基於JPEG的圖像壓縮與解碼技術
摘 要：壓縮的理論基礎是資訊論。從資訊論的角度來看，壓縮就是去掉資訊中的冗餘，即保留不確定的`資訊，去掉確定的資訊(可推知的)，也就是用1種更接近資訊本質的描述來代替原有冗餘的描述。這樣就可以減輕網絡帶寬日益緊張的問題。圖像壓縮1般透過改變圖像的表示方式來達到，因此壓縮和編碼是分不開的。
在圖象傳送過程中，我們經常採用JPEG格式對靜態圖象進行編碼。JPEG基本系統是1種有損編碼，無法完全恢復出原圖象，資訊有1定的丟失，稱爲有損壓縮。儘管我們希望能夠無損壓縮，但是通常有損壓縮的壓縮比(即原圖象佔的字節數與壓縮後圖象佔的字節數之比，壓縮比越大，說明壓縮效率越高)比無損壓縮的高。JPEG編碼先把圖象色彩RBG變成亮度Y和色度Cr、Cb，它利用人的視覺對色度不敏感的特點，減少1部分色度數據，以達到壓縮。
JPEG採取多種編碼方式，包含有行程編碼(Run Length Coding)和哈夫曼(Huffman)編碼，有很高的壓縮比。在編碼前，先對數據進行分塊，離散餘弦變換（DCT）及量化，保留能量大的低頻信號，丟棄高頻信號以達到壓縮。解碼時，進行熵解碼，反量化，反離散餘弦變換（IDCT）。
本論文介紹了用VC++實現JPEG編解碼技術，該程序的編碼部分能把1張BMP格式的24位真彩色圖象進行JEPG編碼，壓縮成以2進制形式儲存的檔案；透過相應的解碼程序又可以把圖象解壓縮出來。
關鍵字：離散餘弦變換；熵編碼；Huffman編碼；bmp；jpeg
The technology of Graphics compression or decompression that is based on JPEG
Abstract:The theory of compression bases on the theory of information. From the theory of information, compress means that to get rid of the redundant information. In other words, it means that to reserve the uncertain information but get rid of the certain information which can be inferred.,And this can relieve the problem that the network band—width is busier day by day. In general, compress picture can be always done by change the express way of the pictures. Therefore, the compress and encode can not be separated.
While conveying pictures we often encode still pictures into the style of JPEG.JPEG is of lossy compression which can not recover all the image data，some data are losed. Although we refer to non-lossy compression， however， the compression ratio of lossy compression( the times that data bytes before compression to that after compression )is bigger than that of non-lossy compression.JPEG encoding translate image color from RBG to luminance Y and chroma Cr、Cb. Because people’ eyes are less sensitive to chroma comparing with luminance ， some of chroma data are abandoned to reduce data size.
JPEG envolves some kinds of ways of encoding including Run Length Coding and Huffman Coding， so it has high compression ratio. Before Coding ，we divide data into blocks ， DCT and quantify each block . The low-frequency signals that have more energy are maintained and others are rejected . While decoding ， data stream are readed ， decoded ， inverse-quantified and then IDCT .
This paper introduces the encoding and decoding of JPEG and the specific realization of program on VisualC++ encoding part of this program can encode a picture whose format is 24-bits-turecolor BMPg and save relevant data as binary decoding program can decode the compression data and reconstruct the origin image.
Keywords: DCT; Entropy encode; Huffman encode; bmp; jpeg
註釋：不含原始碼