A Novel Automatic Weighted Image Fusion Algorithm

Abstract

A new automatic weighted image fusion algorithm is proposed. The current weighted averaging image fusion algorithm has good rapidity, but its fusion quality depends on the choice of weight values to a great extent. In order to choose weight values more reasonably, K-L transform is adopted, which has advantages in multi-component information reduction. The available information in source images is concentrated to new independent principal component matrixes. The redundant information from different source images is decreased and the weighted coefficients are thereby determined. Comparison experiment is performed with infrared and visible images. Experimental results of subjective and objective evaluation show that, the fused images based on the new algorithm have improvements in image clarity, information amount and so on. Furthermore, the algorithm maintains the characteristic of simplicity and practicability of existing algorithms and can be applied in the situation when high speed image fusion is required.