Abstract: A subpixel interpolator includes an input memory capable of storing video information formed from full pixels. The subpixel interpolator also includes at least one interpolation unit capable of performing subpixel interpolation to generate half-pixels and quarter-pixels in parallel. Multiple half-pixels and multiple quarter-pixels are generated concurrently during the subpixel interpolation. In addition, the subpixel interpolator includes an output memory capable of storing at least some of the full pixels, half-pixels, and quarter-pixels. In some embodiments, the at least one interpolation unit includes a horizontal half-pixel interpolation unit, two vertical half-pixel interpolation units, and a quarter-pixel interpolation unit, all of which may operate in parallel. In particular embodiments, the interpolation units are formed from adders and shifters and do not include any multipliers.

Abstract: A processor includes a multi-stage pipeline having a plurality of stages. Each stage is capable of receiving input values and providing output values. Each stage performs one of a plurality of data transformations using the input values to produce the output values. The data transformations collectively approximate at least one of: a discrete cosine transform and an inverse discrete cosine transform. The stages do not use any multipliers to perform the data transformations.

Abstract: A parallel deblocking filtering method, and deblocking filter processor performing such deblocking, for removing edge artifacts created during video compression. The method includes loading luma samples for a macroblock. Filtering is performed on a set of vertical edges of the macroblock using information in the luma samples, with vertical edge filtering occurring concurrently with the loading of the luma samples. The method also includes filtering a set of horizontal edges of the macroblock using information in the luma samples. The horizontal edge filtering occurs in parallel with vertical edge sampling and with loading operations. The use of parallel and concurrent operations significantly enhances the efficiency of the deblocking method. Storing of filtered samples is also performed in the method, and this storing is performed concurrently with some loading operations as well as filtering operations. Edge filtering includes performing filtering to the H.264 standard and its deblocking filtering algorithm.

Abstract: First and second integer transform matrices can be used to approximate the discrete cosine transform. An input matrix of data is multiplied by a first transform matrix of integers to produce an intermediate matrix of data. The intermediate matrix is multiplied by a second transform matrix of integers to produce a transform result matrix of data. The multiplications by the first and second transform matrices can be pipelined to increase throughput. A plurality of transform data paths can also be provided in parallel to increase throughput.

Abstract: An apparatus and method are provided to accelerate error diffusion for color halftoning for embedded applications. High performance is achieved by utilizing functional parallelism within the halftoning error diffusion process, including exploiting data parallelism in different color planes, reducing the number of memory accesses to the error buffer, accelerating the computation by using a parallel instruction set, and improving the throughput of the system by implementing pipelined architecture. A halftoning coprocessor architecture can implement the foregoing. The architecture can be optimized for high performance, low complexity and small footprint. The coprocessor can be incorporated into embedded systems to accelerate the performance of error diffusion halftoning therein.