由Said和Pearlman所提出的階層式集合分割影像壓縮技術(SPIHT)，提供了有效率的漸進式和嵌入式的影像壓縮特性。然而，對於某些除了具有大量低頻信號外，也同時擁有高頻信號的影像，SPIHT的編碼壓縮效率都會顯得低落。這是因為這樣類型的影像，並無法透過小波轉換的處理，徹底地將影像中的能量做有效率的集中，以符合SPIHT的編碼特性。本論文提出了利用小波封包以及區塊重排的方法(BRWP-SPIHT)，以達到能量的集中，使編碼效率以及影像視覺品質得以提升。區塊重排技術把影像的小波係數資訊分成許多區塊，再根據每一個區塊的重要性，重新編排這些區塊的位置。根據實驗結果顯示，BRWP-SPIHT在對於具有大量低頻信號外，也同時擁有高頻信號的測試影像，平均而言，客觀的PSNR可以比SPIHT大約提升0.6 dB，而在主觀的視覺品質上，對重建的影像品質也有一定程度的增強，特別是在影像中的條紋和材質部分。 The set partitioning in hierarchical trees (SPIHT) coding algorithm, proposed by Said and Pearlman, provides effective progressive and embedding property. However, for images with high energy that is randomly dispersed throughout high frequency subbands in the wavelet domain, the SPIHT does not fully exploit energy compaction of the wavelet transform and thus becomes less efficient to represent these images. This paper presents an energy compaction method, block reordering wavelet packet SPIHT (BRWP-SPIHT) coding, to enhance the image visual quality. The block reordering technique divides the wavelet coefficients into blocks and reorders these blocks based on the significance of each block. The simulation results show that BRWP-SPIHT is superior, on average, to SPIHT by 0.6 dB for texture rich images. Subjectively, it also shows significant enhancement to the quality of the reconstructed image, particularly for images with fractal and oscillatory patterns.
