Image segmentation by using K-means clustering algorithm in Euclidean and Mahalanobis distance calculation in camouflage images

Abstract

In camouflage images, the texture of an object is hidden in the background image texture. The hidden object has almost the same color tone and texture as the background image. Since camouflage images show close features with background texture, it is quite difficult to segment and detect the camouflaged object from the image background. In this study, image segmentation is performed on camouflage images using K-means method using Euclidean and Mahalanobis distance calculations. The average value of RMSE 262.47 and the average value of PSNR 24.26 was obtained when using Euclidean distance calculation. Also, the average value of RMSE 799.62 and the average value of PSNR 19,66 was obtained when using Mahalanobis distance calculation. According to the result obtained from this study, while the low RMSE values were obtained with the K-means method by using the Euclidean distance calculation, the lower PSNR values were obtained by using the Mahalanobis distance calculation. In the experimental results; K-means method with Euclidean distance calculation is more successful than the K-means method with Mahalanobis distance calculation.