Lossless Image Compression Method Based on Block Grouping

[[abstract]]本論文提出一無失真灰階影像壓縮技術。此方法先以中間邊緣偵測法(Median Edge Detector, MED)產生一差值影像(Difference image)與一符號影像(Sign image)，為了降低符號影像之複雜度，本論文採用位元轉移法(bits transformation)將符號影像與差值影像之最重要位元(MSB)進行交換，使此符號影像之複雜度能得以降低。接著以區塊歸類法(Blcok classification)將影像區塊歸類至資料容器(Data bin)，每一個資料容器皆包含不同的資料樣本。最後，將各個資料容器的資料用算術編碼法(Arithmetic coding)壓縮。 實驗結果顯示本論文所提出之方法在影像複雜度較低的情形下，所得結果比JPEG 2000之無失真壓縮法好。同時此實驗結果也比其它無失真壓縮法如GIF、PNG、TIFF來得好。[[abstract]]This thesis proposes a lossless image compression technique for gray level images. This method employs a median edge detector (MED) on the input image to produce a difference image and a sign image. Then, in order to reduce the complexity of the sign image, bits transformation is adopted to transform the bits between the sign image and the most significant bits (MSB) in difference image. The resultant image blocks, according to the maximum value of image block, are classified into data bins, each of which has different data patterns. Finally, the arithmetic coding method is applied to compress the data in each data bin respectively. The result of this thesis shows that the proposed method when applied in smooth images provides a better bit rate than JPEG 2000 lossless method does. And the result is also better than other lossless compression methods such as GIF, PNG, and TIFF.[[note]]碩