Estimation of Edge Direction for Block Error Concealment using Hypothesis Testing Problem

S.H. Hyun, S.S. Kim, I.K. Eom, and Y.S. Kim (Korea)

Keywords

Block Recovery, Edge Direction, Hypothesis Testing, Wavelet Coefficient.

Abstract

In this paper, we present an error concealment technique that estimates the edge directions of lost block. We propose an adaptive block pair selection algorithm to estimate the edge direction of lost block in wavelet domain. For this estimation to be made, a hypothesis testing problem is applied to the variance of wavelet coefficients. Here, we use an F-distribution. The strong edge has large amounts of variance because of its large coefficients. Therefore, our method decides the edge directions of the lost block from hypothesis-testing problem using the variance of the wavelet coefficients of the neighbouring blocks. The proposed method outperforms the previous methods that only used a predefined set of neighbouring blocks.

Important Links:



Go Back