Research On The Extraction Technology Of Complex Background Image Preprocessing And Interest Region Of Palm

With the penetration of information technology and networking technology in today’s society, people pay more and more attention to the security of information and system, more and more recognition techniques based on biological characteristics have been widely studied and applied. In a large number of biometric identification technology, palmprint recognition technology has received more and more attention recently. Compared with the face recognition technology, iris recognition technology, retina recognition technology and other biometric recognition technologies, palmprint recognition has advantages in the characteristics of stability, extraction and recognition rate. However, the collecting process of palmprint image have higher requirements on the collecting background, illumination, angle, sharpness and other conditions.In recent years, digital camera and mobile phone camera imaging level unceasing enhancement. Untill now the sharpness of photographes shooted by digital camera and mobile phone is qualified. Therefore, this paper focus on the complex background photograph taken by the digital camera and mobile phone(complex background, illumination is not fixed, the palm angle and size is not fixed), and focuses on the research of Palmprint Feature Extraction of region of interest and the sense of palm lines, as well as palmprint line normalizing. The main work of the study:(1) The palmprint image preprocessing. Selection YCb Cr color space for skin color detection, and proposed a calculation method of improved local inscribed circle, which can extract palmprint region of interest effectivly.(2) Discussed and compared several traditional differential operator, and a method of improved Sobel algorithm.(3) Proposed a method for feature extraction of palmprint based on mathematical morphology and template matching. The method uses mathematical morphology(Bottom-Hat transform) as the extraction of ridge valley value information, by direction template were extracted from the line on 4 directions, and finally adopts an adaptive threshold method of 0-1 valued image processing, to extract the palmprint feature. Experiments show that the method can extract the feature lines continuously and completely.(4) Proposed a method through the geometric features of palmprint, which judgments the palm of your hand. Proposed a method for normalization of arbitrary direction. The experimental results show that, this method can extract the palmprint mainline of complete information, and accurate judgments about the palm of your hand, and at the same time, then normalize the palmprint direction.