Wavelet Transform In Iris Recognition Applications

With the development of the society, security consciousness of human beings is higher and higher, so personal recognition techniques have turned into a hot issue of the advanced science and technology. According to the physiological characteristics of the body which possess universality, distinctiveness, permanence and collectability and so on, personal recognition has become an important topic of the scientific research institutions which pursue biometric recognition at present. Many results of this thesis come from the 863 Program which studies mainly new methods and techniques of multimodal biometric personal authentication based on visual perception and cognitive mechanism (project number: 2007AA01Z423).Iris recognition has more advantages than other biometric recognition, such as high recognition accuracy, non- invasion and difficult forgery etc. According to geometry and physiological characteristics of iris, iris image segmentation and characteristic extraction are studied in this thesis.Iris image pre-processing is a basic and key of the iris recognition system after iris image grabbed, and the pre-processing quality of iris image will be influenced characteristic extraction and matching. According to some traditionally pre-processing method, which computation capacity is big, memory expense is big and search time is long etc, we present a new pre-processing method based on gray curve and improved Hough transform, which method can be achieved simply and the memory space expense is very small. Meanwhile we have carried out the iris image pre-processing experiment on the CASIA database under the Matlab platform and obtained some good results.Iris has abundant texture characteristic, the iris texture characteristics'expression and extraction are very complicated. Meanwhile they are also one of the key points and difficulties of the iris recognition. Wavelet analysis is internationally recognized up to minute tools for analyzing time-frequency due to the"adaptive feature"and"mathematical microtelescope feature". Texture characteristics are extracted by using wavelet transform. And properties of the wavelet transform, high dimension wavelet, and multi-resolution analysis are systemically introduced. Iris characteristic extraction based on wavelet transform zero-crossing point, tree-type wavelet transform and 2D wavelet transform are systemically introduced and experimented. At the iris match stage, we utilize Matlab language (platform) to carry on simulations and experiments. Iris characteristic extraction method based the improved Daubechies-4 wavelet of the 2D wavelet transform is put forward and experimented, and get some good phase results.