A CMOS imager with real-time frame differencing and centroid computation

We present a novel imager that computes real-time frame difference as well as several image statistics. We desire to use a network of these imagers connected to microprocessors for computation on images. To reduce the bandwidth to the processors, each imager has on-chip computation to determine relevant information and sends only that to the microprocessors. We have developed an imager that uses simple analog elements to implement the desired functions. We present experimental results.

[1]  Stephen P. DeWeerth,et al.  Analog VLSI circuits for stimulus localization and centroid computation , 1992, International Journal of Computer Vision.

[2]  Carver Mead,et al.  Analog VLSI and neural systems , 1989 .

[3]  David V. Anderson,et al.  Programmable and adaptive analog filters using arrays of floating-gate circuits , 2001, Proceedings 2001 Conference on Advanced Research in VLSI. ARVLSI 2001.

[4]  E. Fossum,et al.  CMOS active pixel image sensors for highly integrated imaging systems , 1997, IEEE J. Solid State Circuits.

[5]  Paul E. Hasler,et al.  Adaptive circuits using pFET floating-gate devices , 1999, Proceedings 20th Anniversary Conference on Advanced Research in VLSI.