Palmprint authentication in smartphone using Phase-Only Correlation method

Personal authentication in a system security perspective is very important. To overcome the problems of conventional authentication, biometric authentication has been developed that uses the characteristics of human nature. This paper proposes software application of palmprint authentication in smartphone. Authentication is done in four stages: data acquisition, preprocessing, feature extraction and image matching. Band-Limited Phase-Only Correlation (BLPOC)-based local block matching method is selected for image matching because it has the highest accuracy among several methods that have been tested previously. The authentication test results on 30 sample people of each 5 palmprint images indicate that the constructed software application provides a good performance with matching score value of 99.964 % and EER 0.18% in the threshold value of 0.277. Login testing on 10 people samples obtained that the accuracy of 100% with a fixed Region of Interest (ROI). This system has a high tolerance for rotation of 5 ° up to 7 ° and translation displacement tolerant up to 64 pixels in both of x-axis and y-axis.