Face detection process is easily affected by facial details,shape,expression,posture,etc,and the skin color feature,as one of the most recognizable information in human face,is relatively stable.This paper implements face detection and labeling based on the skin color model.The skin color model is con-structed in the YCbCr color space,and the influence of environmental factors such as uneven illumination and brightness changes in the process of skin color extraction is eliminated through light compensation,and the noise in the non-face area is filtered based on morphological filtering.Combined with the geometric fea-tures of the face area,the face positioning and box check mark are realized.The test results of face images under various conditions,such as uneven illumination,occlusion,and large angles,show that the imple-mented algorithm can better achieve face detection and localization.
skin color modelcolor space conversionlight compensationface geometric featuresface detection
万方数据
维普
