Application of feedforward neural network for the deblocking of low bit rate coded images

  • Authors:

  • Kee-Koo Kwon;Man-Seok Yang;Jin-Suk Ma;Sung-Ho Im;Dong-Sun Lim

  • Affiliations:

  • Embedded S/W Technology Center, ETRI, Daejeon, Korea;Embedded S/W Technology Center, ETRI, Daejeon, Korea;Embedded S/W Technology Center, ETRI, Daejeon, Korea;Embedded S/W Technology Center, ETRI, Daejeon, Korea;Embedded S/W Technology Center, ETRI, Daejeon, Korea

  • Venue:
  • AIS'04 Proceedings of the 13th international conference on AI, Simulation, and Planning in High Autonomy Systems
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, we propose a novel post-filtering algorithm to reduce the blocking artifacts in block-based coded images using block classification and feedforward neural network. This algorithm exploited the nonlinearity property of the neural network learning algorithm to reduce the blocking artifacts more accurately. At first, each block is classified into four classes; smooth, horizontal edge, vertical edge, and complex blocks, based on the characteristic of their discrete cosine transform (DCT) coefficients. Thereafter, according to the class information of the neighborhood block, adaptive feedforward neural network is then applied to the horizontal and vertical block boundaries. That is, for each class a different multi-layer perceptron (MLP) is used to remove the blocking artifacts. Experimental results show that the proposed algorithm produced better results than those of the conventional algorithms both subjective and objective viewpoints.