An Efficient Approach for Automatic Palmprint Classification

This paper proposes an efficient approach for automatic palmprint classification based on principle lines that consist of a life line, a head line and a heart line. The extracted principle lines need not be perfect and our features and the criteria for classification are obvious and simple. Experiments are done with Visgraph database and our CU-CGCI hand database. Each database consists of 1,000 palmprint RGB color images from 100 users. Our method classifies palmprints into six groups. The distribution of categories 1-6 in Visgraph database are 28.8%, 34.4%, 24.2%, 3.8%, 4.5% and 4% whereas they are 34.7%, 27.5%, 22.6%, 5.7%, 3.4% and 5.9% in our CU-CGCI hand database. The palmprint distribution from our method is more even with the most population being around 34%.