Filtering decimation technique in a digital video system

Abstract

A video decoder decimates an input image to produce a decimated output image. The video decoder uses approximately every line of pixels in the input image to compute the lines of pixels in the decimated image. The video decoder includes a vertical decimation filter that computes an average, and preferably a weighted average, of luminance (luma) values associated with pixels from each of four lines in the input image. The decimation filter preferably computes a weighted average of lumas from four adjacent lines of pixels from the input image which may represent a frame or a field of video data. The weighted average preferably uses coefficients that weight each luma in the calculation differently. After calculating all of the luma values for a particular line of the decimated image, the line number associated with the first of the four adjacent lines is incremented by four (in a field-based system) to determine the initial line number for calculating the next line in the decimated image. The new initial line number thus represents the first line number of the set of four adjacent lines used to calculate the new line in the decimated image. This technique advantageously results in a high quality decimated image because each line from the initial image is used to compute the decimated image.