Abstract: Two implementations of the inverse wavelet transform for use in an image decompression system do not waste computation power on the zero-valued values inserted into the data stream during an upsampling process. The implementation optimized for low-bandwidth applications toggles between even and odd modes each clock cycle. In even/odd mode, the transformed values are multiplied by the even/odd filter coefficients. The implementation optimized for high-bandwidth applications multiplies the transformed values by the even and odd filter coefficients seperately in two sets of multipliers and outputs two different results each clock cycle.

Abstract: A method and system for decoding compressed MPEG2 bitstream video data utilizing a two row macroblock cache memory and algorithm for determining data overlap between a stored reference macroblock and a stored current macroblock to improve memory bandwidth requirements while producing specific decoding results required for MPEG2 video information.

Abstract: A video encoder typically includes a preprocessor, a frame store, a motion compensator, and a compression module. The preprocessor converts an incoming digital video signal to image macroblocks. The frame store stores the macroblocks from anchor frames, and the motion compensator searches neighborhoods in anchor frames for best matches to macroblocks from a current frame. The compression module receives a vector from the motion compensator indicative of the best match and uses it to compress the macroblocks from the current frame. The compressed macroblocks are provided as components of a compressed video output bitstream. To perform the neighborhood search, the motion compensator accesses many neighborhood macroblocks from the anchor frame for each macroblock from the current frame. To reduce the number of memory accesses, the motion compensator caches the neighborhood macroblocks. Since the search neighborhoods for adjacent macroblocks overlap, caching is effective.

Abstract: Method and apparatus for two row macroblock decoding improves caching efficiency reducing memory bandwidth requirements while simultaneously enabling encoded video information to be decoded in a fast and efficient manner to produce specific results required for MPEG2 video decoding.

Abstract: A method derives edge extensions for wavelet transforms and inverse wavelet transforms of two-dimensional images. The method overcomes the necessity of side computations by treating the two-dimensional matrix of values as a one-dimensional array of values. The use of a one-dimensional array reduces the required flushing and loading of registers by allowing the flushing and loading to be performed in between frames, rather than in between rows or columns of the matrix.

Abstract: A system and method for estimating motion vectors between frames of a video sequence which operates in the DCT domain with improved efficiency and reduced computational requirements. The motion estimation system operates to encode a target block using pointers or motion vectors to a previously encoded block, referred to as the reference block or search block. The system first partitions the target frame into a plurality of target blocks, and DCT transforms the target blocks in the target frame. The motion estimation system then selects a candidate block from the search frame and DCT transforms the selected candidate block. The motion estimation system uses a novel method for selecting candidate blocks which allows re-use of at least a portion of the transformed values of a prior selected candidate block.

Abstract: A system and method implementing a fast wavelet transform by shifting a pair of pixels into a single shift register and using a multimode quadrature mirror filter to eliminate the need of downsampling the filtered signals and to decrease the area required to implement the device on a semiconductor chip.