The Block Transform Coded, JPEG- a lossy image compression format has beenused to keep storage and bandwidth requirements of digital image at practicallevels. However, JPEG compression schemes may exhibit unwanted image artifactsto appear - such as the 'blocky' artifact found in smooth/monotone areas of animage, caused by the coarse quantization of DCT coefficients. A number of imagefiltering approaches have been analyzed in literature incorporatingvalue-averaging filters in order to smooth out the discontinuities that appearacross DCT block boundaries. Although some of these approaches are able todecrease the severity of these unwanted artifacts to some extent, otherapproaches have certain limitations that cause excessive blurring tohigh-contrast edges in the image. The image deblocking algorithm presented inthis paper aims to filter the blocked boundaries. This is accomplished byemploying smoothening, detection of blocked edges and then filtering thedifference between the pixels containing the blocked edge. The deblockingalgorithm presented has been successful in reducing blocky artifacts in animage and therefore increases the subjective as well as objective quality ofthe reconstructed image.
展开▼
