Motion predication method based on image texture and motion features

The present invention provides a method for motion prediction and motion based image texture features, comprising the steps: A: determining a current position of the divided block to be matched (I, J) in the macroblock, and calculate the current to be matched division block edge strength value ESM, I, J: B: determining a motion vector (MVx, MVy) reference blocks, respectively, and the total value of the absolute differences corresponding SADM, I, J (MVx, MVy) MV and the motion vector value, C: set threshold SADTH; D: the total value of the absolute differences determining a reference block SADM, I, J (MVx, MVy) is smaller than the threshold value SADTH, if so, the search is terminated early, the current reference block as the optimal reference block; otherwise returns to step B continue to search for the optimum reference block in the reference block other. The method of the present invention is applied, effectively reducing redundant calculation block matching algorithm, to achieve the purpose of reducing power consumption. The proposed method of the present invention is suitable for hardware implementation, wherein the addition and subtraction, shift operation, and comparison implemented to generate three simple determination threshold value.