Abstract: A program for generating a significant video stream causes a computer to function as coding parameter extraction means for extracting a coding parameter of each macroblock for each frame from an original video stream, macroblock selection means for selecting a significant macroblock that has a coding parameter satisfying a predetermined condition, and significant video stream generation means for generating a significant video stream in which frames of the original video stream temporally synchronized with the frames of the coding parameter in the significant macroblocks are combined in time series.

Abstract: Provided are techniques which offer an advantage of reduced time and reduced storage capacity required to calculate the feature value of AAC-format song data. A feature extraction unit includes: an MDCT coefficient extraction unit which extracts MDCT coefficients from AAC-format song data; a classification unit which locates the MDCT coefficients thus extracted by the MDCT coefficient extraction unit on Mel frequency regions so as to uniformly classify the MDCT coefficients into classes, the number of which is the same as that of a predetermined number of Mel filter banks; an integrating unit which extracts the MDCT coefficients classified by the classification unit by applying a predetermined window function, and integrates the MDCT coefficients thus extracted, in increments of the Mel filter banks; and a feature calculation unit which calculates the feature value by performing logarithmic cosine conversion of the integrated results obtained by the integrating unit.

Abstract: The present invention provides a scene classification apparatus for classifying uncompressed or compressed video into various types of scenes at low cost and with high accuracy using characteristics of a video and audio characteristics accompanied by the video. When video are compressed data, their motion intensity, spatial distribution of motion and histogram of motion direction are detected by using values of motion vectors of predictive coding images existing in respective shots, and the respective shots of the video are classified into a dynamic scene, a static scene, a slow scene, a highlight scene, a zooming scene, a panning scene, a commercial scene and the like based on the motion intensity, the spatial distribution of motion, the histogram of motion direction and shot density.

Abstract: The picture input section 12 fetches image data from the storage device 11, and then, inputs image data on sequential n images into the section 13 for measuring correlation between sequential images and the section 14 for measuring correlation between sampled images. The section 13 for measuring correlation between sequential images measures the correlation between sequential images based on video data. The section 14 for measuring correlation between sampled images measures the correlation between sub-sampled images. The short shot determining section 15 determines short cut points by using the correlation between the sequential images and the correlation between the sub-sampled images. Otherwise, short cut points may be determined in consideration of motion amount between the sequential images or the sub-sampled images.

Abstract: An apparatus for retrieving music that matches with a user preference across a plurality of music is provided. The apparatus comprises a feature vector generating device for generating feature vectors of each of the music, a clustering device for clustering the music into a plurality of clusters according to the feature vectors, a music extraction device for extracting a predetermined number of music from each of the clusters, a music selection accepting device for accepting a selection of music by the user from the music, and a music retrieval device for calculating similarities between a feature vector of the selected music and feature vectors of each of the music, and retrieving music that have higher similarities to the selected music among the music.

Abstract: The picture input section 12 fetches image data from the storage device 11, and then, inputs image data on sequential n images into the section 13 for measuring correlation between sequential images and the section 14 for measuring correlation between sampled images. The section 13 for measuring correlation between sequential images measures the correlation between sequential images based on video data. The section 14 for measuring correlation between sampled images measures the correlation between sub-sampled images. The short shot determining section 15 determines short cut points by using the correlation between the sequential images and the correlation between the sub-sampled images. Otherwise, short cut points may be determined in consideration of motion amount between the sequential images or the sub-sampled images.

Abstract: A first block size motion searching device (4) detects a motion vector with respect to each of blocks by a unit of variable blocks in the smallest first block size. A feature value extracting device (6) detects a feature value between the adjacent blocks. A motion search determining device (7) determines by utilizing the feature value whether or not the motion search should be executed in the second to nth block sizes. A search center generating device (8), a search area generating device (9) and a selected block size motion searching device (10) selects the second to nth block sizes and detects the motion vector. A block size determining device (11) determines the block size having a high efficiency.

Abstract: The present invention intends to provide a classification apparatus for broadcasted sport videos for classifying various scenes constituting a sport video, depending on the importance level without being limited to any particular sport video by using visual and audio features. A sport video classification processing portion determines and classifies a play scene, a close-up shot and the like using an audio and a color feature extracted from motion picture data. Because the determination of the most frequent color can be changed adaptively for each inputted sport video, it is not necessary to determine any particular color to the most frequent color previously.

Abstract: The picture input section 12 fetches image data from the storage device 11, and then, inputs image data on sequential n images into the section 13 for measuring correlation between sequential images and the section 14 for measuring correlation between sampled images. The section 13 for measuring correlation between sequential images measures the correlation between sequential images based on video data. The section 14 for measuring correlation between sampled images measures the correlation between sub-sampled images. The short shot determining section 15 determines short cut points by using the correlation between the sequential images and the correlation between the sub-sampled images. Otherwise, short cut points may be determined in consideration of motion amount between the sequential images or the sub-sampled images.

Abstract: The present invention provides a real-time live picture information compression-transmission apparatus and method using a connection-less type protocol for a computer network. The compression-transmission apparatus is constituted to divide frame data output from a real-time encoder compressing live picture information into packets, and to transmit the live picture information in a real time manner according to the connection-less type protocol while suppressing burst transmission by conducting transmission control. The present invention also provides a compressed video data decoding apparatus capable of reducing compressed video decoding processing load without deteriorating picture quality and decreasing the number of played back picture planes when playing back a picture.

Abstract: Provided are techniques which offer an advantage of reduced time and reduced storage capacity required to calculate the feature value of AAC-format song data. A feature extraction unit includes: an MDCT coefficient extraction unit which extracts MDCT coefficients from AAC-format song data; a classification unit which locates the MDCT coefficients thus extracted by the MDCT coefficient extraction unit on Mel frequency regions so as to uniformly classify the MDCT coefficients into classes, the number of which is the same as that of a predetermined number of Mel filter banks; an integrating unit which extracts the MDCT coefficients classified by the classification unit by applying a predetermined window function, and integrates the MDCT coefficients thus extracted, in increments of the Mel filter banks; and a feature calculation unit which calculates the feature value by performing logarithmic cosine conversion of the integrated results obtained by the integrating unit.

Abstract: The present invention relates to the video playback unit of one or a plurality of videos, the delivery unit and the recording medium for recording the program of the processing of these units. The video playback unit of one video reads in order the video data of the video file from the storage unit. The display displays the video data, the time information sequence data and the still image sequence data positioned before and after the playback time of the video data. In this way, the video scene contained in the video file stored in the storage unit and the video file within the server network-connected can be effectively read or the target scene can be effectively retrieved. On the other hand, the playback unit of a plurality of videos divides a plurality of videos read from the storage unit into the main videos and the proxy videos so that these videos can be simultaneously displayed. In this way, a plurality of videos can be played back even in a limited transmission bandwidth and decoding capacity.

Abstract: An apparatus for retrieving music that matches with a user preference across a plurality of music is provided. The apparatus comprises a feature vector generating device for generating feature vectors of each of the music, a clustering device for clustering the music into a plurality of clusters according to the feature vectors, a music extraction device for extracting a predetermined number of music from each of the clusters, a music selection accepting device for accepting a selection of music by the user from the music, and a music retrieval device for calculating similarities between a feature vector of the selected music and feature vectors of each of the music, and retrieving music that have higher similarities to the selected music among the music.

Abstract: In an information distribution/reproduction control apparatus, a plurality of stream servers distribute real-time reproducible stream information to a plurality of clients through a network like the Internet. A transit control server controls the stream servers for the distribution of the stream information, and also controls the clients for real-time reproduction of the stream information.

Abstract: The present invention provides a real-time live picture information compression-transmission apparatus and method using a connection-less type protocol for a computer network. The compression-transmission apparatus is constituted to divide frame data output from a real-time encoder compressing live picture information into packets, and to transmit the live picture information in a real time manner according to the connection-less type protocol while suppressing burst transmission by conducting transmission control. The present invention also provides a compressed video data decoding apparatus capable of reducing compressed video decoding processing load without deteriorating picture quality and decreasing the number of played back picture planes when playing back a picture.

Abstract: This invention provides a video information transmission apparatus for efficiently transmitting a digital video such as MPEG video at real time while controlling congestion on a QoS non-guaranteed IP network and suppressing degradation of a video quality. A transmission control section 13 on a sender side outputs bit rate feedback information in accordance with congestion information on the network to a real-time encoder 12, and controls a transmission bit rate to change the transmission bit rate in accordance with congestion information on the network. The bit rate feedback information is obtained based on the congestion information on the network on the sender side, or obtained on a receiver side and fed back.

Abstract: A feature description method capable of high-speed, efficiently searching audio data or grasping a summary of the audio data by giving considerations to elements and characteristics peculiar to the audio data, is provided. Also, an audio video data feature description collection construction method for collecting feature descriptions from multiple pieces of audio video data based on a specific feature type makes it possible to efficiently, clearly describe a feature description collection.

Abstract: A variance between sequential video pictures is extracted, and then, a GOP boundary position is decided based on inter-frame variance information. Furthermore, simple motion estimation is carried out with respect to video pictures inside one GOP. If a motion variation between the video pictures is large, a small predictive frame interval is taken; to the contrary, if the motion variation is small, a large predictive frame interval is taken. The simple motion estimation is carried out between two downscaled feature planes at a timewise fixed interval with respect to a video picture which is discriminated to be an interlaced video picture, wherein a motion compensatory prediction error at that time is output as image variance information. If the image variance is small, coding is conducted by a frame structure; to the contrary, if the image variance is large, the coding is conducted by a field structure.

Abstract: The present invention provides a real-time live picture information compression-transmission apparatus and method using a connection-less type protocol for a computer network. The compression-transmission apparatus is constituted to divide frame data output from a real-time encoder compressing live picture information into packets, and to transmit the live picture information in a real time manner according to the connection-less type protocol while suppressing burst transmission by conducting transmission control. The present invention also provides a compressed video data decoding apparatus capable of reducing compressed video decoding processing load without deteriorating picture quality and decreasing the number of played back picture planes when playing back a picture.

Abstract: A variance between sequential video pictures is extracted, and then, a GOP boundary position is decided based on inter-frame variance information. Furthermore, simple motion estimation is carried out with respect to video pictures inside one GOP. If a motion variation between the video pictures is large, a small predictive frame interval is taken; to the contrary, if the motion variation is small, a large predictive frame interval is taken. The simple motion estimation is carried out between two downscaled feature planes at a timewise fixed interval with respect to a video picture which is discriminated to be an interlaced video picture, wherein a motion compensatory prediction error at that time is output as image variance information. If the image variance is small, coding is conducted by a frame structure; to the contrary, if the image variance is large, the coding is conducted by a field structure.