Image and Video Processing research has been disrupted by the emergence of Deep Learning as a ubiquitous tool. This is especially the case in pixel level manipulation of video and images e.g. denoising. Nevertheless, the traditional topics of Digital Signal Processing (DSP) still have a role to play in the development of efficient processing pipelines, and especially in creating and supporting explainable systems. Video quality in capture and display has also never been higher, and devices are pushing the boundaries of brightness (displays), pixels (resolutions), and speed/battery life. The pandemic has brought timeliness and relevance to technologies that support multistream video conferencing over variable bandwidth connections, ensuring that DSP remains at the heart of practical systems for enhancement and streaming.
In this Research Topic, we address the impact that DSP continues to have in the areas of computational video and video streaming. We want to expose the ideas, originating in the field of DSP, that continue to have success in these areas because of the ability to deliver explainable, reliable, and energy efficient pipelines. We bring the whole pipeline to this issue: streaming because that is what the immediate needs of videoconferencing have pushed to the front, and computational video because that underpins the making of “better pixels”.
We invite articles in the areas of:
• Video quality improvement: denoising, enhancement, superresolution, stabilisation, frame rate conversion
• Video quality measurement, especially perceptual quality measurement in the context of practical pipelines
• Energy efficient/aware compression and enhancement for video
• Techniques that combine Machine Learning and traditional DSP toolsets
• Pre and post processing for improving efficiency of streaming workflows
• Real time visual effects in video conferencing e.g. auto-background blurring, background replacement, selective foreground manipulation
• Tools in video compression, rate control, pre and post processing
• Machine Learning and DSP algorithms for controlling the infrastructure for streaming
• Tools for emerging video compression schemes
• Adaptive bitrate streaming
• Video denoising in post production and streaming
• Video frame interpolation combining traditional DSP with Machine Learning
• High dynamic range video processing
Topic Editor Feng Yang is a Senior Staff Software Engineer at Google Research. The remaining Topic Editors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Image and Video Processing research has been disrupted by the emergence of Deep Learning as a ubiquitous tool. This is especially the case in pixel level manipulation of video and images e.g. denoising. Nevertheless, the traditional topics of Digital Signal Processing (DSP) still have a role to play in the development of efficient processing pipelines, and especially in creating and supporting explainable systems. Video quality in capture and display has also never been higher, and devices are pushing the boundaries of brightness (displays), pixels (resolutions), and speed/battery life. The pandemic has brought timeliness and relevance to technologies that support multistream video conferencing over variable bandwidth connections, ensuring that DSP remains at the heart of practical systems for enhancement and streaming.
In this Research Topic, we address the impact that DSP continues to have in the areas of computational video and video streaming. We want to expose the ideas, originating in the field of DSP, that continue to have success in these areas because of the ability to deliver explainable, reliable, and energy efficient pipelines. We bring the whole pipeline to this issue: streaming because that is what the immediate needs of videoconferencing have pushed to the front, and computational video because that underpins the making of “better pixels”.
We invite articles in the areas of:
• Video quality improvement: denoising, enhancement, superresolution, stabilisation, frame rate conversion
• Video quality measurement, especially perceptual quality measurement in the context of practical pipelines
• Energy efficient/aware compression and enhancement for video
• Techniques that combine Machine Learning and traditional DSP toolsets
• Pre and post processing for improving efficiency of streaming workflows
• Real time visual effects in video conferencing e.g. auto-background blurring, background replacement, selective foreground manipulation
• Tools in video compression, rate control, pre and post processing
• Machine Learning and DSP algorithms for controlling the infrastructure for streaming
• Tools for emerging video compression schemes
• Adaptive bitrate streaming
• Video denoising in post production and streaming
• Video frame interpolation combining traditional DSP with Machine Learning
• High dynamic range video processing
Topic Editor Feng Yang is a Senior Staff Software Engineer at Google Research. The remaining Topic Editors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.