This paper performs Compressed Sensing (CS) super-resolution for images sampled using pixel binning. It extends Image Sensor Engineering Toolbox (ISET) to accurately simulate the CS acquisition and reconstruction process. Using ISET allows designers to easily explore large design space and measure the performance of CS under different conditions such as lighting, noise etc. without going through the cycle of full system development. CS using binning can make CS programmable and reduce power consumed in image acquisition process by half while maintaining good reconstruction quality. This work shows the results of CS reconstruction with different sensor noise and lighting conditions. It shows that even at low Peak Signal to Noise Ratio (PSNR) of approximately 26 dB and Structural Similarity (SSIM) index of .7, the perceptual quality of image is quite good.

Date Published: 13 May 2019
PDF: 9 pages
Proc. SPIE 10980, Image Sensing Technologies: Materials, Devices, Systems, and Applications VI, 1098012 (13 May 2019); doi: 10.1117/12.2518826

Published in SPIE Proceedings Vol. 10980:
Image Sensing Technologies: Materials, Devices, Systems, and Applications VI
Nibir K. Dhar; Achyut K. Dutta; Sachidananda R. Babu, Editor(s)