Feature-level image fusion is to synthesize the feature information obtained after preprocessing and special extraction of the source image, such as edges, shapes, contours, regions, etc. Feature-level image fusion is information fusion at the level between the gates. It not only retains a sufficient amount of important information, but also compresses the information, which is conducive to real-time processing. It uses parameter templates, statistical analysis, pattern correlation and other methods to complete the functions of geometric association, feature extraction and target recognition, so as to facilitate the system decision. The typical feature information generally extracted from the source image is line type, edge, texture light, similar brightness area, similar depth of field area, etc. In the feature-level image fusion process, since the extracted features are directly related to the decision-making analysis, the fusion result can give the maximum feature information required for the decision-making analysis.