Affiliations: [a]
IK Gujral Punjab Technical University, Jalandhar, Punjab, India
| [b]
Faculty Department of Electronics and Communication Engineering, DAVIET, Jalandhar, Punjab, India
| [c]
Faculty Department of Applied Sciences, MIMIT, Malout, Punjab, India
Correspondence:
[*]
Corresponding author. Amanpreet Kaur, Ph.D, Research Scholar, IK Gujral Punjab Technical University, Jalandhar 144601, Punjab, India. Tel.: +919872257573; E-mail:[email protected].
Abstract: With the higher compression ratio, the decoded image produces annoying compression artifacts near block boundaries, ringing artifacts near original edges and corner outliers at block corner. These coding artifacts are caused by quantization and transformation process of discrete cosine transform (DCT). This paper proposes a novel deblocking algorithm that removes block discontinuities by taking into account the ringing, blurring and corner outlier artifacts. The proposed deblocking technique consists of two frequency related modes (smooth and detailed region mode) and corner outlier mode have been proposed and then applied median filter. The proposed technique has been applied to a number of reconstructed images and their performance is compared with conventional methods on the basis of standard metrics such a PSNR-B, BBM and MOS. Experimental simulation results illustrate that proposed technique improves the perceptual quality of reconstructed images and thus outperforms the all existing methods.
