Research On Palmprint Image Processing And Recognition Methods

With the rapid development of information technology, the reliability and safety of various systems are put a higher demand. The essential means of biometric identification technology has become a hot research topic at home and abroad. Palmprint identification is a new biometric technology which has lower cost, higher stability, recognition speed, noise immunity and acceptability. To improve the certification system reliability and security, this technology has a very important research significance and practical value.Palmprint identification technology includes image acquisition, preprocessing, feature extraction and matching modules. In palmprint image acquisition module the quality which is obtained is very important. In palmprint preprocessing effective region of interest to obtain is critical. In palmprint feature extraction the core technology is to extract efficient and distinguishing feature and reduct dimensions. In palmprint feature matching the vital technology is how to select and design matching algorithm. Therefore, how to improve the efficiency of palmprint identification have been complex and meaningful topics to the palmprint identification technology. Base on the opinion and previous research, this dissertation deeply study palmprint identification technology. The main work of this paper is as follows:(â…°) For the features of palmprint identification technology, the operating mode and palmprint identification system performance assessment indicators have been deeply study. And on the basis, the operation mode and system performance assessment indicators are given. It has been study principal component analysis algorithm, independent component analysis algorithm, linear discriminant analysis algorithm, non-linear dimension reduction algorithm of approximate linear techniques. Furthermore, it has been analyzed of the Gabor wavelet technology with good time-frequency characteristics. These technologies provide a theoretical basis for palmprint identification technology study.(â…±) Because the placed positions of palm and finger aperture are different, it can result in the difficult problem of positioning segmentation. This paper proposes palm positioning segmentation algorithm to find automatically points. Firstly, palm images are coverted binary image according to the palm outline characteristics. Then, the fingers and palms are separated and the two positioning points are found. Finally, the palm of a certain degree of freedom is normalized and positioned, and the region of interest is extracted.(â…²) In terms of small sample problems which exist in the subspace palmprint identification technology and the direct irrelevance to optimize linear discriminant analysis criterion function and the maximization of the recognition rate, this paper proposed a two-dimensional linear discriminant dimension reduction algorithm which merges two dimisional principal component analysises. This algorithm is a direct projection of the palmprint image matrix and is achieved very good results for dimension reduction which can eliminate the relevance of rows and columns to get the best projection vector and extract optimal discriminant information.(â…³) According to the characteristics of Gabor transform, the time-frequency domain characteristics of two-dimensional Gabor wavelet and the dimension reduction method of principal component analysis are deeply studied. The two dimensional principal component analysis algorithm based on the Gabor is designedl. Firstly, the algorithm is done the convolution between Gabor filters of different scale and direction and training samples of all palmprint images. Therefore the high-dimensional Gabor feature space is obtained. Then less the coefficients to characterize the image is extracted under the Gabor feature space dimensionality reduction through principal component analysis.(â…´) Since the traditional two-dimensional linear discriminant algorithm exists to calculate the high complexity and whithin class distances are limited, an improved two-dimensional linear discriminant algorithm is proposed. Then an improved two-dimensional linear discriminant algorithm is designded which combine with Gabor wavelet and principal component analysis...