The Research And Design Based On The DSP Face Recognition System

With the progressive development of the society, people’s demand for accurately determining a person’s identity information is becoming higher and higher in the field of identification. The traditional authentication methods such as keys, magnetic cards can not meet such high demand, while face recognition method has its own advantages so that it has got rapid development. The traditional face recognition system mainly depends on computer hardware and operating systems. Its cost is high, real-time performance is bad and miniaturization is hard to realize. While face recognition system based on DSP(Digital Signal Processor) is combined with the design of embedded systems. It has a lot of advantages, such as low cost, small size and high reliability. Therefore, this paper is about my research on face recognition system based on DSP, the specific contents are as follows:Firstly, this paper introduces foreign research status of face recognition and basic links of face detection contained in face recognition system.Secondly, this paper introduces the basic means of image preprocessing. We use the face detection algorithm to detect human face. YCb Cr color space is used for color segmentation on the human face detection algorithm and simulation test is completed in Matlab.Then, we have discussed the advantages and disadvantages of several commonly used algorithms for face recognition. The improving PCA algorithm is used in this paper by adding the wavelet transform in order to reduce the dimensionality of the image and enhance the speed of recognition. Face recognition algorithm test is simulated in Matlab. The recognition effect of ORL face image database and self made facial image database are given out.Finally, this experiment builds face recognition system based on DSP. The characteristics of DSP hardware and CCS software are described in this paper, and the flow of hardware and software are provided. The camera is controlled by DSP to capture face image. And the algorithm transplant is achieved from Matlab to DSP. This test has given out the effect of the recognition and verified the feasibility of the system.