A New Multi-source Image Sequence Fusion Algorithm Based on SIDWT

Abstract

A new fusion method of infrared and visible video sequence is proposed based on the shift-invariant discrete wavelet transformation (SIDWT). Firstly the approximate target regions of each single-frame infrared image are detected by weighted information entropy. Secondly the new fusion algorithm based on SIDWT is proposed to fuse the target regions of the registered infrared and visible images. Lastly the fused target regions are fused with the background regions of visible sequences. The comparison experiments with the traditional static fusion algorithms show that the new proposed algorithm not only meets the real-time requirements of video fusion, but can effectively avoid the problem of the weakened target information in the fused image, which realize the complementary advantages of the two types of source images.