A biologically-motivated CMOS vision chip for edge detection robust to device mismatches

Human retina is able to detect the edge of the object by effective mechanism. We designed a CMOS vision chip by modeling retinal cells involved in edge detection. There are several fluctuation factors which affect characteristics of MOSFETs during CMOS fabrication process and this effect appears as output offset of the vision chip. The vision chip detecting edge information from input image is used for input stage of vision systems. Therefore, the output offset affects the efficiency of entire vision system. In order to eliminate the output offset, we designed a vision chip utilizing CDS (correlated double sampling) technique. This vision chip could be used at the input stage for actual applications such as target tracking, fingerprint recognition and human-friendly robots, since it is robust to device mismatches.