Use of multiple motion vectors in video compression

With the recent advancements and changes in technology, the demand for different standards and algorithms in video compression has increased. Therefore, the primary purpose of using multiple motion vectors in video compression is to generate the block that is most similar to the original block during the motion estimation process between frames. Although the method used requires additional complexity (computation) to calculate the weights of multiple reference blocks on the encoder side and extra bits to transmit the motion vectors, using multiple reference blocks, particularly in complex and detailed scenes, can provide more accurate motion predictions, reduce motion blur, and enable the acquisition of clear and high quality images. In the doctoral thesis, various tests were conducted to calculate the weights of multiple motion vectors using the least squares method. These tests are summarized in three points; the first point involves calculating each weight using the least squares method, the second point involves keeping one weight constant and calculating the remaining weights, and the third point involves clustering the calculated weights using the K-means algorithm to find sets of weights. Through the applied tests and evaluations, it has been observed that the use of multiple motion vectors in video compression leads to an improvement in video image quality.