Abstract: An apparatus for decoding coded video data is provided which is capable of decreasing the storage size of its frame memory while attenuating the drift noise and minimizing lowering of the resolution during the reconstruction of the coded video data. The apparatus comprises a data size reducer for reducing the size of a decoded video data reconstructed by subsampling, pixel differential coding, or other methods, a frame memory for saving a reduced video data released from the data size reducer, and a data size recover for recovering the original size of the decoded video data from the reduced data saved in the frame memory. Accordingly, the storage size of the frame memory is decreased by a combination of compression of pixel data and IDCT processing in a reduced block size and will successfully attenuate unwanted drift noise and minimize declination of the resolution.

Abstract: Coding bit rate converting method and apparatus for coded audio data are disclosed to convert input coded audio data to output coded data of a coding bit rate lower than a target coding bit rate. A control output is taken out when the frame size of the input coded audio data is larger than the frame size determined by the target coding bit rate, and the output coded data is provided as a converted output by controlling parameters defining the frame size of the input coded audio data by the use of the control output in a predetermined procedure until the frame size of the input coded audio data becomes smaller than the frame size determined by the target coding bit rate.

Abstract: A system for detecting a moving object in a moving picture, which can detect a moving object by block directly with a highly-accurate, simple process without completely decoding the compressed moving picture data. The picture block coding mode, the motion compensation mode and the motion vector information (information b) decoded in a variable-length decoder 1 and the texture information d detected by a texture information detector 7 are applied to a moving object detection processor 8. The moving object detection processor 8 judges whether the particular block is a moving object or not using the information b or the combined information b and the texture information d. This judgement is effected by a motion vector, a spatial similarity judgement and a temporal similarity judgement. The moving object detection information e resulting from the judgement is sent to a picture image processor 5 for processing the picture information f.

Abstract: A video query and editing system which is capable of querying images without fast-forward and playback operations, to specify an editing start or end point easily, and to lower editing cost of compressed data. An edit control window is displayed (11) first, and then a file for query and editing is selected in query and editing file processing 12. In motion picture query processing 13, image query is performed in macro or micro units for the selected file and then the images to be under the query and editing are displayed. In editing section specification 14, a start point and an end point of a section for the editing are specified out of the images displayed in the motion picture query processing. Editing of a plurality of sections or other files are performed in the same manner as for the above processing if desired. After all the editing sections are determined, the editing sections are played back in editing section playback 17 to confirm the editing parts.

Abstract: To provide a fast video browsing system in which a fast playback suitable for keeping track of the contents can be performed by adaptively determining the number of skipped pictures according to the motion vector of picture, and displaying the picture after the skipped pictures. In step S1, pictures to be fast viewed are sequentially inputted, and in step S2, n pictures are skipped. In step S3, the picture after the n skipped pictures is displayed. In step S4, using the picture before or after the displayed picture with respect to time, a motion vector at one frame interval is detected, and in step S5, the number n of skipped pictures is determined from the motion vector. As a result, the number of skipped pictures can be adaptively determined by the motion vector of a picture.