Naresh Kumar Kachhi
Department of Computer Scienc and Engineering, Indian Institute of Technology Kanpur, Kanpur India, 208016
Figure 2: Schematic diagram of corresponding sub-image matching (a) (b) (c) ACC(%) FAR(%) FRR(%) IITK N-Wgt Wgt 88.65 98.74 4.27 1.81 18.42 1.93 CASIA N-Wgt Wgt 98.66 98.74 0.74 0.58 1.93 1.93 PolyU N-Wgt Wgt 99.29 99.31 0.24 0.28 1.16 1.08
Table 1: Performance of the proposed system of the proposed system and which are stable for the user, The extracted palmprint region with different orientation of placement for the same user remains the unchanged. Hence the proposed palmprint extraction procedure of the system makes the system robust to rotation. The extracted palmprint is subjected to enhancement procedure to obtain uniform brightness and contrast enhanced image. The enhanced palmprint portioned into equal sized sub-blocks of size m × m and zernike moments of each block is extracted as their features. The zernike moments of corresponding blocks of live and enrolled palmprint are matched ( as shown in Fig. 2 ) using Euclidean distance and weighted based on average discrimination level of the sub-block relative to other blocks. The matching score of each block is fussed using summation rule and decision of acceptance rejected is made based on predeﬁned threshold. Palmprint has rich texture. so each sub-image can be classiﬁed into into occluded / non-occluded based on its randomness (Entropy). Since the extracted Zernike moment of a sub-image is independent of other subimages, ignoring the features of the occluded sub-images makes the system robust to occlusion. The system is evaluated on Indian Institute of Technology Kanpur (IITK) database of 549 hand images from 150 users, The Hong Kong Polytechnic University (PolyU)database...