Human Face Feature Extraction Based on a Partition Threshold Model

Image feature extraction technology has been applied to all aspects of our lives, such as science, education, aerospace, national defence, and industrial production. For example, a painting robot requires a face extraction method that meets human aesthetic standards. This paper mainly aims to solve the problem of conditional binary extraction of the human face against changeable and complex backgrounds, and under different lighting environments, in order to achieve better human face image feature extraction and higher image processing speeds - the average time to process one image is 3.3 seconds. Different from the traditional partition method, which only uses the pixel value information, our method combines a skin colour model, the edge structure of the face, and pixel information in the partition. The partition threshold model proposed in this paper is used to calculate the appropriate threshold value in each partition, so that each image partition gets the corresponding threshold value. Experimental results demonstrate the robustness of the scheme.