Parallel signal preprocessing techniques in an image plane, as suggested by the mechanism of human vision, are used to improve the performance of a solid-state imaging sensor. In this paper a 3 × 3 neighborhood operator, by which the gray level of a particular pixel in a transformed image can be determined from the gray levels of the corresponding pixel and its neighborhood in the original image by using appropriate algorithms, is discussed. Dynamic trade-off of signal-to-noise ratio for resolution depending on the light level is realized by combining a newly designed CCD spatial filter with variable weighting circuitry. Improvement in sensitivity by as much as 11 dB at low light levels and edge enhancement at high light levels is obtained only by varying the weighting circuit gain. The design concepts and circuit layouts are also shown together with the performance data on test imagery.
© 1992 Optical Society of America
Original Manuscript: February 19, 1990
Published: April 10, 1992
Takao Ando and Kazuji Wada, "Design and performance of a programmable spatial CCD filter," Appl. Opt. 31, 1762-1768 (1992)