Determination of the cause and amount of image degradation using a reduced reference approach

摘要

Multimedia data transmission, particularly sending images, is an important feature of modern communication systems. These activities usually occur on a very large scale and hence, reduction of the data volume and storage requirements become crucial. Reduction is achieved through compression, typically JPEG2000. Since this is a lossy compression technique, the image quality is degraded, the extent depending on the amount of compression. The transmission channel is another source of quality loss of images. At the receiving end, the absence of a clean reference image makes the task of quality assessment difficult. Since image quality assessment is required for different applications, various reduced and no reference approaches have evolved to meet the demand. The job of restoring the image quality is facilitated by the knowledge of the exact source of degradation. There has been however, little work in this area. This article proposes a reduced reference approach, which, starting with a received degraded image, is able to distinguish between two different types of degradation - Gaussian noise (commonly arising from transmission channels) and compression due to JPEG2000. The approach can, not only correctly determine the cause of the degradation but also the amount of it, with only minimal information regarding the original image. This is the most significant contribution of this article. The performance of the proposed approach is established through extensive simulations using images from well-known databases.