The Application of Graph Kernels in Face Recognition
Jiang Qiang-rong, Huan-jun Chen, Bo Liu · 2012
Face recognition has become one of the hot research topics in pattern recognition and image processing in the recent several years, as a result of the wide application in the areas of security control and human-machine interaction. And it has been recognized as the most simplest and non-intrusive technology without hazardous problems, compared to other biometric recognition technology, such as fingerprint recognition, iris recognition and et al. Many scholars dedicated to do research and propose various methods to improve the accuracy and speed of face recognition. As most real-world data is structured, the kernel method, investigated for various kinds of structured data after the successional propose of statistical learning theory and support vector machine, reduced computation and performances well. One of the most widely used tools for modeling structured data are graph and an interesting and important challenge is thus to investigate kernels on instances that are represented by graphs. But so far, only very specific graphs such as trees and strings have been considered. In this paper, we propose two graph kernels including maximum spanning-tree kernel and cycle kernel based on minimum spanning-tree. The experimental results show that the two graph kernels achieve good performance.