New no-reference blocking artifacts metric based on human visual system | IEEE Conference Publication | IEEE Xplore

New no-reference blocking artifacts metric based on human visual system


Abstract:

Block-based discrete cosine transform (BDCT) is an important tool in the image and video compression. Blocking artifacts caused by BDCT may be noticeable in the decoded i...Show More

Abstract:

Block-based discrete cosine transform (BDCT) is an important tool in the image and video compression. Blocking artifacts caused by BDCT may be noticeable in the decoded image under low bit-rate conditions. Measuring blocking artifacts is of great significance in the evaluation of image and video image. To ensure reliability of the model outputs, a new no-reference blocking artifacts metric is proposed in this paper, based on human vision sensitivity by the research on a model of luminance and texture masking. Furthermore, blocking artifacts are measured with different weighting coefficient in flat regions and edge regions, respectively. Compared with Wang's blind measurement of blocking artifacts, experimental results illustrate that the proposed metric has better consistency to the visual perception, and it can reflect the image quality more effectively.
Date of Conference: 13-15 November 2009
Date Added to IEEE Xplore: 31 December 2009
ISBN Information:
Conference Location: Nanjing, China

I. Introduction

With the widespread development of video communication, many efficient image compression methods have been developed and standardized. Especially, high quality image communication with low bit-rate is widely used for video conferencing, videophone, etc. Block-based discrete cosine transform (BDCT) is the most widely used technique for the compression of both still and moving images [1]. Most of image and video coding standards, such as JPEG and MPEG, are based on BDCT. One of the most noticeable artifacts caused by coding algorism is an artificial discontinuity across the block boundaries called “blocking artifact”, due to the independent quantization of the transform coefficients. Peak signal to noise ratio (PSNR) and mean square error (MSE) are popular image quality evaluation criterions, but they are ineffective measurements for blocking artifacts that are structural disturbance [2]. It is therefore essential to develop metrics that have good consistency to the visual perception.

References

References is not available for this document.