Video preprocessing method and apparatus with selective filtering based

Abstract

Video preprocessing methods and apparatus which utilize motion detection to control selective filtering of pels in a sequence of video frames wherein a video preprocessor generates first and second motion metrics for a given pel in a current frame N by taking the difference between the given pel and corresponding pels in a previous frame N-1 and a subsequent frame N+1, respectively. The motion metrics are converted to first and second bitmaps by thresholding the motion metric value against a first threshold. A motion detection metric is then generated for a given pel by summing the values of the first and second bitmaps for a group of pels which includes the given pel, and comparing the result to a predetermined threshold. The group of pels may include five pels on each of two lines above the given pel, and five pels on each of two lines below a given pel. The motion detection metric may be used to determine whether or not temporal median filtering and/or temporal lowpass filtering operations should be applied to the given pel. The motion detection metric may also be used in conjunction with an edge detection metric and a frame-wise motion activity measure to generate an address into a look-up table. The look-up table specifies a set of filter coefficients for use in a spatial lowpass filtering operation applied to the given pel.