By Peter M. Kuhn
MPEG-4 is the multimedia regular for combining interactivity, average and artificial electronic video, audio and computer-graphics. common functions are: web, video conferencing, cellular videophones, multimedia cooperative paintings, teleteaching and video games. With MPEG-4 the next move from block-based video (ISO/IEC MPEG-1, MPEG-2, CCITT H.261, ITU-T H.263) to arbitrarily-shaped visible gadgets is taken. this crucial step calls for a brand new technique for approach research and layout to satisfy the significantly larger flexibility of MPEG-4.
movement estimation is a valuable a part of MPEG-1/2/4 and H.261/H.263 video compression criteria and has attracted a lot consciousness in learn and undefined, for the next purposes: it truly is computationally the main hard set of rules of a video encoder (about 60-80% of the whole computation time), it has a excessive effect at the visible caliber of a video encoder, and it's not standardized, hence being open to festival.
Algorithms, Complexity research, and VLSI Architectures for MPEG-4Motion Estimation covers intimately each step within the layout of a MPEG-1/2/4 or H.261/H.263 compliant video encoder:
- Fast movement estimation algorithms
- Complexity research instruments
- Detailed complexity research of a software program implementation of MPEG-4 video
- Complexity and visible caliber research of quick movement estimation algorithms inside MPEG-4
- Design house on movement estimation VLSI architectures
- Detailed VLSI layout examples of (1) a excessive throughput and (2) a low-power MPEG-4 movement estimator.
Algorithms, Complexity research and VLSI Architectures for MPEG-4Motion Estimation is a vital advent to various algorithmic, architectural and approach layout features of the multimedia commonplace MPEG-4. As such, all researchers, scholars and practitioners operating in photo processing, video coding or approach and VLSI layout will locate this booklet of interest.
Read Online or Download Algorithms, Complexity Analysis and VLSI Architectures for MPEG-4 Motion Estimation PDF
Similar computer vision & pattern recognition books
An rising expertise, Speaker reputation is turning into famous for delivering voice authentication over the phone for helpdesks, name centres and different company companies for enterprise procedure automation. "Fundamentals of Speaker popularity" introduces Speaker id, Speaker Verification, Speaker (Audio occasion) type, Speaker Detection, Speaker monitoring and extra.
Many individuals have helped the writer with reviews and corrections, and that i wish to point out D. E. McClure, I. Frolow, J. Silverstein, D. city, and particularly W. Freiberger for his important feedback and encouragement. The paintings in Chapters 6 and seven has been stimulated and motivated through discussions with different participants of the heart for Neural Sciences, in particular with L.
Die Autoren geben eine fundierte Einf]hrung in die wichtigsten Methoden der digitalen Bildverarbeitung. Dabei steht die praktische Anwendbarkeit im Vordergrund, formale und mathematische Aspekte sind auf das Wesentliche reduziert, ohne dabei auf eine prdzise und konsistente Vorgehensweise zu verzichten.
Computerized Autocorrelation and Spectral research supplies random info a language to speak the knowledge they comprise objectively. within the present perform of spectral research, subjective judgements must be made all of which impact the ultimate spectral estimate and suggest that varied analysts receive diversified effects from a similar desk bound stochastic observations.
- Multiple View Geometry in Computer Vision
- Computer vision and Fuzzy-Neural Systems
- Advanced Quantum Communications: An Engineering Approach
- Pattern Recognition: 6th Chinese Conference, CCPR 2014, Changsha, China, November 17-19, 2014. Proceedings, Part II
- Computational and Cognitive Neuroscience of Vision
Additional resources for Algorithms, Complexity Analysis and VLSI Architectures for MPEG-4 Motion Estimation
The selection ofsl=4, s2=2, and s3=1 allows a search distance of7 pels for each coordinate. Subpel MY calculation is regarded to be oflow complexity, as the interpolation of projections of one dimension is equivalent to projections after interpolation [Sauer 96]. As the theory of integral projections would require images of infinite extent, [Sauer 96] introduced a weighting of the horizontal and vertical pel sum H, k(m) and VI k(m) to reduce boundary effects. H, k( m) and V, k( m) are weighted according to the PDF (probability density function) of the motion 'vectors, which results in smaller weights at the block boundary compared to the block center.
VLSI implementation. Basically the following distance criteria can be used in any fast motion estimation algorithm. CCF (Cross-Correlation Function) The CCF (Cross-Correlation Function) is derived from the correlation between the two random variables Ik(m, n) and Ik_J(m + dx, n + dy), and is depicted in eq. 20) [Rao 90] p242, [Fuhrt 97] pS8. As the CCF suffers from high computational 30 CHAPTER 2 complexity, the practical importance for real-time video codecs is negligible. 20) MSE (Mean Square Error Function) The MSE (Mean Square Error Function) is known to produce superior results, as the MSE can be interpreted as Euclidean distance between two MBs, which is close to the human visual perception [Haus 94].
NTSS: New Three Step Search (= N3SS) The NTSS algorithm was proposed by [RLi 94] in order to improve the well-known TSS algorithm especially for video content with slow motion as found in typical headand shoulder-sequences. As TSS uses a uniform search area subsampling pattern with 2 pel between the search locations in the first step, TSS was regarded to be inefficient in terms of computational complexity for the estimation of small motion [Po 96]. Experimental results in [RLi 94] showed that the block motion field of real world image sequences was usually gentle, smooth, and slowly varying, resulting in a center-biased global minimum motion vector distribution, instead of an uniform distribution.