Concept, development, mass production, and applications of artificial retina chips

Abstract

Images of the real world contain a very large amount of information which present image processing systems cannot analyze in real time with reasonable cost and low power consumption. In order to solve these problems, we have proposed and developed an artificial retina chip (AR chip) which combines video camera function (image sensing) and image processing function, in a similar way to the human eyes. The basic principle of our AR chip is based on a novel type of optoelectronic vector/matrix multiplication. The core of the AR chip consists of a two-dimensional array of variable sensitivity photodetection circuits. The matrix corresponding to the input image can be processed in analogue by applying an appropriate electric vector to the chip. The possible image processing functions include conventional image sensing, edge extraction, variable resolution, 2D to 1D image compression, pattern matching, noise elimination, and random access. Several types of the AR chips with different resolution have been developed by using CMOS technology. Among them, we have succeeded in the mass production of 128_!_(J128 pixels) in February 1998. The features of the AR chips over conventional CCDs are fast image processing (on chip image processing), low power consumption, low cost. With use of these advantages, 500 million chips have been used up to date in several application areas which include games, security systems, communication systems, etc. The concept, structure, operation principle, some applications, and future trends of the AR chips are introduced.