A Comparative Study on DeBlocking Algorithms for Highly Compressed Image

• Main Authors: Hsueh-Jung Hung, 洪雪榕
• Other Authors: Chin-Hsing Chen
• Format: Others
• Language: en_US
• Published: 2008
• Online Access: http://ndltd.ncl.edu.tw/handle/95586097847859438681

碩士 === 國立成功大學 === 電腦與通信工程研究所 === 96 === In recent years, the demand of images has been increasing because the Internet and multimedia prevails all over around. In order to transmit and store data, there are many compression standards. JPEG image compression standard is very popular nowadays which has two kinds of compression modes. One is lossless, the other is lossy. In order to get high compression rate, we often use the lossy mode to compress images and multimedia. However, the higher the compression rate, the more serious the blocks appear blocky. Degradation caused by this mosaic like artifact is called the blocking effect. We will focus on improving images with blocking effect after JPEG compression in this thesis. There are two classes of deblocking methods: one is removing blocking effect as noises by filtering the compressed image, the other is removing blocking effect in the wavelet transform domain. The Kuo’s method only deals with visible blocks by filtering and the Ramamurthi’s method filters pixels of different classes by one-dimensional and two-dimensional filters. The Shi’s method weakens edge features in the wavelet domain. We design a deblocking algorithm by combining the merits of the above three algorithms. At first we remove the block boundary by the one level Shi’s wavelet-based deblocking algorithm. Then we blur lightly the image by a lowpass filter in order to blur the block boundary. Then we filter the pixel at visible block boundaries but not inside blocks. Finally we filter pixels in the remaining blocks by the Ramamurthi’s method. By combining the merits of the three well known algorithms, we design an algorithm which eliminates the blocking edge and meanwhile maintains the edge information of images. In the future, we will focus on further improving the sharpness of the image.