基于Gabor滤波器和不变矩的掌纹识别算法研究计算机科学与技术专业论文.docxVIP

Abstract Palmprint identification, which means using the prints of human palms to perform personal recognition, is a new technology of biometrics. Compared with the previous biometrics technologies, palmprint identification has his advantages which make it have a good future. In this paper, there are three algorithms given. First, it is a palmprint verification algorithm based on wavelet transformation and moment invariant. Second, it is also a palmprint verification algorithm based on Gabor filter and moment invariant. At last, it is a fusion method of 2D palmprint gray scale image and infrared image. In chapter two, we briefly introduce a previous attempt of palmprint verification algorithm. It utilized the multi-resolution technology to analyze the palmprint image in different level and reconstructed a similar image using the horizontal, vertical and diagonal coefficients. This algorithm did not get a good result, but it is very helpful to the following work, so at the beginning of our paper, we illuminate it shortly. In chapter three, an effective algorithm was presented, which is a algorithm of palmprint feature extraction based on Gabor filter and moment invariant (MI). First, the algorithm performs wavelet transform on the original region of interest (ROI) of palmprint images to get the approximation image (ALROI). Later, we apply the Gabor filter to obtain the texture information of the ALROI, and then compute each pixel’s magnitude of the transformed image to get a Gabor Magnitude Picture (GMP). Afterwards, we calculate the MIs of the GMP and the MIs are used as features for palmprint identification. Our experiment on open database downloaded from PolyU Biometrics Research Center obtains very high right recognition rate. In the last chapter, we introduce a fusion algorithm of 2D gray scale palmprint image (palmprint for short) and infrared palmprint image (palm vein for short). This algorithm also uses the wavelet transform on each original region of inter