Detection of blocking artifact on satellite image and its new evaluator
Risnandar Risnandar, Masayoshi Aritsugi · 2016
New algorithms for blocking artifact detection on a compressed satellite image by using a 7×7 block and for analyzing its characteristics are proposed. Blocking artifacts often appear on an image when it has a very high compression rate based on either the loss-compression or lossless compression. Localization of blocking artifact is useful for finding the image's characteristics. This study also proposes methods of image quality metric that will analyze the characteristic of distorted satellite image caused by blocking artifact based on the rate-distortion and the similarity. In addition, we propose a new evaluator, which combines precision and recall, as a more suitable measure than MSE for error measurement in imbalanced data. Experimental evaluations show that our proposal works well for satellite images.