영상압축을 위한 웨이브릿 패킷 알고리듬의 개선에 관한 연구

Abstract

본 논문에서는 코딩성능을 향상시키는 base 선택 알고리듬을 제안하였다. 제안한 알고리듬은 실험을 통하여 정해진 기준값에 의하여 알고리듬의 계산량이 제한되고, top-down tree search를 이용하며 주어진 영상에 따라서 자식 서브밴드에서 상대적으로 에너지 분포가 많은 suboptimal base를 선택한다. 본 연구의 목적은 웨이브릿 패킷 변환의 성능을 알아보는 것이므로 양자화기와 코딩 방법의 최적화는 고려되지 않았고, 제안한 알고리듬의 실험을 위하여 쿼드트리를 이용한 코팅을 적용하여 비트율과 PSNR을 조사하였으며, 비적응적 웨이브릿 변환과 비교하여 상대적으로 적은 계산량의 증가와 1.1㏈의 코딩 효율의 향상을 보였다.

In this paper, the base selection algorithm, which improves the performance of the coding gains, is proposed. The resolved threshold value restrict the computational complexity and the proposed algorithm decompose the child subband with the top-down tree search and the relative energy between the parent and child subband. The object of this paper is the performance of wavelet packet transform and optimization of quantisers or coders is not considered. For the present experiments, we used the quadtree coder scheme and showed the bit-rate, peak signal-to-noise ratio (PSNR) distortion measure, and the reconstructed image. Compared to discrete wavelet transform, PSNR improvements up to 1.1㏈ are achieved for a set of standard test images.

In this paper, the base selection algorithm, which improves the performance of the coding gains, is proposed. The resolved threshold value restrict the computational complexity and the proposed algorithm decompose the child subband with the top-down tree search and the relative energy between the parent and child subband. The object of this paper is the performance of wavelet packet transform and optimization of quantisers or coders is not considered. For the present experiments, we used the quadtree coder scheme and showed the bit-rate, peak signal-to-noise ratio (PSNR) distortion measure, and the reconstructed image. Compared to discrete wavelet transform, PSNR improvements up to 1.1㏈ are achieved for a set of standard test images.