Dynamic Stochastic Resonance Based Blocking Artifacts Removal from Compressed Images in DCT Domain

EasyChair Preprint 961

Abstract

In this paper, we have introduced a dynamic stochastic resonance (DSR)-based blocking artifacts removal of compressed images in DCT domain. Blockiness in DCT based compressed images after reconstruction are a very challenging problem. In the proposed technique it has been observed that blockiness is coming due to a few DCT coefficient of the first row of each block, so we focused only on target coefficient and modified them with a weighted average of neighboring block coefficients. To improve the performance of the blocking artifacts reduction methods, we have used the dynamic stochastic resonance (DSR) by exploiting the clamor or decadence introduced during the quantization procedure. DSR is an iterative process which modulates the quantized factors so that the impact of the quantization procedure is compressed and image information is enhanced. To find a better outcome and minimize the computational complexity, an adaptive optimization method has been taken for selection of bistable parameters. The main advantage of the proposed method is, it overcomes the blocking artifacts introduced during the DCT based compression. The qualitative and quantitative results are also shown.

Keyphrases: Dicrete cosine transform (DCT), Dynamic Stochastic Resonance (DSR), Stochastic Resonance (SR), noise

@booklet{EasyChair:961,
  author    = {Rajib Kumar Jha and Onkar Krishna and Saurabh Kumar Pandey and Sumit Kumar and Vivek Singh Verma},
  title     = {Dynamic Stochastic Resonance Based Blocking Artifacts Removal from Compressed Images in DCT Domain},
  howpublished = {EasyChair Preprint 961},
  year      = {EasyChair, 2019}}