This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. The early vision system of insects as well as many higher level organisms exhibit interesting phenomena and features such as analog preprocessing, parallel structure, and sub-pixel resolution. Early vision is defined as the vision processes that occur within the first few cellular synapses beyond the photoreceptor layer. These features allow for the rapid extraction of image primitives: object edges, boundaries, image segmentation, and movement parameters. This analog, parallel approach to vision provides advantages over current digital-based imaging system. The same type of object information can be extracted with a digital-based system;however, extraction usually requires multiple passes of image processing techniques that must be exhaustively applied pixel-by-pixel to an image. We propose a new approach to the challenge of vision sensor development which takes its inspiration from the obvious success of biological vision systems. This project will use a similar evolutionary, system-level development that has resulted in robust, adaptable vision for so many biological organisms. In this biologically-based systems approach, the sensor (the """"""""eye"""""""") and the computational subsystem (the """"""""visual cortex"""""""") will be developed together. The sensor design and the computational algorithm design will be made to evolve together as a synergistic, mutually optimized pair;we believe this will greatly increase the probability that successful computer vision will be achieved for a wide variety of medical, commercial, and military applications. This project will advance the state of science in several ways. First, greater understanding of biological vision will be a benefit. The preprocessing in retinal neural layers and the final processing in the visual cortex is only partially understood today. By creating analog circuitry that accomplishes some of the preprocessing and computational algorithms that implement the final processing, more complete knowledge of biological vision will be obtained. Second, a more capable and robust computer vision system suitable for intelligent navigation to be achievable for a wide variety of mobile applications such as autonomous wheelchairs and robot movement in hazardous areas is expected. Furthermore, we believe this research will provide the foundation for the development of a vision prosthetic system.
Showing the most recent 10 out of 84 publications
