Design of Military Camouflage Patterns Using Image Processing and Interactive Genetic Algorithm

Military camouflage is important in the protection of personnel, shelters, and equipment from observation by enemy forces. It is challenging and time-consuming to design efficient camouflage patterns manually without extensive experience. The resulting manual patterns may also have low accuracy and effectiveness; thus, a computerized image processing software can be used to facilitate and speed-up the design of such patterns. To achieve this goal, meta-heuristic image processing algorithms can be employed. In this study, the K-Means clustering method and the interactive genetic algorithm were utilized for the first time to produce and optimize military camouflage images scored by expert observers. The input images were reduced to 10 colors by the K-Means clustering method and every pattern was reduced to 3-10 colors with regard to maximum printable colors on the fabric. The introduced algorithm replaced background colors with the ones existed in a pattern according to their redundancy. The camouflage patterns were produced by the interactive genetic algorithm and their suitability was evaluated by 30 military experts. 80% of the expert evaluators were satisfied with the generated camouflage patterns which showed the applicability of the proposed design.