誤差訊號編碼已經被廣泛的運用在影像壓縮上，尤其是近無失真影像壓縮
的領域上。為了準確的預測，這類演算法需要將上一行已經編碼過的像素
值存放在記憶體中。然而，隨著影像的解析度越來越大，影像的尺寸也都
隨之增加以至於所需要的記憶體也跟著增加。更何況，這些用來預測的資
訊通常是以無失真的方式儲存，以資料壓縮的觀點來說，這是一個很沒有
效率的作法。本論文中，提出一個誤差訊號編碼的演算法，此演算法可以
在只降低一點效能的情況下減少記憶體的需求。本論文的演算法有兩個主
要的部分，一個是改進在JPEG-LS中斜向邊緣的預測方式；另一個部分則是
利用小波轉換以及一個新提出的位元控制方法，以減少記憶體的使用。實
驗的結果將會說明此論文提出的演算法可以節省記憶體的需求，並且壓縮
比幾乎沒有改變。

DPCM is widely used in image compression, especially in lossless and near
lossless image compression. For accurate prediction, it need memory bu er
to store pixels encoded in the lastes lines. However, as the resolution of im-
ages become larger, the image size increses, more memory is needed to store
context information. Furthermore, this context information is usually stored
losslessly which is not a e cient method in the view of compression. In this
thesis, a DPCM algorithm is proposed to reduce the memory requirement
without loss of performance greatly. There are two major parts in the pro-
posed algorithm, the rst one is the improvement of JPEG-LS diagonal edge
prediction and the other one is that reduce the memory requirement based
on DWT and a proposed rate control method. Experimental results show
that the proposed algorithm low the usage of memory and the comopression
ratio is almost the same.

[1] Rafael C. Gonzalez and Richard E. Woods Woods. Digital Image Pro-
cessing. Robbins, Tom, second edition edition, 2002.
[2] Rafael C. Gonzalez and Richard E. Woods Woods. Digital Image Pro-
cessing. Robbins, Tom, second edition edition, 2002.
[3] C. Grecos, jiang J, and E.A. Edirisinghe. Two low cost algorithms
for improved diagonal edge detedtion in jpeg-ls. IEEE Transaction on
Consumer Electronics, 4:579{586, August 2001.
[4] Jaehan In, Shahram, and Faouzi Kossentini. On rd optimized progres-
sive image coding using jpeg. IEEE Transaction on Image Processing,
9:1630{1638, November 1999.
[5] Jianmin Jiang. A low-cost content-adaptive and rate-controllable near-
lossless image coded in dpcm domain. IEEE Transaction on Image
Processing, 12:543{554, April 2000.
[6] Taekon Kim, Hyun Muna, Ping-Sing Tsai, and Tinku Acharya. Memory
e cient progressive rate-distortion algorithm for jpeg2000. IEEE Trans-
action on circuit and system for vedio technology, 7:181{187, January
2005.
[7] Jong-Sen Lee. The loco-i lossless image compression algorithm: Princi-
ples and standization into jpeg-ls. IEEE TRANSACTION ON IMAGE
PROCESSING, 16:1309{1324, Auguest 2000.
[8] Viresh Ratnaker and Miron Livny. An e cient algorithm for optimizing
dct quantization. IEEE Transaction on Image Processing, 4:267{270,
February 2000.
[9] Amir Said and William A. peaarlman. A new, fast, and e cient image
codec based on set partitioning in hierarchical trees. IEEE Transaction
on circuit and system for vedio technology, 8:243{250, June 1996.
[10] Xiaolin Wu and Nasir Memon. Context-based, adaptive, lossless image
coding. IEEE Transaction on Communication, 8:437{444, April 1997.
[11] Zongze Wu and Nanning Zheng. E cient rate-control system with three
stages for jpeg2000 image coding. IEEE Transaction on circuit and
system for vedio technology, 17:620{636, July 2006.
[12] Y.M. Yeung and Oscar C. Au. E cient rate control for jpeg2000 image
coding. IEEE Transaction on circuit and system for vedio technology,
10:335{344, March 2005.
[13] Wei Yu, Fangting Sun, and Jason E. Fritts. E cient rate control for
jpeg2000. IEEE Transaction on circuit and system for vedio technology,
13:577{589, May 2006.