Paradigm Shifts in Video Technologies: Introduction to T-CSVT Future Special Issues

2014 IEEE transactions on circuits and systems for video technology (Print)  
R ECENTLY, we have witnessed many exciting technology and methodology advances that will deeply impact the way we do research and development. These advances imply many important new research directions in the next 5-10 years, and potentially introduce paradigm shifts in video technologies-visual signal processing, communication, and computing. We have observed new trends ranging from breakthroughs in information theory and machine learning, big data, crowd sourcing, cloud computing, low-cost
more » ... t powerful computing end points, Internet of things (IoT), mobile and wireless computing, augmented reality, green computing, to integrated system research. Let us elaborate on each of these trends. Breakthroughs in Information Theory and Machine Learning: With recent developments in compressive sensing theory, machine learning theory, and related aspects of information theory, we envision a complete paradigm shift in visual signal processing, communication, and computing. Signal acquisition may depart from the traditional Nyquist-Shannon sampling theorem. Instead, it may be based on a compressive sensing theory, which exploits the intrinsic highly correlated property in natural visual signals. Many traditional visual processing pipelines will be revisited or changed to embrace this new trend. Visual signal processing, communication, and computing may also shift away from pixel-wise processing in the traditional Hilbert space to structure-, region-, or object-based processing in a new metric space. Advanced developments in machine learning may lead to a mass-knowledge based visual signal processing, communication, and computing systems, instead of traditional memoryless or memory-based signal processing pipelines. A new theory on distributed coding and distributed processing is also emerging to enable independent compression and processing of visual signals with significantly reduced communication overhead, but without degradation of the overall performance. Shannon's information theory may be updated to reflect recent changes in modern visual signal processing, communication, and computing. For example, if we redefine information as new knowledge perceived by the user, a web search operation might be formulated as a filtering process. Big Data-Vast Amount of Data With Embedded Knowledge: The Internet is a massive computing platform with a vast amount of data containing an increasing amount of human knowledge. It has the largest collection of text documents, photos, and audio and video data ever compiled in human history, including a tremendous amount of annotated _____________
doi:10.1109/tcsvt.2014.2332410 fatcat:zdevzkgx6bdkxik5wcxbh2ad3i