查詢結果分析
相關文獻
- 應用小波係數編碼之四分樹結構
- 以DSP晶片為架構之影像處理電路設計及其應用
- Analysis of DWT Coefficients and Its Application
- A Fast Convolution Algorithm for Biorthogonal Wavelet Image Compression
- The Global Minimum of Scalar Quantization Errors by Discrete Wavelet Transforms in Image Compression
- 進階雙預測器DPCM影像傳輸系統
- 加強式中心映射法於相似醫學影像集壓縮之研究
- 應用三維小波轉換於高光譜影像壓縮之研究
- 小波轉換影像壓縮模式之研究
- 結合小波轉換與多層級向量量化作醫學影像壓縮
頁籤選單縮合
題名 | 應用小波係數編碼之四分樹結構=A New Quad-Tree Structure for Encoding of Wavelet Coefficients |
---|---|
作者 | 沈岱範; 康世宗; | 書刊名 | 科技學刊 |
卷期 | 8:4 1999.10[民88.10] |
頁次 | 頁281-289 |
分類號 | 312.9 |
關鍵詞 | 零樹編碼; 小波轉換; 影像壓縮; 四分樹; JND; Zero-tree coding; Wavelet transform; Image compression; Quad-tree; |
語文 | 中文(Chinese) |
中文摘要 | 本篇論文提出一種新的zero-tree編碼改良技術,稱之為JZW_1,當影像經過小波 編碼及次頻係數經過量化後,各次頻的係數分佈呈現不同空間走向,因此,JZW_1依據此 係數分佈的特性與重要性在zero-tree structure上有突破,而這樣編碼方式的改進可以提昇編 碼效率。在1998年Shen and Yan以在陰極射線管顯示器上量測之人類視覺特性JND(Just Noticeable Difference)為基礎提出JZW(JND based Zero-tree Wavelet)其在影像視覺品質上 優於SPIHT。而JZW_1是以JZW為基礎繼續在編碼方法上改進,JZW_1的主要編碼優點 在於,它對於小波轉換之高頻係數的編碼處理,提出一套更有效率的方式,使得影像在高品 質高位元率下有更優越的壓縮率。經電腦模擬,我們所提的新方法,可以使整體的bit rate 比原先的JZW編碼方式平均降低3∼17%,且影像品質越(高頻係數越多),位元率越高則 此方法之優勢更明顯,本文方法雖以JZW為基礎做改進,但其概念可以推廣到其它wavelet zero-tree based 編碼。 |
英文摘要 | The proposed JZW_1 coding technique is an improved version of JZW. Based on the measurements of JND (Just Noticeable Difference) property on CRT monitor as well as the zeor-tree encoding method, Shen and Yan proposed JZW (JND based Zero-tree Wavelet), which is proven to achieves better image quality than SHIHT perceptually as well as in PSNR. When an image is wavelet transformed and quantized, the coefficients in different spatial directional subbands have their own characteristics. Based on these characteristics, JZW_1 adopts three methods to improve the coding efficiency: 1. A novel quad-tree structure 2. Improved Zero-tree coding and 3.Removal of isolated coefficients. Compared to JZW and SPIHT, JZW_1 provides higher efficiency for encoding coefficients in AC subbands. For common seen images, JZW_1 reduces bit rates up to 3~17% over JZW at the same PSNR. Although the method is based on JZW, the concept can extend easily to other wavelet zero-tree based coding. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。