Abstract
It is found in infrared military targets’detection that some of the collected images are greatly affected by the environment, and they are still not provide targets’detailed information after preprocessing, which limits the detection effect. In this paper, we establish a military infrared-visible dataset and propose a military image fusion strategy based on composite decomposition and multi-fuzzy theory. Firstly, the source infrared and visible images are decomposed by using the two-scale and Latent Low-rank representation composite method, and the underlying optimal information of the images is mined. Secondly, for low-frequency detail images, the Gaussian fuzzy function is used to adjust the visual saliency map weighting function; for low-frequency salient images, the Cauchy fuzzy function is used to adjust the image energy weighting; and for high-frequency images, an improved intuitionistic fuzzy set function is used as the fusion rule as proposed. Finally, four sets of typical images are used to test the proposed method and make both subjective and objective comparisons with the other four mainstream fusion methods. The application experiment results show that the proposed military image fusion strategy can more effectively improve the clarity of the data source and thus improve military targets’detection accuracy.