Visual Systems -- Interests

Interests

Image and Video Coding: We have developed a general-purpose still-image coding scheme based on quincunx pyramids. The method has been licensed and adopted by industry partners in Europe and North America, and work is continuing on applying it to moving pictures. We are also studying quincunx wavelets and mixture-of-principal-components representations for general-purpose image compression. Other recent work has concentrated on object-based coding and error resilience for video.

Augmented Reality: We are investigating both wearable augmented reality, where a user views the world through a head-mounted display with graphical overlays registered to the scene, and projected augmentation. In the latter case we use algorithms for finding objects (such as documents) that may be haphazardly arranged on a table top, recognizing and reading those objects, then projecting down augmentation information, or playing audio, in response to a user's actions.

Interactive Television: We are exploring ways to enable multiple viewers of a single iTV system to each be provided with an individualised experience without disenfranchising the other viewers. We are looking at the use of an augmented reality display system, consisting of a head-mounted display and a wearable computer for each user, to provide additional 'virtual' screens around the main presentation device (television, cinema screen, etc.) This would enable viewers to access and interact with additional information such as subtitles, graphics or alternative camera angles to customise and enhance the presentation without affecting it for others.

Automated video content analysis and indexing: We are exploring automated video indexing and scene characterization methods that meet the usage requirements of broadcast television and feature film editors. In particularly, we are looking at the application of perspective estimation methods, originally developed for augmented reality applications, for cut detection and camera movement parsing, in conjunction with temporal filtering to resolve complex subject motion. Ultimately we will be examining ways of obtaining semantic information from footage as well.

Investigative virtual reality: We are interested in the design of virtual reality (VR) and 3D display systems that simulate and investigate complex forms of human activity, such as walking, hand reaching, cycling or driving. Understanding the underlying physiological (sensory/motor) and psychological (perceptual/cognitive) mechanisms and their interactions illuminates the nature of the neural pathways and benefits application in engineering and medicine, such as the design and evaluation of image augmentation devices for the visually impaired. We also study simulator after-effects and the mal-adaptation that sometimes occurs in VR, for instance as a result of visual lags, towards the overall aim of improving the design and safety of virtual systems and enhancing the realism and presence within them.

Visual interpretation of complex images: We are investigating the visual properties of complex images including natural scenes. Models of human vision and psychophysical experiments on still images and motion sequences identify the local and global properties of images, and the mechanisms by which the user detects luminance contrast and high-order object features, such as distance, colour, shape and size. These studies are motivated by applications in display design and image coding, including the detection of digitisation artefacts, embedding watermarks, and optimising the use of computing resources.

Please contact us to learn more.



				Interests Image and Video Coding: We have developed a general-purpose still-image coding scheme based on quincunx pyramids. The method has been licensed and adopted by industry partners in Europe and North America, and work is continuing on applying it to moving pictures. We are also studying quincunx wavelets and mixture-of-principal-components representations for general-purpose image compression. Other recent work has concentrated on object-based coding and error resilience for video. Augmented Reality: We are investigating both wearable augmented reality, where a user views the world through a head-mounted display with graphical overlays registered to the scene, and projected augmentation. In the latter case we use algorithms for finding objects (such as documents) that may be haphazardly arranged on a table top, recognizing and reading those objects, then projecting down augmentation information, or playing audio, in response to a user's actions. Interactive Television: We are exploring ways to enable multiple viewers of a single iTV system to each be provided with an individualised experience without disenfranchising the other viewers. We are looking at the use of an augmented reality display system, consisting of a head-mounted display and a wearable computer for each user, to provide additional 'virtual' screens around the main presentation device (television, cinema screen, etc.) This would enable viewers to access and interact with additional information such as subtitles, graphics or alternative camera angles to customise and enhance the presentation without affecting it for others. Automated video content analysis and indexing: We are exploring automated video indexing and scene characterization methods that meet the usage requirements of broadcast television and feature film editors. In particularly, we are looking at the application of perspective estimation methods, originally developed for augmented reality applications, for cut detection and camera movement parsing, in conjunction with temporal filtering to resolve complex subject motion. Ultimately we will be examining ways of obtaining semantic information from footage as well. Investigative virtual reality: We are interested in the design of virtual reality (VR) and 3D display systems that simulate and investigate complex forms of human activity, such as walking, hand reaching, cycling or driving. Understanding the underlying physiological (sensory/motor) and psychological (perceptual/cognitive) mechanisms and their interactions illuminates the nature of the neural pathways and benefits application in engineering and medicine, such as the design and evaluation of image augmentation devices for the visually impaired. We also study simulator after-effects and the mal-adaptation that sometimes occurs in VR, for instance as a result of visual lags, towards the overall aim of improving the design and safety of virtual systems and enhancing the realism and presence within them. Visual interpretation of complex images: We are investigating the visual properties of complex images including natural scenes. Models of human vision and psychophysical experiments on still images and motion sequences identify the local and global properties of images, and the mechanisms by which the user detects luminance contrast and high-order object features, such as distance, colour, shape and size. These studies are motivated by applications in display design and image coding, including the detection of digitisation artefacts, embedding watermarks, and optimising the use of computing resources. Please contact us to learn more.