Improved differential box counting with multi-scale and multi-direction: A new palmprint recognition method

A novel palmprint recognition called improved differential box counting (IDBC) with multi-scale and multi-directional is proposed in this paper. At present, fractal dimension as feature vectors cannot accurately reflect the characteristics of image information, and the algorithm complexity is high. Firstly we set out to improve the method of differential box counting, putting forward fractal characteristics as eigenvector. Next, for effective description of accurate orientations and scale, we combine multi-scale and multi-direction of Gabor and Curvelet with IDBC (GIDBC and CIDBC), further proving that Curvelet is more effective than Gabor for palmprint recognition. Experimental results on PolyU palmprint experiment show that the proposed method can obtain state-of-the-art recognition accuracy (99.78%), reduce algorithm complexity and meet the real-time requirements that time of feature extraction and matching is less than 300 ms.