以SPIHT為基礎實現膠囊內視鏡影像無失真壓縮

SPIHT- Based Lossless Compression for Capsule Endoscope Images

姚皓勻(Hao-Yun Yao)

指導教授 ： 繆紹綱

中原大學/工學院/電子工程研究所/碩士(2005年)

https://doi.org/10.6840/cycu200500453

摘要

中文摘要

在無失真靜態影像壓縮技術中，針對壓縮率的提升一直是一道困難度頗高的課題。我們針對醫學上的膠囊內視鏡影像，發現其獨特的影像特性，其中佔了27%的黑色外框，經醫師証實，對診斷來說是沒有意義的，只有內圈的圓形部分才有診斷價值。所以，我們針對這項特點，以階層樹狀結構分組SPIHT (Set Partitioning In Hierarchical Trees)這個已被証明對於影像壓縮相當有成效的演算法為基礎而進行提升壓縮效能的研究。使用此方法來做壓縮有以下兩個優點：(1) 不需要另外再發展一個新演算法來做壓縮，節省開發成本。(2) 此演算法廣被驗証，既簡單、快速又可靠。
在本篇論文中我們提出一種利用多個區塊的影像壓縮方法對256 256大小的膠囊內視鏡影像壓縮。首先將一張影像經由切割的程序分為多個區塊，接著按各區塊的大小進行SPIHT的壓縮，而提高壓縮率的關鍵，在於切割影像時，將黑色外框完全切除，只對中間有用的資訊給予編碼。在解碼端再將黑色外框與中間圓形內視鏡影像進行重建，以回復完全無失真之影像。本論文所提出的方法，其壓縮效能比起整張影像直接做SPIHT可提高約27%，比起大家公認之最佳無失真影像壓縮法之一的JPEG-LS，絲毫不遜色。在無失真壓縮的前提下，如此的壓縮率更顯得可貴。

關鍵字

小波係數 ； 膠囊內視鏡影像 ； 影像無失真壓縮 ； SPIHT

並列摘要

Abstract

The enhancement of the compression ratio is a challenging topic in lossless static image compression. We find out a characteristic of capsule endoscope image, which includes a useless outer boder area for doctors’s diagnoses. This useless area is in black and occupies 27% of entire image, while the rest circular area is valuable for diagnoses. We make use of this feature and SPIHT (Set Partitioning In Hierarchical Trees), a state-of-the-art image compression method, to enhance the compression ratio. There are two advantages of using this existing technology for image compression. (1) No new algorithm is required to be developed for achieving the goal of compression. (2) The method is based on the algorithm that is widely verified, easy to use, and comprehensive.
In this thesis, we propose a multiple block image compression method for 256 256 capsule endoscope images. We segment the source image into many blocks first, and proceed to perform SPIHT coding for the blocks having different sizes. The key point to enhance the compression ratio is that we cut the black border of image off and encode only the remaining useful portion of image. At the decoder side, the image is reconstructed with black border and the circular image to recover the lossless image. The method proposed in this thesis enhances the compression ratio by about 27% in comparison with using SPIHT for one complete image. In the field of lossless compression, this result is valuable and commendable and can compare with the JPEG-LS, one of the leading lossless image compression standands.

並列關鍵字

capsule endoscope image ； SPIHT ； lossless image compression ； wavelet coefficient

參考文獻

[1] H. Haneish and Y. Miyake, “Distortion compensation of electronic endoscope image,” IEEE Conf. Rec. Med. Image Conf., vol. 3, pp. 1717-1721, 1994.

[2] H. Hideaki, Y. Yagihashi and Y. Miyake, “A new method for distortion correction of electronic endoscope image,” IEEE Trans. Med. Imaging, vol. 14, pp. 548-555, Sept.1995.

[3] M. Li and L. Jean-Marc, “Some aspects of zoom lens camera calibration,” IEEE Trans. Pattern Anal. Machine Intell., vol. 18, pp. 1105-1110, Nov. 1996.

[5] K. Inamura, H. Kondoh, and H. Takeda, “Development and operation of PACS,” IEEE Communications Magazine, vol. 34, pp. 46-51, July 1996.

[6] R. Noumeir, “DICOM structured report document type definition,” IEEE Trans. Inform. Technol. Biomed., vol. 7, no. 4, pp. 318-328, Dec. 2003.