Evaluation method of camouflage effect based on image feature similarity/fusion degree

Abstract

The evaluation of camouflage effectiveness is crucial for validating and optimizing camouflage strategies. However, most evaluation approaches based on image features ignore the impact of locational information and the data normalization method on the results. In this study, brightness, color, and texture features are selected as evaluation indices according to the principles of visual perception characteristics and camouflage index selection, and a novel camouflage evaluation method based on image feature similarity/blending is proposed. This method fully considers the spatial location and statistical properties of pixels. Further, a weighted Euclidean metric is introduced to achieve comprehensive camouflage effect evaluation by calculating the brightness, color similarity, and texture integration between the target and the environment. A Sigmoid function normalizes the calculation results. The observer experiment verifies the method, demonstrating that its results align with human visual judgments, thereby proving its scientific rationality.