Biometric recognition based on the texture along palmprint principal lines

Palmprint recognition has been in the focus of biometric research over the last ten years. Here, a biometric system for palmprint recognition based in the texture of palmprint principal lines is proposed. Two novel methods for detection of principal lines using graph search algorithms are developed. One of the schemes performs particularly well at detected fully connected lines, which is an advantage over other methods that only employ edge detection algorithms. The lines detected with the developed scheme are used to extract line textural information using Haralick’s features. The concept of texton dictionary is used to find representative textures in principal lines. A palmprint is then represented by the histogram of texton frequencies. The performance of the method is inferior to other documented procedures but, considering that no spacial information is present in histogram representation, and that each histogram contains a small amount of data, the achieved accuracy of 85% highlights the importance of texture in palmprint recognition. The implemented methods are innovative in palmprint recognition and constitute a framework for future approaches in this field.