Artifacts Removal and Edge Detection of Digitally Compressed Images
Shanty Chacko, Lijo T Joseph, J. Earnest Jayakumar · 2012
Block discrete cosine transform compressed image exhibits visually annoying compression artifacts. In this paper a method for reducing the compression artifacts and to increase the visual quality of image is been presented. The compressed image is filtered using Gaussian filter to reduce the amount of artifacts. In order to reproduce the high frequency coefficients a differential image is been obtained. From the differential image the artifact contents are removed and smoothened edge parts are retained. The edge detected image is added with the filtered image to improve the sharpness of the image. Results show that the image thus restored achieves perceivable image quality. Image compression schemes are extremely used now a day's in order to accommodate with the bandwidth of sto- rage medias. To better compress the image, block discrete cosine transform is a method that is widely used. First the image is divided into 8×8 nonoverlapping blocks and each block is transformed independently to convert image into dct coefficients followed by quantization and variable length encoding. Thus binary data streams are generated for data transmission. The BDCT is the recommended transform technique for both still and moving image coding standards, such as JPEG, H.261.H.263 and MPEG. Block discrete cosine transform along with the application of dc quantization give rise to discontinuities between the blocks termed as blocking artifacts. This is due to the fact that the two low frequency dct coefficients in the adjacent blocks similar in value gets quantized to different quantization bins. Removing the high frequency components result in ringing artifacts around the strong edges. Thus the image quality gets decreased and to improve the quality of images image restoration schemes are been used. There are various approaches to suppress the artifacts in transform domain and spatial domain. Zeng (1) models the blocking artifact as 2-D step functions and reduces it by applying zero masking to the dct coefficients of some shifted image blocks. A signal adaptive filtering scheme is used to reduce blocking artifacts in (2) by means of an adaptive weighting mechanism and quantization. Here local region filtering of each pixel can be performed with- out the consideration of the predicted coordinates of block- iness and size of the smoothing window can be varied. ————————————————