Infrared and Visible Light Image Fusion Algorithm Based on Compressed Sensing

Abstract

The Nonsubsampled Contourlet Transform (NSCT) is translation invariant. It also has good line singular characteristics capturing feature in the image processing. The compressed sensing theory has good noise suppression capability. On the basis, this study proposed a novel compressed sensing image fusion algorithm in NSCT transform domain for the infrared image and visible light image fusion. This algorithm adopted the pixel feature energy weighted fusion rule and the neighborhood variance feature information fusion rule for the low-frequency coe‐cients and high-frequency coe‐cients fusion. The compressed sensing algorithm used the Toeplitz matices to observe the high-frequency sub-band coe‐cients. The experiment results show that this method efiectively decreases the amount of processing data, improves the convergence rate, and raises the fusion efiect.