Camouflage is the art of disguising or blending objects into a natural background so as to make them more difficult for viewers to see. Traditional camouflage is usually based on the designer's experience and includes macro patterns spots and stripes irregular whose outlines or boundaries are sharp and are easier to see. To overcome this main drawback, digital camouflage combines macro and micro patterns with computer assistance. Most works related to the digital camouflage are in the field of color pattern design. However, designing a suitable color pattern that can best match the target in terms of shape and color characteristics with the background is a major challenge in the field of digital camouflage. Nowadays, the digital camouflage is based on the principles of visual psychology and uses digital image processing techniques to characterize background features. The common digital camouflage techniques are based on the fuzzy, the neural network and the greedy methods. The main problem to use these methods is that the number of main colors is chosen manually or experimentally, while it is different in each image. Therefore, the optimal colors cannot be obtained for appropriate blending targets into their backgrounds.  
The main objective of this study is to provide a novel method of designing a digital template which automatically extracts the number of original colors based on the specific features of each image.  The proposed method is based on the conventional greedy algorithm. The greedy algorithm tries to minimize the difference between the shape perceived by the viewer and the shape patterned on the target. The proposed method first uses the minimum description length (MDL) criterion for determining the number of optimum clusters of the image. Then, it uses the well-known K-means clustering method to extract the original colors from the image. Finally, the proposed method uses the greedy algorithm to obtain an optimal distribution or arrangement of the combination of pattern templates stored in a database.
In this study, the proposed method is compared to the color similarity algorithm proposed by yang and Yin (2015). The quantitative and qualitative assessments of both the methods are based on the saliency map, which is a common criterion for the camouflage assessment. The saliency map is originally intended to model covert attention. It attaches a value to each location in the visual field given the visual input and the current task, with regions of higher salience being more likely to be fixated.
For our comparison, 11 different images captured in different conditions have been used in this study. The images used are in different times (spring, summer, autumn, and winter seasons) and different location (desert, forest, sea, urban, etc.) conditions. Experimental results show that, the mean value of the saliency measure in the 11 images are, respectively, 53% and 42% for the color similarity algorithm method and the proposed method. This indicates that the proposed method is superior to the color similarity algorithm for distinguishing the targets in their backgrounds.