Abstract
Nowadays, for better security, we must select a more difficult password in the ordinary passphrase method, but it is hard to remember it. We research and develop a novel biometric authentication to compare the advantage of the technique with typical methods. In the previous works, we presented a personal identification method with the image of the hand shapes on the keyboard. It is available to construct in a low-cost way and it is less stressful for users than ordinary passphrase method. In this article, we described the proposal of a biometric authentication method based on Bhattacharyya distance and Chi-square test between color histograms in hand image, and also we discussed the implementation using OpenCV (Open Source Computer Vision Library) for the practical use. In addition, we experimented to evaluate the proposal method, and as the result, it was confirmed the best precision of the authentication using the all 3 channels in HSV color space was 100% under the condition with high luminance lighting, and also it was over 95% under the conditions using zoom. Both calculations on Bhattacharyya distance and Chi-square test were almost similar results. It was suggested that the implementation using of our method was valid computational cost.
