Abstract: A method encodes a video by first measuring a variance of pixel intensities in a current frame. A number of bits to encode the current frame is assigned according to rate and buffer fullness constraints. A multiplier value is determined directly as a function of only the variance and the number of bits assigned to the current frame. Motion vectors between a reference frame and the current frame are estimated, and a sum of absolute difference (SAD) is based on a motion compensated residual between the reference frame and the current frame. An encoding mode is determined for each macro block in the current frame based on the sum of absolute difference, the motion vectors and the multiplier value. Then, the motion compensated residual is encoded based on the encoding mode, multiplier value and the number of allocated bits.

Abstract: A method segments a video into objects, without user assistance. An MPEG compressed video is converted to a structure called a pseudo spatial/temporal data using DCT coefficients and motion vectors. The compressed video is first parsed and the pseudo spatial/temporal data are formed. Seeds macro-blocks are identified using, e.g., the DCT coefficients and changes in the motion vector of macro-blocks.

Abstract: A method and system transcodes an input video to a lower spatial resolution. The input video is first decoded into pictures. Each picture includes a set of macroblocks. Each picture sub-sampled to a downscaled picture having a lower spatial resolution. A quantization scale is selected for each macroblock in the downscaled picture. A set of motion vectors is generated for each macroblock in the downscaled picture. A multiplier value based on the quantization scale is determined for each macroblock in the downscaled picture. One of a plurality of encoding modes is selected for each macroblock in the downscaled picture according to the quantization scale, the motion vectors, and the multiplier value. Then, each macroblock in each downscaled picture is encoded according to the quantization scale, the selected encoding mode, and the set of motion vectors to produce an output video having a lower spatial resolution that the input video.

Abstract: A compressed bit-stream represents a corresponding sequence having intra-coded frames and inter-coded frames. The compressed bit-stream includes bits associated with each of the inter-coded frames representing a displacement from the associated inter-coded frame to a closest matching of the intra-coded frames. A magnitude of the displacement of a first of the inter-coded frames is determined based on the bits in the compressed bit-stream associated with that inter-coded frame. The inter-coded frame is then identified based on the determined displacement magnitude. The inter-coded frame includes macro-blocks. Each macro-block is associated with a respective portion of the inter-coded frame bits which represent the displacement from that macro-block to the closest matching intra-coded frame. The displacement magnitude is an average of the displacement magnitudes of all the macro-blocks associated with the inter-coded frame.

Abstract: A method extracts high-level features from a video including a sequence of frames. Low-level features are extracted from each frame of the video. Each frame of the video is labeled according to the extracted low-level features to generate sequences of labels. Each sequence of labels is associated with one of the extracted low-level feature. The sequences of labels are analyzed using learning machine learning techniques to extract high-level features of the video.

Abstract: A method adaptively encodes a video including a sequence of images, where each image is a picture of two fields. Each image is first separated into a top-field and a bottom-field. Motion activity is extracted from the top-field and the bottom-field, and each image is encoded using either frame encoding or field encoding depending on the extracted motion activity.

Abstract: A method adaptively encodes a video including a sequence of images, where each image is a picture of two fields. Each image of the video is encoded as a frame and rate-distortion characteristics are extracted from the encoded frames, while concurrently encoding each image of the video as two fields and rate-distortion characteristics are extracted from the fields. A parameter value &lgr; of a cost function is determined according to the extracted rate-distortion characteristics, and a cost function is constructed from the extracted rate-distortion characteristics and the parameter &lgr;. Then, either frame encoding or field encoding is selected for each image depending on a value of the constructed cost function for the image.

Abstract: A method extracts an intensity of motion activity from shots in a compressed video. The method then uses the intensity of motion activity to segment the video into easy and difficult segments to summarize. Easy to summarize segments are represented by any frames selected from the easy to summarize segments, while a color based summarization process extracts generates sequences of frames from each difficult to summarize segment. The selected and generated frames of each segment in each shot are combined to form the summary of the compressed video.

Abstract: A method transcodes groups of macroblocks of a partially decoded input bitstream. The groups of macroblocks include intra-mode and inter-mode macroblocks. Each macroblock includes DCT coefficients, and at least one motion vector. The modes of each group of macroblocks are mapped to be identical only if there is an inter-mode block and an intra-mode macroblock in the group. If any of the macroblocks in the group are mapped, then the DCT coefficients and the motion vector for such mapped macroblocks are modified in accordance with the mapping to generate reduced-resolution macroblock for an output compressed bitstream to compensate for drift.

Abstract: A method determines distortion in a video by measuring a spatial distortion in coded frames, and by measuring a temporal distortion and spatial distortion in uncoded frames. The spatial distortion of the coded frames is combined with the temporal distortion and the spatial distortion of the uncoded frames to determine a total average distortion in the video.

Abstract: A video is first partitioned into video objects. If the video is uncompressed, then the partitioning is done with segmentation planes. In the case where the video is compressed, a demultiplexer is used for the partitioning. Over time, shape features are extracted from each partitioned object. The extracted shape features are combined to determined a temporal resolution for each object over time. The temporal resolutions are subsequently used to encode or transcode the video objects as an output compressed video.

Abstract: A method decodes a video encoded as a compressed bitstream including frames constructed as blocks. The bitstream is variable length decoded and inverse quantized to determine motion vectors and DCT coefficients of the blocks of the bitstream. The DCT coefficients are then inverse discrete cosine transformed. A low-resolution block is generated for each reconstructed block, and the reconstructed blocks are motion compensated with the low-resolution blocks to decode the video.

Abstract: A surveillance and control system includes a feature extraction unit to dynamically extract low-level features from a compressed digital video signal, a description encoder, coupled to the feature extraction unit, to encode the low-level features as content descriptors. An event detector is coupled to the description encoder to detect security events from the content descriptors, and a control signal processor, coupled to the event detector, to generate control signals in response to detecting the security events.

Abstract: A method transcodes a compressed multi-layer video bitstream that includes a base layer bitstream and an enhancement layer bitstream. The base and enhancement layers are first partially decoded, and then the partially decoded signals are combined with a motion compensated signal yielding a combined signal. The combined signal is quantized into an output signal according to a quantization parameter, and the output signal is variable length encoded as a single layer bitstream. In a preprocessing step, the enhancement layer can be truncated according to rate control constraint, and the same constraints can also be used during the quantization.

Abstract: This invention relates to methods of feature extraction from MPEG-2 and MPEG-4 compressed video sequences. The spatio-temporal compression complexity of video sequences is evaluated for feature extraction by inspecting the compressed bitstream and the complexity is used as a descriptor of the spatio-temporal characteristics of the video sequence. The spatio-temporal compression complexity measure is used as a matching criterion and can also be used for absolute indexing. Feature extraction can be accomplished in conjunction with scene change detection techniques and the combination has reasonable accuracy and the advantage of high simplicity since it is based on entropy decoding of signals in compressed form and does not require computationally expensive inverse Discrete Cosine Transformation (DCT).

Abstract: A method describes motion activity in a video sequence. A motion activity matrix is determined for the video sequence. A threshold for the motion activity matrix is determined. Connected regions of motion vectors at least equal to the threshold are identified and measured for size. A histogram of the distribution of the sizes of the connected areas is constructed for the entire video sequence. The histogram is normalized to characterize the spatial distribution of the video sequence in a motion activity descriptor.

Abstract: A method analyzes a high-level syntax and structure of a continuous compressed video according to a plurality of states. First, a set of hidden Markov models for each of the states is trained with a training video segmented into known states. Then, a set of domain specific features are extracted from a fixed-length sliding window of the continuous compressed video, and a set of maximum likelihoods is determined for each set of domain specific features using the sets of trained hidden Markov models. Finally, dynamic programming is applied to each set of maximum likelihoods to determine a specific state for each fixed-length sliding window of frames of the compressed video.

Abstract: A method for transcoding a compressed video partitions the compressed video into hierarchical levels, and extracts features from each of the hierarchical levels. One of a number of conversion modes of a transcoder is selected dependent on the features extracted from the hierarchical levels. The compressed video is then transcoded according to the selected conversion mode.

Abstract: A dynamically configurable video signal processing system including an encoder and decoder processes data in the form of hierarchical layers. The system partitions data between hierarchical layers and allows variation in the number of layers employed. Data is automatically partitioned into one or more hierarchical layers as a function of one or more parameters selected from available system bandwidth, input data rate, and output signal quality. In addition, the image resolution and corresponding number of pixels per image of the data may be varied as a function of system parameters.

Abstract: A multi-media delivery system for delivering a compressed bitstream through a network to a user device includes a transcoder and a manager. The transcoder is configured to operate on the bit stream using in any one of a plurality of conversion modes. The manager is configured to selecting a particular one of the plurality of conversion modes dependent on semantic content of the bitstream and network characteristics. The system also includes a content classifier to determine the content characteristics, and a model predicator to determine the network characteristics, and user device characteristics. An integrator of the manager generates an optimal rate-quality function to be used for selecting the particular conversion model for a given available bit rate of the network.