Encoder Optimization using Collaborative Filtering with Dictionary Learning to Improve PSNR

In this paper, we propose inculcation of collaborative and adaptive filtering for a H.264 /AVC deblocking filter. For this, an improved PSNR–B model including optimal adaptive filter with weight adjustment is designed for improved PSNR calculation. Block matching 3D algorithm is used for the image decomposition and form a dictionary of blocking and deblocking components for post processing. The dictionary is capable of removing the blocking artifacts effectively and avoids unwanted blurring and maintains edges found in the original image. Proposed method enhances the filtration and reduces deblocking time. The proposed technique is able to reduce about 46% deblocking time with improvement in PSNR.