Abstract: An MPEG based video signal compression system reduces picture degradation caused by sub-optimal selection of a prediction mode. A disclosed method produces a compressed video output signal from a video input signal. The video input signal is converted (57) into pixel blocks. Then, a plurality of predicted blocks are formed (60) as representations of an input block. Each predicted block is produced by a different prediction function and has an associated Correlation Factor CF which indicates the degree of correlation between the input block and a predicted block. This factor is obtained (35) by subtracting a measure of a low frequency varying correlation component from an index of correlation. The Correlation Factors are compared (45), and a compressed block associated with the predicted block with the lowest CF value is selected as an output block. This process is repeated for the other input blocks, and the selected compressed output blocks are combined to form the output signal.

Abstract: Video signal compression apparatus includes a memory for storing two fields of image data and providing fields of image data separated by one frame intervals. Frames of video signal are composed from the respective ones of the fields. The field types (e.g., odd or even) of respective fields are determined. Flags are associated with composed frames to indicate if a field of respective frames may be repeated upon reproduction, and which field of the composed frames should be displayed first.

Abstract: A video signal compressor includes apparatus for generating motion vectors for respective blocks of video data. Motion vectors, other than those which provide direction to co-located blocks in other frames, are determined to be candidates for conversion to vectors which do provide direction to co-located blocks. However, conversion is precluded if the vector is determined to be correlated with one of the neighboring motion vectors.

Abstract: Video signal compression apparatus includes a memory for storing two fields of image data and providing fields of image data separated by one frame intervals. Image data from corresponding fields of successive frames are subtracted to generate field differences. The field differences are accumulated over respective field intervals, and the sum is compared against a predetermined value. If a sum of differences over a field is less than the predetermined value, the most recent field is considered to be redundant and may be excised. After respective fields are excised, frames of video signal are composed from the remaining fields. The field types (e.g., odd or even) of the fields in the memory are determined, and the data is accessed from the respective fields in memory such that data from odd and even fields occupy odd and even lines in the composed frames respectively.

Abstract: Video signal compression apparatus includes a memory for storing two fields of image data and providing fields of image data separated by one frame intervals. Image data from corresponding fields of successive frames are subtracted to generate field differences. The field differences are accumulated over respective field intervals, and the sum is compared against a predetermined value. If a sum of differences over a field is less than the predetermined value, the most recent field is considered to be redundant and may be excised. After respective fields are excised, frames of video signal are composed from the remaining fields. The field types (e.g., odd or even) of the fields in the memory are determined, and the data is accessed from the respective fields in memory such that data from odd and even fields occupy odd and even lines in the composed frames respectively.

Abstract: A variable length encoding/decoding system for use with a layered signal format, and for adaptively providing VLC changes without additional signal overhead, includes a plurality of VLC encoding tables, respective ones of the tables being preprogrammed with VLC codes associated with anticipated signal statistics. The layered signal includes blocks of data subject to common parameters which parameters may be associated with signal statistics. The blocks of data include information indicative of such parameters. Each packet is examined for the common parameter and responsive thereto a default VLC coding/decoding table is assigned. In a further embodiment the VLC table used for each successive codeword is made dependent upon the table utilized for the lastmost codeword and the relative value of the lastmost codeword.

Abstract: Apparatus for quantizing compressed viedo data arranged in macroblocks calculates a quantizing parameter independently for each macroblock. The apparatus determines the relative coding complexity of respective macroblocks, and then determines which of the more complex macroblocks can withstand relatively severe quantization without significantly affecting image quality. This determination is made on the basis of the relative amount of codewords in blocks associated with respective macroblocks. In general, the data required to encode each one of the associated blocks is compared to find the block with the minimum data associated with each macroblock. A scale factor is then generated which is a function of the amount of data in the minimum block associated with a macroblock and the average of the data for all the minimum blocks in an image frame. The scale factor is utilized to scale a global quantization value applied to at least a portion of a frame of compressed video data.

Abstract: In a video signal coder blocks of pixels are generated, scanned in a pseudo-random order and transformed blocks of coefficients which are stored in a buffer. The resulting blocks of coefficients are quantized individually by the use of a quantization level and the coefficients of each block are individually quantized by the use of a quantization matrix. The quantization level is determined by the fullness of the buffer.