A camouflage generation algorithm based on modified K-means clustering
Yinsong Kong, Congwei Liao, Shengxiang Huang, Lei‐Lei Qiu, Lianwen Deng · 2023
The advent of high-tech means of detection posed a huge challenge to traditional camouflage imaging. It is important to develop a more efficient and better performing digital camouflage algorithm to improve the poor camouflage effects. The performance of camouflage generation is mainly affected by the camouflage color and camouflage texture. In this paper, we propose a novel design of digital camouflage based on he K-means clustering optimized by genetic algorithm. First, we randomly call the plaque of the target neighborhood to retain texture details, and then smooth the removal of abrupt boundaries. Then, we extract primary colors from the background and precisely reduce the influence of randomization of the initial cluster center using a clustering method. By comparing with the other reported camouflage patterns, we find that the output camouflage patterns generated by our proposed method greatly match the background and have good camouflage effect.