A New Way for Extracting Region of Interest from Palmprint by Detecting Key Points

Extracting region of interest (ROI) from palmprint is the important and key link of palmprint recognition. The quality of ROI directly determines the recognition rate. We propose a new algorithm for extracting palm ROI, and show the validity of our algorithm with numerical experiments on PloyU database and CASIC database, achieving recognition rates respectively 100% and 99.527%. The core idea of our algorithm is to obtain the key points from the palmprint. To get the first key, we firstly construct a circle with radius r slide along the edge of the palm, then calculate the center of the circle when the intersection area of the circle and the palmprint reaches maximum, so that the center is the first key point we need. To get the second key, we remove the first key point and its neighborhood, then detect the second key point using the same method. Other key points are obtained using the same method. In the step of generating ROI, the length of sides of the square ROI is based on the approximate half width of the palm.