Key Technology Research Of Face Feature Extraction System And Implementation Based On DSP

Face recognition is a challenging research subject in the field of pattern recognition and artificial intelligence, feature extraction is a very important part in face recognition. Face recognition involves to artificial intelligence, computer graphics learning, pattern recognition, computer vision and other fields, and has a very wide range of application in the fields of public safety, commercial enterprises, in and out establishments places, such as police forensic recovery, public security, access control, certificate identification, video monitoring, etc.At present, most applications of face recognition are still based on a PC, big volume and cost is high, it is difficult to meet the practical application requirements. Face is non-rigid objects with rich facial expressions, so when face facial expressions is changed, face recognition results will be affected. Therefore to eliminate the impact of changes in the expression on the face recognition process is essential. The studies purpose of this paper is to find a feature extraction algorithm, that not only can satisfy the realtime requirements and suitable for facial expression changes, but also achieve simple and can improve recognition rate, that proposed a face feature extraction system solution based on DSP. This pape first expounded face recognition research background, present situation and its application, introduces some digital image processing techniques before facial feature extraction and several commonly used feature extraction algorithm, on the basic of analysis and comparison of several common character extraction algorithm, then proposed feature extraction method based on wavelet transform and improved eigenfaces, first face images extracted low-frequency components by wavelet transform, and low-frequency image is mapped to a low-dimensional space through PCA transform, then in the low-dimensional FLD space using feature extraction methods. Finally, to achieve the proposed algorithm face recognition system based on DSP platform. And the program were optimized to ensure the system's real-time.