Multi-algorithm fusion method for bimodal infrared image difference characteristic index measurement

Abstract

The invention discloses a multi-algorithm fusion method for bimodal infrared image difference characteristic index measurement. The multi-algorithm fusion method comprises the following steps that: firstly, selecting difference characteristic types between infrared polarization and a light intensity image, wherein the difference characteristic types are mainly luminance, details, edges, outlines and the like; according to the difference characteristic types, selecting local energy maximization, non-subsampling shear wave and multiscale guidance filtering to independently carry out fusion on a source image; independently calculating the local mean value, the local Laplacian energy and the local standard deviation of two classes of image; utilizing the local mean value, the local Laplacian energy and the local standard deviation of two classes of image to calculate each difference characteristic index measurement; and constructing a difference characteristic index measurement covariance matrix, calculating the feature value and the feature vector of the covariance matrix, and selecting the feature vector corresponding to a maximum feature value as each algorithm weight to keep the function of the edge and the outline feature of the high luminance, multiple details and high definition of infrared polarization and light intensity images while multiple algorithms are fused while multi-algorithm fusion is realized. Therefore, the fusion effect of multi-algorithm fusion can be obviously improved.

Description

technical field [0001] The invention belongs to the field of infrared image processing, and specifically relates to a multi-algorithm fusion method for measuring difference characteristic indices of dual-mode infrared images. Background technique [0002] Infrared polarization and light intensity imaging use the polarization and intensity attributes of infrared rays to detect targets respectively. The two modal images are highly complementary, and their fusion can describe the target information more comprehensively and further enhance the detection capability of the system. However, infrared polarization and light intensity images include different features such as brightness, details, edges, and contours. It is difficult for a single fusion algorithm to achieve fusion of multiple differential features. Therefore, combining different fusion algorithms with complementary advantages and performances is beneficial to brightness, detail It has become a research hotspot in the f...

AI Technical Summary

Problems solved by technology

[0004] In order to solve the problem that the existing multi-algorithm fusion of infrared polarization and light intensity images cannot simultaneously retain the characteristics of high brightness, multiple details, and high-definition edges and contours of the source image, resulting in poor fusion effect, a method is proposed. The multi-algorithm fusion method of dual-mode infrared image difference feature index measurement selects the algorithm according to the difference feature type to improve the complementarity between the fusion algorithms; establishes the exponential difference feature measure and accurately measures the difference feature amplitude; uses the difference feature Based on the index measure, the weight of the algorithm is determined by decorrelation processing of the difference feature measure, and the proportion of the fusion results of each algorithm in the final fusion image is clarified, so as to prevent excessive loss of image features and oversaturation, and finally achieve multi-algorithm fusion while retaining infrared polarization and The fusion of features such as high brightness, multiple details, and high-definition edges and contours of light-intensity images can significantly improve the fusion effect of multi-algorithm fusion

Images