New algorithms to enhance 3D imaging resolution
TOF cameras are relatively new devices that capture a whole scene in three dimensions with a dedicated image sensor, and therefore have no need for moving parts. In the project POCS-DSR (POCS based depth super resolution), researchers developed innovative algorithms in a bid to increase depth resolution while reducing depth error in a range-imaging camera. Based on a projection-onto-convex-sets (POCS) technique, researchers used multiframe data correlation to produce a super-resolution depth map. A newly developed calibration-based technique enabled the formation of constraint sets that compensate for spatially varying depth errors within the work volume. An innovative alignment technique enabled the registering of real-time images on a digital signal processor platform. The team devised a multi-exposure depth enhancement algorithm to capture and fuse valuable depth information in the scene using alternating exposure durations. Furthermore, parallel image reconstruction algorithms were developed for POCS. As TOF cameras provide distance images in real time, it is easy to track movements. Researchers developed new algorithms to track moving objects from an aerial platform. Using the longest increasing subsequence technique, they also developed autofocus algorithms. In addition, the team enhanced video contrast algorithms for low-power processors by sampling the original histogram with the help of a massively parallel co-processor. A paper discussing the problem of constrained interest point selection was submitted to a journal. Range sensing applications vary, including the high-end automobile industry and in production lines and biomedical imaging. Regarding biomedical imaging, researchers applied the enhanced resolution algorithms to improve diffusion tensor imaging. Through innovative and efficient algorithms, researchers successfully met the real-time image processing requirements for several applications in the host organisation. Project activities were published in several conference papers at local and international level.
Keywords
3D imaging, time-of-flight, TOF cameras, POCS-DSR, depth resolution, projection-onto-convex-sets
