Edge statistics-based image scale ratio and noise strength estimation in DCT-coded images

This paper presents an algorithm for estimating the scale ratio of a compressed image and the strength of compression noise that is associated with a block-based compression scheme. The proposed basic idea was based on analyzing the block edge statistics resizing across two neighboring blocks. For this purpose, the block boundary discontinuity was analyzed statistically. The scale ratio of an input image was then estimated. The location of the discontinuity was detected using the computed scale ratio. Finally, the strength of compression noise was estimated by combining the computed statistics. The main advantages of the proposed algorithm are (1) it estimates the image scale ratio and block offset associated with image compression without prior information, (2) statistics calculated during the offset detection can be used to estimate the compression noise strength, and (3) the estimated noise strength can guide available deblocking algorithms to adjust their filtering strength. Several experiments were performed under different scale ratios and compression strengths to evaluate the proposed scheme.