Vein Image Feature Extraction Based on Box Counting Dimension and Multiple Wavelets
Ding Wan-qiang · Journal of Northeastern University · 2010
A global/local feature extraction algorithm is proposed on the bases of box counting dimension and multiple wavelets to improve the matching rate in vein recognition process.In this algorithm the vein texture roughness is extracted as global feature by the box counting dimension,with the local features of vein image extracted by multiple wavelets decomposition and described by the encoding with one-dimension coefficients and that with multi-scale coefficients.It is proved by the test results of existing vein images that the proposed feature extraction algorithm is effective,and the recognition accuracy of its relevant matching algorithm improved.