Improving image quality for block coding by image division method
Zhong Zheng, Masayuki Nakajima, Takeshi Agui · Systems and Computers in Japan · 1992
Abstract In block coding, the original image is divided into square blocks composed of N × N pixels, and the coding is applied to each block. When the image is coded at a low bit‐rate, however, there is often produced a block noise in the reconstructed image. When a block noise is produced in the part of the time where the change of the gray‐level is slow, the noise deteriorates the image quality greatly. To reduce the block noise, this paper proposes three methods of image division. Let the blocks in a horizontal row in the conventional divided image be called a block row. In the first method, the division is executed by moving each block row in the horizontal direction. The amount of translation is determined by a random variable. The second method divides the original image by the rectangular block, and then each block row is moved as in the first method. The third method divides the image using an octagonal block. It is verified by experiment that the block noise can be reduced by the proposed image division.