Vision is arguably our most crucial sense. Todayâ€™s cameras use a flat CMOS focal plane array that requires complex multi-element optics that flatten the Petzval field curvature of a simple lens. This adds weight and complexity, while limiting the field of view and f/number of the optics. By curving the focal plan array to match the field of curvature of the lens in a way that mimics the shape of the human eye, the imaging system can be made much lighter, smaller, and with minimal optical aberrations. I am developing a hybrid multispectral hemispherical focal plane array (HFPA)
imaging system with extremely low optical aberration and operating at wavelengths from the visible to the NIR (i.e. 400 nm â€“ 1700 nm) in both extremely low light levels and in IR rich environments. The hemispherical focal plane array will have a 140 degree field-of-view(FOV) providing near complete vision in the forward direction. This would, among other advantages, enable noise suppression via averaging over receptive fields, optical flow sensing, and light level adaptation that is variable across the retina. A notable advantage of this curved imaging architecture are its rapid, high resolution, exceptionally large FOV, and night-vision data acquisition capabilities that have the potential to revolutionize our concepts of what is possible in the field of imaging.
Undergraduate student is expected to involved in characterization of artificial eye image sensor by building a PCB board that enable to train the system and to read a signal from sensors.
If student have experience with Matlab and basic circuit, it would be helpful. But, this is not a requirement.
'- Optics for imaging
- Machine learning for self-recognition
- Thin-film semiconductor devices