PURPOSE: To improve the picture quality of a reproduced image and to shorten processing time by reducing block distortion or mosquito noises by performing encoding/decoding while dividing an image into blocks corresponding to the contents of that image.
CONSTITUTION: An inputted 10 image signal is stored in a block part 11, block- deformed, quantized 15 and encoded 16 after designating apex coordinates with a pair of horizontal and vertical numerals so that the image can be divided into plural rectangular blocks. On the other hand, at the decoder, the transmitted signal is transmitted from a decoding part 20 to a reverse quantizing part 21 and a shape parameter generating part 25, an inverse DCT 22 is executed to the reversely quantized data, and the rectangular blocks are restored. Then, the data are transmitted to an inverse shape transforming part 23 and rectangular blocks are formed. The rectangular blocks are horizontally scanned, the one-dimensional arrangement is extracted, and the rectangular blocks are restored from the data of correspondent rows. Then, a block synthesizing part 24 sticks the rectangular blocks at positions corresponding to the apex coordinates, and the reproduced image is obtained.
HAYASHI TAKAAKI
TSUKIJI HIROSHI