Most existing high performance image coders in applications are transform based coders because they provide better performance for a fixed complexity as compared to the spatial methods. The two popular transforms for image compression are the Discrete Cosine Transform (DCT) and Subband Coding (SBC) such as the Discrete Wavelet Transform (DWT) based systems. The DCT yields nearly optimal energy compaction. However, since the cosine basis functions are non-local, it is not practical to apply the DCT to an entire image because of spatial ringing artifact when high frequency coefficients are quantized. To localize the cosine bases, the DCT is usually implemented as block-based transform by which the image is normally divided into a number of non-overlapping blocks (e.g. 8 × 8 pixels) and the 2-Diomentional DCT (2-D DCT) is applied to these blocks individually. In this way, the high frequency ringing artifacts are reduced. But unfortunately, at low bit rates, block-based DCT suffers from the blocking artifacts, i.e., discontinuities at the block boundaries as shown in Figure (1) for “Lena” test image. In addition, block-based DCT is unable to exploit the correlations between the DCT coefficients across block boundaries. Nevertheless, Block-based DCT is the basis of many international multimedia compression standards such as Joint Photographic Expert Group (JPEG) for still images compression and the Motion PEG (MPEG) for video compression. Low computational complexity and reasonable compression performance are the main reasons for the popularity of the block-based DCT systems [Salomon 04], [Shi 08].
علي كاظم الجنابي | Ali Kadhim Al-Janabi | 861