Iterative affine transform (IAT) based block matching motion estimation
M K Pushpa, S. Sethu Selvi · 2016
Video transmission or storage needs best encoder to generate motion vectors and decoders to compensate the frame. The block matching motion estimation algorithm such as exhaustive search (ES) algorithm requires more computations to generate motion vectors and compensate the frame in a video sequence. Most of the standard encoders assume translation motion. Using only translation motion model is not providing acceptable image quality. Hence, the proposed algorithm uses iterative affine transform (IAT) based motion estimation, which iteratively predicts affine coefficients for translation, scaling and rotation of every macroblock and compensate the frame using these coefficients. Compared to ES algorithm performance, the saving in average time consumption to estimate and compensate the frame is 95.54% with a slight degradation of 3% in average PSNR quality. IAT is suitable for small and medium motion video sequences.