Human Identification Based On The Ear And Profile Face

Due to the tremendous development of computer industry, biometric identification technology has made enormous progress over the last few years. The special effects that people can only see in the movie now become reality. It’s hard to avoid negative impact by merely using one biometric identification technology due to the instability of feature extraction. Thus multi-biometric feature fusion and recognition technology is adopted to avoid the shortcomings of using single feature, which have been explored to mass applications presently.In this paper, face profile and human ear adopted for multi-biometric feature fusion and recognition technology is also studied. We use an algorithm named force field transform for the extraction of the ear feature while Gabor algorithm is simply used for that of profile face. Accordingly, both algorithms are fused and the recognition rate turns out to be higher than using single biometrics.AdaBoost algorithm is used for the profile face and the ear detection. With the use of OpenCV toolkit and VS2008integrated development environment, we firstly design interface and get the coordinates of profile faces and human ear. The next step is to correct the ear using force field transform where down sampling is used in order to reduce the calculation time. Then by using force field transform with canny edge detection and morphological processing, we get the ear area of binary image. With setting the rotation angle and calculating the minimum enclosing rectangle which bounding the ear, the index of the biggest aspect ratio will be identified the best angle.Regarding the force field transform, we get force field vector diagram first. Then we calculate the divergence of the vector diagram and get the divergence image. The divergence image in fact depicts the location of the potential energy and potential energy trap of the human ear. Then K-means algorithm is applied to divide different human ears. Finally, we use SIFT algorithm for the extraction and matching of human ear features.We use color segmentation algorithm with a regard to the profile face. Firstly the RGB space is transformed to HSI space and YCbCr space. Then the region of interest is extracted by using H, Cr and Cb components. Finally, Gabor feature is used for the matching process. To sum up, feature fusion algorithm is used for the human ear characteristics and profile face features. Experimental results show that SIFT algorithm can better describe the human ear divergence characteristics than other methods. Furthermore, adopting multiple biometric fusion technique can achieve higher recognition rate than using single biometric.