A Face Detection Method Based On Color And Template Matching

Face detection is an image processing course, ascertaining face's location, size and amount in an image. And face detection is the precondition of face recognition, human-computer interface and intelligent scene supervision system. Research on face detection will be paid more attention, following with computer application being popular and its capabilities improved and researches growing up in image processing and mode recognition field.It aimed to face detection on the complex background. A face detection method combined cursory detection and validation was proposed on the base of analyzing the advantages and disadvantages of the face detection methods. Improved complexion and thresholds of lip color were used in cursory detection, and template matching method was used in validation.The main contributions are as follows:Complexion is important character of face. Through sampling statistic and clustering analyzing, subsection Gauss model in Ycbcr color space about complexion segmentation was proposed. In Ycbcr color space, a Gauss model about complexion distributing was built up, then we can get probability image about complexion resemble degree. Twain-value image was gained according to the best threshold. At last, some potential complexion areas were got, through complexion segmentation.Lip color is an obvious character in the face. Ternary threshold was received through lip color sampling statistic and clustering analyzing. Search lips in above potential complexion areas, and ascertain the size and rotation degree of face according to the size and position of lip.An average face template matching was used for face detection in this thesis, after mastering the template matching principle and the realization arithmetic. The templates were handmade, and there are frontal, lateral and various sizes of templates. This arithmetic was used for validating in face detection stage, especially a solving method was proposed for non-upright faces.At last, a new face detection arithmetic was proposed, through analyzing advantages and disadvantages of color segmentation and template matching arithmetic in face detection. Template matching was used for validating in potential face areas, and received the final results.