Content adaptive image de-blocking

Abstract

Blocking artifacts may appear in block based digital image and video coding systems like JPEG and MPEG, especially when compression is done at a low bit-rate. The blocking artifacts can degrade the perceived image quality. In this paper, a method for reducing the blocking artifacts in the decoded image using adaptive filters is proposed. The selection of the proper filter is based on the local image pattern and/or the relative pixel position in the coding block. The results show a significant image quality improvement. To evaluate the methods, both Mean Square Error (MSE) scores and Block-edge Impairment Metric (BIM) are given for objective comparison. Screen shots are also shown to demonstrate the subjective quality.