Effect of Registration Error on Infrared Image and Visible Image Fusion Quality

Abstract

In this research, the effect of registration error on infrared image and visible image fusion quality was discussed. The registration errors from different directions were adopted to simulate the real registration error. Meanwhile, the infrared and visible images under different registration error were analyzed. The result showed that among the Laplacian Pyramid fusion, Gradient Pyramid fusion, Contrast Pyramid fusion and SIDWT with Haar fusion, the Laplacian Pyramid fusion can better maintain the infrared and visible image fusion quality under a certain registration error. Moreover, when dealing with different registration errors, single integration indicator was not sufficient in describing the infrared and visible image fusion quality. Instead, the multi integration indicators shall be taken into consideration.