Model-based discontinuity evaluation in the DCT domain

Abstract

The discrete cosine transform (DCT) has been used for compressing videos or images with standards like MPEG and JPEG. In this paper, we derive DCT properties related to a standard discontinuity and propose a model-based discontinuity evaluation technique in the DCT domain. This technique consists of a direction verification and a position alignment method with an evaluation criterion. The direction verification and position alignment causes the DCT coefficients to be of the centralized form, which enables an evaluation regardless of various positions and directions of discontinuities. The evaluation criterion examines the standard position and evaluates the magnitude of a discontinuity by using the properties of the ideal step model. Although the detected discontinuities are rough in a low-resolution image for the size (8 x 8 pixels) of DCT blocks, experimental results show that this technique is fast in processing and robust against noise. (C) 2001 Elsevier Science B.V. All rights reserved.