Abstract: A system and method for reducing the memory requirements of a decoding subsystem by decoding some pictures multiple times. In a preferred embodiment a video decoder stores only two decoded frames plus a small fraction of a third decoded frame in memory.

Abstract: The objects of this invention are accomplished by combinging the inverse DCT of an appropriate size with scaling on the resulting reconstructed image. In particular, if the data is stored as DCT values of blocks of size P.times.Q and an output image is to be scaled by a factor of R in one dimension and S in the second dimension, then the process is performed in two stages. First a scaling of factor K1/P in the first dimension and a scaling of factor L1/Q in the second dimension are done by inverse transforming with 2-dimensional DCTs of size K1.times.L1. A factor .sqroot.(K1/P).times..sqroot.(L1/Q) is absorbed into a dequantization process prior to the inverse transorm process. Then a scaling of factor K2/K3 in the first dimension and a scaling of factor L2/L3 in the second dimension is done in the spatial domain. The integers K1, K2, K3, L1, L2, L3 are chosen so the (K1K2/K3)=R, (L1L2/L3)=S, (K1/P).gtoreq.R, (L1/Q).gtoreq.S, and the ratios (K2/K3) and (L1/L2) are close to 1.

Abstract: A method and apparatus are provided for reducing the memory requirements of a decoding subsystem by storing reference frames in compressed form. A reference picture in a sequence is decoded, compressed and stored in memory. When the reference frame is needed for motion compensation, it is decompressed.

Abstract: A video decoder/format/frame rate converter with common memory is provided. This device decodes a compressed, interlaced video stream and converts the decoded output to a non-interlaced format and/or a format with a different frame rate.

Abstract: A system and method for performing motion estimation. Interlaced frames are reduced in resolution using field based scaling, and a first search is performed on the reduced resolution frames. This first search is performed by computing motion errors between the fields of a current frame and the fields of a reference frame to obtain vectors that approximately describe inter-field motion. These same motion errors are used to compute frame errors, which are used to obtain vectors that approximately describe inter-frame motion. The approximate field and frame vectors are then refined to get more precise vectors.

Abstract: A system and method for coding a digitized picture is described. The picture is divided into a plurality of regions and then each region is further divided into a plurality of subregions. Linear combinations of pixels in each subregion are then computed. A subregional adaptive quantisation parameter for each of the subregions is then determined based on the courseness tolerance of that subregion as established by processing the linear combinations. A regional quantisation parameter for each region is then determined as a function of the subregional adaptive quantisation parameters and then a quantisation level for each of the regions is determined based on the regional quantisation parameter. Once the regional quantisation level for a region has been determined, it is transform encoded. The transform encoded representations of the regions are concatenated to obtain an encoded representation of the entire picture which can then be decoded for display or storage.

Abstract: Systems and methods are provided for allocating bits to pictures in accordance with the bit allocation constraints for operation at both constant and variable bit rates. The encoding complexity of each picture is modeled based on its spatial and temporal properties. Based on the modeling, a number of bits is allocated in advance to each picture. Each picture is then compressed according to the bits allocated. Based on the difference between the achieved and desired bit rates and on the availability of computational resources, a second pass or multiple passes may be performed to improve the coding.

Abstract: Array processing for carrying out scaled-transform/descaled-inverse-transform procedures on data arrays encoding a visual image or other physical variable makes effective use of fused multiply/add instructions in computer workstations or other digital processors having a fused multiply/add capability. To reconstruct data representative of a physical variable from data originally representative of the variable which had been transformed according to a scaled discrete cosine transform procedure using a predetermined scaling matrix Q, an inverse transform procedure of the invention involves computing a descaling constant and a sequence of descaling/inverse-transform matrices which incorporate elements of the scaling matrix Q. One point of the input data array is multiplied by the descaling constant. The result is matrix multiplied by the sequence of descaling/inverse-transform matrices in turn.

Abstract: A system and methods for implementing a video encoder which for each picture guarantees that the number of bits produced will not exceed a limit. In a preferred embodiment, the system includes two cooperating components or subsystems that together with other components of a video encoder operate to encode the input video, detect when the limit on the number of bits produced so far is close to exceeding the limit on the number of bits that may be produced and reduce the bit rate when such a condition is detected. One aspect of the current invention embodies a macroblock (MB) bit stream generator that can operate in a "low bit rate mode" Another aspect embodies a subsystem for detecting when the number of bits in the picture is close to exceeding the limit.

Abstract: A system and methods for implementing an encoder suitable for use with the motion picture compression. The system includes a motion estimation apparatus which performs temporally interpolative motion estimation. The temporally interpolative motion estimation is performed by first performing unidirectional motion estimation between the current picture and one or more other pictures. The results of the unidirectional motion estimation alone, and not any picture data, are used to perform temporally interpolative motion estimation.