702234A
- Authority
- CN
- China
- Prior art keywords
- human face
- viewing angle
- lens camera
- area
- optimal viewing
- 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
Landscapes
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a method for selecting an optimal viewing angle of a video human face by a multi-lens camera, comprising the following steps of: (1) acquiring a human face video image by the multi-lens camera from different viewing angles; (2) converting colorful spaces; (3) extracting complexion areas; (4) determining a human face area; (5) calculating viewing angle evaluation factors; and (6) determining the optimal viewing angle of the camera. In the method, the favorable clustering characteristic of human face complexion in a color space of YCrCb is utilized, the human face complexion area is extracted from video images captured by the multi-lens camera, the specific valve of the human face complexion area to a rectangular area corresponding to the human face complexion area is calculated, and the determination of the optimal viewing angle of the multi-lens camera is realized, thereby the method has simple, convenient, flexible and easily-realized operation, solves the problems that the selecting method of the optimal viewing angle for the video human face by a multi-lens camera system needs relative scenes, head three-dimensional geometric structures and light source information and improves the robustness of the optimal viewing angle selection for videoing the human face by the multi-lens camera system.
Description
Technical field
The invention belongs to the intelligent information processing technology field, relate to a kind of system of selection of optimal viewing angle of video human face by multi-lens camera, be used for video digital images analysis and understanding.
Background technology
In the practical application scene, because the single camera visual angle is limited, and existence is blocked in the scene, therefore, uses single camera to be difficult to scene is implemented effectively monitoring, discerns, analyzes and understood.The multi-lens camera system can cover scope with great visual angle, and can show from scene image in the different visual angles scope and information, overcomes and blocks influence.But how from the multi-lens camera system, it is crucial choosing the video camera with optimal viewing angle.Because optimal viewing angle video camera capturing information amount is big, help the analysis and the understanding of scene information, and the choosing of optimal viewing angle video camera, help to give up redundant information from other video camera, save computing cost, improve efficient scene monitoring, identification, analysis and understanding.Because the most important visual signature of fundamental sum that people's face is the human visual system to have, it also is the basis that human vision exchanges, human by people's face visual signature, can discern identity each other, understand the other side's emotion and intention, therefore, carry out video human face detection, tracking, identification, behavioural analysis and understanding etc. and become in recent years one of focus of research both at home and abroad, and the optimal viewing angle facial image comprises the maximum fault information of people's face, helps the solution of the problems referred to above.
The system of selection of many optimal viewing angle of video human face by multi-lens camera has been proposed at present.Mainly contain based on model method with based on characterization method.The former sets up the statistical model of head under different visual angles according to the Given information of relevant scene, head three-dimensional geometrical structure and light source, chooses the optimal viewing angle video camera.These information all are difficult to obtain in practical application, and the algorithm complexity.The latter adopts characteristics of image, judges as information such as how much, brightness, color, saturation degree, calculates the characteristic information of each pixel and respective background point relatively; Carry out the feature judgement according to background that detects and prospect, when image background changes, can't effectively select the optimal viewing angle video camera, this method applicability and poor accuracy.Existing optimal viewing angle of video human face by multi-lens camera system of selection need relevant scene, head three-dimensional geometrical structure and light source information and to dynamic scene change responsive, noise greatly, computing is complicated.
Summary of the invention
Problem and shortage in view of above-mentioned prior art existence, the present invention proposes a kind of system of selection of optimal viewing angle of video human face by multi-lens camera, this method can be able to be obtained the video camera of best capture people face optimal viewing angle easy, quickly from multi-lens camera, its computing is easy, flexible, the robustness that multi-lens camera system video people face optimal viewing angle is selected can be improved, the automatic selection of the optimal viewing angle of video human face under the multiple condition can be adapted to.
For reaching above-mentioned purpose, design of the present invention is: this method has good color cluster characteristic based on face complexion, catch the different visual angles area of skin color of people's face by multi-lens camera, its face complexion area has different ratio amplitudes with its corresponding rectangular area, catch people's face quantity of information according to this ratio amplitude reflection different cameras, from the multi-lens camera system, select to have the optimal viewing angle video camera automatically.
According to above-mentioned design, the present invention adopts following technical scheme:
The system of selection of above-mentioned optimal viewing angle of video human face by multi-lens camera, concrete steps are as follows:
(1), starts the multi-lens camera image capturing system: gather the video human face image from different visual angles by multi-lens camera;
(2), color space conversion, calculate color-values Cr, Cb:, determine the color-values Cr of YCbCr color space, Cb by the red R of RGB color space, green G, blue B three-component:
Cr＝0.5×R-0.4187×G-0.0813×B
Cb＝-0.1687×R-0.3313×G+0.5×B
(3), area of skin color extracts: determine color-values Cr, the threshold value T of Cb and Cr/Cb respectively
1, T
2, T
3, T
4, T
5, the zone with all pixels that satisfy following formula are formed is defined as area of skin color S
S＝(Cr≥T
1∩Cr≤T
2)∩(Cb≥T
3∩Cb≤T
4)∩(Cr/Cb≥T
5)
Wherein, ∩ is " logical and " operational character.
(4), determining of human face region: the bianry image that satisfies (b) is carried out the connected region search, calculate long and wide ratio of connected region and the hole number in the connected region, the area of skin color that all pixels that satisfy following formula are formed is defined as human face region
F＝(S
1/S
w≥T
6∩S
1/S
w≤T
7)∩(H＞1)
Wherein, S
1, S
wBe respectively the length and the width of colour of skin connected region, T
6, T
7Be threshold value, H is the hole number in the colour of skin connected region.
(5), the visual angle estimates the factor and calculates: according to the face complexion area F that extracts, will comprise rectangular area M that all pixels of this human face region form as the face complexion rectangular area, and be calculated as follows the visual angle and estimate the factor
R＝F/M
(6), determining of optimal viewing angle video camera:, select to have the video camera that factor values is estimated at maximum visual angle, as the video camera of best capture people face according to the size of visual angle evaluation factor values.
The system of selection of optimal viewing angle of video human face by multi-lens camera of the present invention compared with prior art, have following conspicuous outstanding substantive distinguishing features and remarkable advantage: this method is because face complexion presents good cluster characteristic in the color space of YCrCb, be subjected to the dynamic environment variable effect little, and people's face distribution characteristics is stable, thereby can make full use of face characteristic, easy, obtain the area of skin color of people's face quickly, and and then determine to comprise the rectangular area size of this area of skin color, obtain the visual angle and estimate factor values.According to the size of visual angle evaluation factor values, from multi-lens camera, select to have the video camera that factor values is estimated at maximum visual angle automatically, as the video camera of catching people's face optimal viewing angle; This method solved multi-lens camera system video people face optimal viewing angle system of selection need relevant scenes, head three-dimensional geometrical structure and light source information and to dynamic scene change responsive, noise greatly, the problem of the complicated deficiency of computing; Adopt this method can improve the robustness that multi-lens camera system video people face optimal viewing angle is selected, can adapt to the automatic selection of the optimal viewing angle of video human face under the multiple condition, be with a wide range of applications.
Description of drawings
Fig. 1 is the FB(flow block) of the system of selection of optimal viewing angle of video human face by multi-lens camera of the present invention;
The figure of Fig. 2 original image that to be step in the embodiments of the invention (1) catch from 6 video cameras;
Fig. 3 is the figure of the definite two-value facial image of step in the embodiments of the invention (4);
The figure of Fig. 4 facial image that to be one embodiment of the invention step (6) caught by the optimal viewing angle video camera;
Embodiment
As shown in Figure 1, a specific embodiment of the system of selection of optimal viewing angle of video human face by multi-lens camera of the present invention is: the multi-lens camera image capturing system that is made of 6 video cameras, the video human face image library that obtains by 6 camera acquisitions, comprise the video image that 6 people take in different time, different location in the facial image database, image size 640 * 480.Comprise in everyone video image that people's face is the back side, side and front, rotation, inclination and expression shape change image, the digital picture of gathering is carried out determining automatically of optimal viewing angle video camera.Though the human colour of skin varies with each individual, but face complexion presents good cluster characteristic in the color space of YCrCb, and multi-lens camera catches the difference at people's face visual angle, and face complexion area has different ratio sizes with its corresponding rectangular area and this ratio is correlated with the size that video camera obtains people's face quantity of information; The factor is estimated at the visual angle that this ratio is caught people's face as video camera, according to the size of visual angle evaluation factor values, from multi-lens camera, select to have the video camera that factor values is estimated at maximum visual angle automatically, as the video camera of catching people's face optimal viewing angle, its concrete steps are as follows:
(1), starts the multi-lens camera image capturing system: 6 video cameras are installed on the appropriate location, carry out image acquisition, obtain containing the original image of facial image, as shown in Figure 2, (numbering is respectively 1# to 6 video cameras of employing, 2#, 3#, 4#, 5#, the original image of the different visual angles of 6#) catching;
(2), color space conversion, calculate color-values Cr, Cb:, determine the color-values Cr of YCbCr color space, Cb by the red R of RGB color space, green G, blue B three-component:
Cr＝0.5×R-0.4187×G-0.0813×B
Cb＝-0.1687×R-0.3313×G+0.5×B
(3), area of skin color extracts: determine color-values Cr, the threshold value T of Cb and Cr/Cb respectively
1, T
2, T
3, T
4, T
5, the zone with all pixels that satisfy following formula are formed is defined as area of skin color S
S＝(Cr≥T
1∩Cr≤T
2)∩(Cb≥T
3∩Cb≤T
4)∩(Cr/Cb≥T
5)
Wherein, ∩ is " logical and " operational character, T
1～T
5Value be respectively 133,178,78,127 and 1.10.
(4), determining of human face region: the bianry image that satisfies (3) is carried out the connected region search, calculate long and wide ratio of connected region and the hole number in the connected region, the area of skin color that all pixels that satisfy following formula are formed is defined as human face region
F＝(S
1/S
w≥T
6∩S
1/S
w≤T
7)∩(H＞1)
Wherein, S
1, S
wBe respectively the length and the width of colour of skin connected region, T
6, T
7Be threshold value, H is the hole number in the colour of skin connected region, T
6, T
7Value be respectively 0.75,2.5, as shown in Figure 3.
(5), the visual angle estimates the factor and calculates: according to the face complexion area F that extracts, will comprise rectangular area M that all pixels of this human face region form as the face complexion rectangular area, and be calculated as follows the visual angle and estimate the factor
R＝F/M
As shown in Figure 3, the numeral that 6 width of cloth image upper right side are annotated among Fig. 3 is that factor numerical value is estimated at the visual angle, is respectively 0.555,0.563,0.585,0.568,0.667,0.596;
(6), determining of optimal viewing angle video camera: the size of estimating factor values according to the visual angle, selection has the video camera that factor values is estimated at maximum visual angle, as the video camera of best capture people face information, the optimal viewing angle facial image that is obtained for 5# optimal viewing angle video camera as Fig. 4.
Claims (1)
1. the system of selection of an optimal viewing angle of video human face by multi-lens camera, it is characterized in that, has good color cluster characteristic based on face complexion, catch the different visual angles area of skin color of people's face by multi-lens camera, its face complexion area has different ratio amplitudes with its corresponding rectangular area, catches people's face quantity of information according to this ratio amplitude reflection different cameras; Automatically select to have the optimal viewing angle video camera from the multi-lens camera system, its concrete steps are as follows:
(1), starts the multi-lens camera image capturing system: gather the video human face image from different visual angles by multi-lens camera;
(2), color space conversion, calculate color-values Cr, Cb:, determine the color-values Cr of YCbCr color space, Cb by the red R of RGB color space, green G, blue B three-component:
Cr＝0.5×R-0.4187×G-0.0813×B
Cb＝-0.1687×R-0.3313×G+0.5×B
(3), area of skin color extracts: determine color-values Cr, the threshold value T of Cb and Cr/Cb respectively
1, T
2, T
3, T
4, T
5, the zone with all pixels that satisfy following formula are formed is defined as area of skin color S
S＝(Cr≥T
1∩Cr≤T
2)∩(Cb≥T
3∩Cb≤T
4)∩(Cr/Cb≥T
5)
Wherein, ∩ is " logical and " operational character;
(4), determining of human face region: the bianry image that satisfies (3) is carried out the connected region search, calculate long and wide ratio of connected region and the hole number in the connected region, the area of skin color that all pixels that satisfy following formula are formed is defined as human face region
F＝(S
1/S
w≥T
6∩S
1/S
w≤T
7)∩(H＞1)
Wherein, S
1, S
wBe respectively the length and the width of colour of skin connected region, T
6, T
7Be threshold value, H is the hole number in the colour of skin connected region;
(5), the visual angle estimates the factor and calculates: according to the face complexion area F that extracts, will comprise rectangular area M that all pixels of this human face region the form rectangular area as face complexion, and be calculated as follows the visual angle and estimate the factor
R＝F/M
(6), determining of optimal viewing angle video camera:, select to have the video camera that factor values R is estimated at maximum visual angle, as the video camera of best capture people face according to the size of visual angle evaluation factor values.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910198221A CN101702234A (en) | 2009-11-03 | 2009-11-03 | Method for selecting optimal viewing angle of video human face by multi-lens camera |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN200910198221A CN101702234A (en) | 2009-11-03 | 2009-11-03 | Method for selecting optimal viewing angle of video human face by multi-lens camera |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101702234A true CN101702234A (en) | 2010-05-05 |
Family
ID=42157147
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN200910198221A Pending CN101702234A (en) | 2009-11-03 | 2009-11-03 | Method for selecting optimal viewing angle of video human face by multi-lens camera |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101702234A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102254338A (en) * | 2011-06-15 | 2011-11-23 | 西安交通大学 | Automatic obtaining method of three-dimensional scene optimal view angle based on maximized visual information |
| CN103149214A (en) * | 2013-02-27 | 2013-06-12 | 上海交通大学 | Method for detecting flaw on surface of fruit |
| CN105939456A (en) * | 2010-03-30 | 2016-09-14 | 株式会社尼康 | Image processing device and image estimating method |
| CN107680386A (en) * | 2017-11-07 | 2018-02-09 | 潘柏霖 | A kind of intelligent traffic monitoring system |
| CN109214363A (en) * | 2018-10-23 | 2019-01-15 | 广东电网有限责任公司 | A kind of substation's worker's face identification method based on YCbCr and connected component analysis |
-
2009
- 2009-11-03 CN CN200910198221A patent/CN101702234A/en active Pending
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105939456A (en) * | 2010-03-30 | 2016-09-14 | 株式会社尼康 | Image processing device and image estimating method |
| CN102254338A (en) * | 2011-06-15 | 2011-11-23 | 西安交通大学 | Automatic obtaining method of three-dimensional scene optimal view angle based on maximized visual information |
| CN103149214A (en) * | 2013-02-27 | 2013-06-12 | 上海交通大学 | Method for detecting flaw on surface of fruit |
| CN103149214B (en) * | 2013-02-27 | 2015-04-01 | 上海交通大学 | Method for detecting flaw on surface of fruit |
| CN107680386A (en) * | 2017-11-07 | 2018-02-09 | 潘柏霖 | A kind of intelligent traffic monitoring system |
| CN109214363A (en) * | 2018-10-23 | 2019-01-15 | 广东电网有限责任公司 | A kind of substation's worker's face identification method based on YCbCr and connected component analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107578035B (en) | Human body contour extraction method based on super-pixel-multi-color space | |
| CN106874871B (en) | Living body face double-camera identification method and identification device | |
| CN103763515B (en) | A kind of video abnormality detection method based on machine learning | |
| CN101443791B (en) | Improved foreground/background separation in digitl images | |
| CN105187785B (en) | A kind of across bayonet pedestrian's identifying system and method based on choice of dynamical notable feature | |
| CN109359634B (en) | Face living body detection method based on binocular camera | |
| JP4979033B2 (en) | Saliency estimation of object-based visual attention model | |
| US20200250840A1 (en) | Shadow detection method and system for surveillance video image, and shadow removing method | |
| CN103248906B (en) | Method and system for acquiring depth map of binocular stereo video sequence | |
| CN106991370B (en) | Pedestrian retrieval method based on color and depth | |
| Zhang et al. | Detecting photographic composites using shadows | |
| CN104598905B (en) | A kind of license plate locating method and device | |
| CN101702234A (en) | Method for selecting optimal viewing angle of video human face by multi-lens camera | |
| CN102385753A (en) | Illumination-classification-based adaptive image segmentation method | |
| CN106599880A (en) | Discrimination method of the same person facing examination without monitor | |
| CN109587394A (en) | A kind of intelligence patterning process, electronic equipment and storage medium | |
| CN110674884A (en) | Image identification method based on feature fusion | |
| CN106934338A (en) | A kind of long-term pedestrian tracting method based on correlation filter | |
| CN110399785A (en) | A kind of detection method that the leaf based on deep learning and traditional algorithm blocks | |
| CN103729624B (en) | A kind of light measuring method and photometric system based on skin color model | |
| CN109472826A (en) | Localization method and device based on binocular vision | |
| CN109583330A (en) | A kind of pore detection method for human face photo | |
| CN102542304B (en) | Region segmentation skin-color algorithm for identifying WAP (Wireless Application Protocol) mobile porn image | |
| CN103120581A (en) | System and method for integrating heartbeat measurement and identity identification | |
| Garg et al. | Color based segmentation using K-mean clustering and watershed segmentation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Open date: 20100505 |