| 摘要 |
Disclosed is a time-domain information-based adaptive video pre-processing method, comprising the following steps: step S1, reading a frame of video image data; step S2, obtaining a time-domain information-based weighting coefficient of each pixel; step S3, calculating a bilateral filtering value of each pixel; and step S4, enabling a processed value of a pixel point to be taken from an original pixel and a weighted sum of a bilateral filtering value of this pixel according to the weighting coefficients and the bilateral filtering values, so as to obtain a video pre-processing value. The present invention processes an original image based on a bilateral filtering value and a weighting coefficient of each pixel, an obtained value being a weighted sum of the bilateral filtering value and an original pixel value. A boundary saving effect of maintaining time-domain information thereof unchanged in an accurate motion estimation and relatively-stationary region and achieving noise reduction by means of bilateral filtering in a severe motion estimation and relatively-inaccurate residual estimation region is achieved. A time-domain correlation between front and back video frames can be retained to the greatest extent, a code rate is obviously reduced, and the subjective quality is obviously improved. |
