Please use this identifier to cite or link to this item: http://hdl.handle.net/10266/2930
Title: Blocking Artifacts Reduction in JPEG Compressed Images Using Overlapped Block Processing
Authors: Simranjit, Singh
Supervisor: Kulbir, Singh
Keywords: Blocking artifacts;Overlapped block processing
Issue Date: 14-Aug-2014
Abstract: In recent years, there has been a huge interest in compression of digital image data due to the fact that a large number of communication applications contain pictures or video. Thus, image compression is considered very useful in multimedia applications and in distributed information systems that operate in network environments. One of the most commonly used image format, JPEG, uses the concept of transform coding or more specifically Block based Discrete Cosine Transform (BDCT). Despite their widespread popularity BDCT compression scheme leads to noticeable blocking artifacts at high compression levels due to discontinuities between sub-images. Prior research shows that incorporating systems that reduce blocking artifacts into codecs is useful because visual quality is improved. Existing methods reduce blocking artifacts by applying various pre and post processing techniques to the image. This dissertation examines a framework where blocking artifacts are reduced by means of overlapped block processing. In the proposed method, the blocking artifacts problem is dealt with at the source end, namely, the segmentation process. In a typical block processing, an image is divided into mutually exclusive regions. Instead of constraining the regions to be exclusive of each other, it is rational that a slim overlap around the boundary of each region could reduce the blocking artifacts. Objective quality of post processing images is measured using human visual system based peak signal to noise ratio (PSNR-HVS-M) instead of PSNR due to ineptness of the latter for real analysis purpose. An increment of 1.917 dB of PSNR-HVS-M at 90% compression level is observed for 256×256 Lena image by applying this algorithm using DCT with 7×7 inside block. However, it also increases the number of real arithmetic computations by 26.5%. An attempt is also made to apply the algorithm by means of Discrete Fractional Cosine Transform (DFrCT) as a transform coding technique which is generalised form of DCT (Type-I) and considered to play a vital role in the area of image processing in future because of its fractional order acting as an extra degree of freedom. However, DCT (Type-II) comes out to be better than DFrCT in case of image compression. The subjective and objective results for various images indicate that the proposed method performs better in terms of both visual quality and PSNR-HVS-M for all ranges of compression level.
Description: ME, ECED
URI: http://hdl.handle.net/10266/2930
Appears in Collections:Masters Theses@ECED

Files in This Item:
File Description SizeFormat 
2930.pdf3.71 MBAdobe PDFThumbnail
View/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.