本論文是基於LZW壓縮原理,設計應用於小型嵌入式系統的無失真壓縮/解壓縮程式,由LZW方法中延伸出兩個新的變形壓縮方法,一個是利用PNG相鄰兩點色差值壓縮原理,用在LZW相鄰字元差值,並以縮減位元編碼,用在灰階圖形上有不錯的效果,另一個是改善LZW重建字典時,字串取代的順序,在壓縮率上可比原本的LZW演算法更好,而且字典需求空間更小,對於小型靜態無失真圖形檔上,壓縮效果比其他格式好,而且也能用於壓縮非圖形的檔案。另外本論文亦實作整合了RLE的方法於應用程式中,提供了小型嵌入式系統多重的選擇,使用者可自由選擇壓縮方法,及調整本論文所提供壓縮方法的參數,以符合各式不同嵌入式系統記憶體及速度上不同的需求。 This research is based on the LZW compression principle. It extends the principle to two new methods. One of the methods exploits different values with bytes, then compresses as reducing code-bit. Another uses reference values to reserve common use strings when the LZW dictionary need to be rebuilt. For some small images, the new method is better than the PNG image format. And the new method can compress non-imaged files. This research also integrated RLE into the application program that supporting multi-method selection for embedded systems. The user can adjust different parameters for various embedded systems with different demands on speed and memory.