YCrCb Color Space Face Detection Algorithm Based On The Design And Implementation

The endless stream of new science and technology in the21st century, in biometricidentification technology, face set other technology has made considerable development.Face device as a new identity to set up another technology, commonly used passwords,card recognition technology has no need to carry is not lost, difficult to counterfeitadvantages. Face detection face recognition system, which includes algorithm design,image processing, pattern recognition, computer graphics, artificial intelligence, andother related disciplines. Face detection data for face recognition system to provide corecritical data, so the effect of the face detection directly determines the level of laterecognition rate and recognition speed, improve face detection accuracy and detectionspeed is to improve the face recognition effect the key. Therefore, the study facedetection algorithm to improve people face set have important significance for thereliability and accuracy of the other systems.The main problems of the current face detection system, this paper carried outin-depth research in the following areas:1.Improved the conventional RGB color space to the YCrCb color spaceconversion algorithm. In this paper, by analyzing the contents of the the traditional RGBconversion YCrCb algorithm proposed an improved integration of linear and non-lineartransformation of the RGB color space to the YCrCb color space conversion algorithm.And proved by experimental data obtained by the improved conversion algorithmvariation range of the Cr, the CB value is more concentrated, a higher recognition rate.Horizontal and vertical projection and template matching combination of face detection.2. Horizontal and vertical projection and template matching combination of facedetection. In this paper, the combination of the vertical projection of the template, thelevel of regional connectivity matching method face detection, fixed dimensions shrinkto the size of the face, and in the vertical and horizontal direction striping certain areascalculated for each region pixels the number of points generated projection mapprojection map, with standard front face contour contrast to the similarity as the basis todetermine whether human face. In addition, the the face template color distributioncharacteristics establish color matrix template based template matching colordistribution, determined to match the rate of the level of the human face. At last we used Microsoft Visual Studio2005design, face detection reliability testsystem, and the theoretical research and application in the design of the system.