A Fusion Method of Infrared Image and Low Light Image

Abstract

The invention relates to a novel infrared image and low-light-level image fusion method. The method comprises the steps of firstly performing multi-scale and multi-direction NSCT transform decomposition on an infrared image and a low-light-level image to obtain low-frequency and high-frequency sub-bands of the infrared image and low-frequency and high-frequency sub-bands of the low-light-level image; on the low-frequency sub-bands, preprocessing a low-frequency coefficient by adopting a smooth linear filter firstly, and performing fusion by adopting a local energy ratio and local energy weighting combination method, wherein a weight value of the smooth linear filter located in the center of a mask is greater than the weight values of the smooth linear filter located on the four sides and the four corners, the weight values of the smooth linear filter located on the four corners are minimal, and an output response of the smooth linear filter is a weighted average in a template; and on the high-frequency sub-bands, preprocessing a high-frequency coefficient by adopting a second-order differential-based Laplace operator firstly, and finally performing NSCT multi-resolution reconstruction on a fusion coefficient to obtain a fused image, wherein a Laplace filter belongs to isotropic filters.

Description

technical field [0001] The invention belongs to the technical field of image fusion, and in particular relates to a fusion method of an infrared image and a low-light image. Background technique [0002] In the field of image fusion, different types of sensors can be used to provide different types of image information for the same detection area, so as to obtain a more accurate image of the same scene that meets the detection needs. In night or low-light detection occasions, infrared sensors and low-light sensors are often used together. The infrared image receives the infrared information of the target, and the low-frequency information of the target is more prominent; the signal-to-noise ratio of the low-light image is high, and the detailed information provided is more abundant. . Through the image fusion algorithm, the characteristics of each sensor can be used to obtain more ideal target indication scene information. [0003] Traditional image fusion algorithms have ...

AI Technical Summary

Problems solved by technology

[0003] Traditional image fusion algorithms have disadvantages such as poor contrast between the target and the background, unobvious target features, and weak detail information.

For example, Qian Xiaoyan et al. (Qian Xiaoyan, Han Lei, Wang Bangfeng. Fast fusion algorithm for infrared and visible light images [J]. Journal of Computer-Aided Design and Graphics, 2011,23(7):1211-1216.) proposed Combining local histogram equalization to enhance infrared and low-light images, and using median filtering to denoise, the fusion is fast and the details are prominent, but the target is not obvious; Xing Suxia et al. (Xing Suxia, Xiao Hongbing, Chen Tianhua, etc. based on target extraction and Research on NSCT image fusion technology [J]. Optoelectronics Laser, 2013 (3): 583-588.) proposed to use information entropy to segment infrared images to extract targets, and use NSCT algorithm to enhance the low-frequency components of low-light images for fusion. Good, but the details of the image are slightly weaker during its enhancement process

Images