小腦模型控制器於串級影像壓縮之研究

A Study of Cascade Compress Image based on Cerebellar Model Articulation Controller

王智仁

指導教授 ： 洪 欽 銘

國立臺灣師範大學/科技學院/工業教育學系/碩士(2002年)

摘要

摘 要
本論文提出一個影像串級壓縮技術，係連結小腦模型控制器(CMAC)與傳統影像處理技術而成，影像串級壓縮技術兼備了二者的優點。相較於傳統影像壓縮技術，影像串級壓縮技術能有較優異的影像壓縮效能，有效率的壓縮影像資料是有助於加速影像資料傳送的速度與精簡影像資料儲存的空間。
再者，本論文提出一個影像修補的原理，這個影像修補原理是藉由小腦模型控制器優越的類化與學習能力來達成。影像修補原理運用簡捷的修正程序，即可以使影像受損的部分獲得不錯的辨識效果。事實上，影像串級壓縮的成效與影像修補的良莠都是完全取決於小腦模型控制器類化的程度與學習的精度。然而，為提高壓縮比率伴隨而來之影像精確度降低的結果是無法避免的，所以小腦模型控制器類化的程度是必須被限定在一定的範圍內。
最後，經由實驗證明，應用小腦模型控制器於影像串級壓縮與影像修補技術上都能獲致良好的效能。

關鍵字

小腦模型控制器 ； 串級影像壓縮 ； 影像修補

並列摘要

Abstract
The thesis proposes a cascade of image compression technique that joins Cerebellar Model Articulation Controller (CMAC) with conventional image processing method in image compression procedure. This cascade of image compression technique combines the merits of two together and has more effective capability of image compression compared to some conventional image compression techniques. Thus it can effectively decreases image data for reducing transmission speed or storage utilization efficiency.
Moreover, the thesis presents a novel method of image retrieval that using the perfect generalization and learning properties of CMAC. This image retrieval method simply gets good recognition for the part of damaging images. In fact, the capability of the image compression and the performance of the image retrieval all are based on the degree of generalization and exactness of learning. However, there is a trade off between advantages and accuracy of image. Thus the demand of the degree of generalization must be limited.
Finally, from experiential results that apply CMAC to cascade image compression and image retrieval are able to achieve advantages and good performance.

並列關鍵字

CMAC ； cascade image compression ； image retrieval

參考文獻

[2] R. J. Clarke, " Digital Compression of still Image and Video", Academic Press, 1995.

[3] W. H. Fang, N. C. Hu and S. K. Shih, " Recursive fast computation of the two-dimensional discrete cosine transform," IEEE Proceedings on Vision, Image and Signal Processing, Vol. 146, No. 1, Feb. 1999, pp.25-33.

[4] De Natale, F.G.B., Perra, C., Vernazza, G.,"DCT information recovery of erroneous image blocks by a neural predictor", Selected Areas in Communications, IEEE Journal , Vol. 18, June. 2000, pp. 1111 –1121.

[6] W. B. Pennebaker and J. L. Mitchell, "JPEG: Still Image Data Compression Standard, " New York: Van Nostran Reinhold, 1993.

[7] B. L. Yeo and Liu, "A unified approach to temporal segmentation of motion JEPG and MPEG compressed video, " in Proc. of Int. conf. On Multimedia Computing and Systems, May 1995, pp.81-88.