Lossless video compression

Includes bibliographical references (leaf [61])The objective of this study is to evaluate the compression performance of several variations of the JPEG-LS compression algorithm for video sequence compression, when interframe predictions are employed in addition to the standard intraframe predictions. Intraframe predictor of the JPEG-LS standard uses neighboring pixels of current pixel x to predict its value. Interframe predictions are implemented via the block matching technique. Optimal and sub-optimal search algorithms are investigated for block matching to evaluate the level of compression degradation caused by sub-optimal algorithms. Sub-optimal algorithms used in this study are two-dimensional logarithmic, three-step, orthogonal, and one-at-a-time search algorithms. Fixed and variable-size block matching techniques are employed to reveal their effects on the compression performance of JPEG-LS. High and low-motion video sequences are utilized to evaluate the performance of the algorithms. Performance results for three different cases are presented in this thesis. In the first case, only the intraframe predictions are used in the JPEG-LS. It is important to note that the compression rates obtained from the first case reflect the performance of the standard JPEG-LS algorithm and help us to form a benchmark for comparison purposes. The second case involves using only interframe predictions in the JPEG-LS scheme. In the third and final case, both interframe and intraframe predictions are calculated and the predictions yielding less prediction error are used.M.S. (Master of Science