| 摘要 |
PURPOSE: An image data segmentation method and a device thereof are provided so that an interesting section of the image data can have relatively good image quality, by extracting time and space properties of the image data, segmenting an object and background thereof, and applying different quantization variables to the object and background. CONSTITUTION: In a time property extraction process, motion vectors are obtained by motion-estimating an image data in block units, and a dynamic block having a motion vector which is not zero is obtained(22). In a space property extraction process, the dynamic block and the image data are discrete cosine transformed in block units, and a space property including DCT coefficients is obtained in the respective blocks(24). In a segmentation process, a moving object and background of the image data are segmented by using the space property including the motion vectors and the DCT coefficients(26). Thereafter, in a quantization process, different quantization variables are applied to the object and background of the image data segmented in the segmentation process(28). Accordingly, the object and background of the image data are segmented by using the time and space properties of the image data, and are provided with different quantization variables. |
