Face Recognition with Multiscale Data Fusion of Visible and Thermal Images

Abstract

This paper presents face recognition with activity level fusion of visible and thermal image using multiscale decomposition. Image fusion combines images with different features to obtain a single composite image with extended information for better recognition performance. The proposed fusion technique adoptively controls the fusion ratio between visible and thermal information with multiscale analysis. Experimental results demonstrate that the proposed method effectively overcomes the weaknesses of visible and thermal images. The performance of the fusion is evaluated by classification rate with support vector machines