Video technology promises to be the key for the transmission of motion video. A number of video compression techniques and standards have been introduced in the past few years, particularly the MPEG-1 and MPEG-2 for interactive multimedia and for digital NTSC and HDTV applications, and H.2611H.263 for video telecommunications. These techniques use motion estimation techniques to reduce the amount of data that is stored and transmitted for each frame. This book is about these motion estimation algorithms, their complexity, implementations, advantages, and drawbacks. First, we present an overview of video compression techniques with an emphasis to techniques that use motion estimation, such as MPEG and H.2611H.263. Then, we give a survey of current motion estimation search algorithms, including the exhaustive search and a number of fast search algorithms. An evaluation of current search algorithms, based on a number of experiments on several test video sequences, is presented as well. The theoretical framework for a new fast search algorithm, Densely-Centered Uniform-P Search (DCUPS), is developed and presented in the book. The complexity of the DCUPS algorithm is comparable to other popular motion estimation techniques, however the algorithm shows superior results in terms of compression ratios and video qUality. We should stress out that these new results, presented in Chapters 4 and 5, have been developed by Joshua Greenberg, as part of his M.Sc. thesis entitled "Densely-Centered Uniform P-Search: A Fast Motion Estimation Algorithm" (FAU, 1996).

Real-Time Video Compression: Techniques and Algorithms introduces the XYZ video compression technique, which operates in three dimensions, eliminating the overhead of motion estimation. First, video compression standards, MPEG and H.261/H.263, are described. They both use asymmetric compression algorithms, based on motion estimation. Their encoders are much more complex than decoders. The XYZ technique uses a symmetric algorithm, based on the Three-Dimensional Discrete Cosine Transform (3D-DCT). 3D-DCT was originally suggested for compression about twenty years ago; however, at that time the computational complexity of the algorithm was too high, it required large buffer memory, and was not as effective as motion estimation. We have resurrected the 3D-DCT-based video compression algorithm by developing several enhancements to the original algorithm. These enhancements make the algorithm feasible for real-time video compression in applications such as video-on-demand, interactive multimedia, and videoconferencing. The demonstrated results, presented in this book, suggest that the XYZ video compression technique is not only a fast algorithm, but also provides superior compression ratios and high quality of the video compared to existing standard techniques, such as MPEG and H.261/H.263. The elegance of the XYZ technique is in its simplicity, which leads to inexpensive VLSI implementation of any XYZ codec. Real-Time Video Compression: Techniques and Algorithms can be used as a text for graduate students and researchers working in the area of real-time video compression. In addition, the book serves as an essential reference for professionals in the field.

Real-Time Video Compression: Techniques and Algorithms introduces the XYZ video compression technique, which operates in three dimensions, eliminating the overhead of motion estimation. First, video compression standards, MPEG and H.261/H.263, are described. They both use asymmetric compression algorithms, based on motion estimation. Their encoders are much more complex than decoders. The XYZ technique uses a symmetric algorithm, based on the Three-Dimensional Discrete Cosine Transform (3D-DCT). 3D-DCT was originally suggested for compression about twenty years ago; however, at that time the computational complexity of the algorithm was too high, it required large buffer memory, and was not as effective as motion estimation. We have resurrected the 3D-DCT-based video compression algorithm by developing several enhancements to the original algorithm. These enhancements make the algorithm feasible for real-time video compression in applications such as video-on-demand, interactive multimedia, and videoconferencing. The demonstrated results, presented in this book, suggest that the XYZ video compression technique is not only a fast algorithm, but also provides superior compression ratios and high quality of the video compared to existing standard techniques, such as MPEG and H.261/H.263. The elegance of the XYZ technique is in its simplicity, which leads to inexpensive VLSI implementation of any XYZ codec. Real-Time Video Compression: Techniques and Algorithms can be used as a text for graduate students and researchers working in the area of real-time video compression. In addition, the book serves as an essential reference for professionals in the field.

Video technology promises to be the key for the transmission of motion video. A number of video compression techniques and standards have been introduced in the past few years, particularly the MPEG-1 and MPEG-2 for interactive multimedia and for digital NTSC and HDTV applications, and H.2611H.263 for video telecommunications. These techniques use motion estimation techniques to reduce the amount of data that is stored and transmitted for each frame. This book is about these motion estimation algorithms, their complexity, implementations, advantages, and drawbacks. First, we present an overview of video compression techniques with an emphasis to techniques that use motion estimation, such as MPEG and H.2611H.263. Then, we give a survey of current motion estimation search algorithms, including the exhaustive search and a number of fast search algorithms. An evaluation of current search algorithms, based on a number of experiments on several test video sequences, is presented as well. The theoretical framework for a new fast search algorithm, Densely-Centered Uniform-P Search (DCUPS), is developed and presented in the book. The complexity of the DCUPS algorithm is comparable to other popular motion estimation techniques, however the algorithm shows superior results in terms of compression ratios and video qUality. We should stress out that these new results, presented in Chapters 4 and 5, have been developed by Joshua Greenberg, as part of his M.Sc. thesis entitled "Densely-Centered Uniform P-Search: A Fast Motion Estimation Algorithm" (FAU, 1996).