Analog pattern classifier with flexible matching circuitry based on principal-axis-projection vector representation

An analog CMOS classifier circuit has been developed for pattern recognition based on a robust image representation algorithm called the Principal Axis Projection method [1]. The short-cut current in a CMOS inverter is modulated by a variety of capacitance coupling techniques, thus enabling real-time altering the functional form of the similarity evaluation. Not only the peak position and the peak height in the bell-shaped current-voltage characteristics but also the sharpness of the bell shape are made alterable for flexible pattern matching operation. Test circuits were designed and fabricated in a 0.6µm double-poly CMOS process. Handwritten pattern matching experiments were carried out and the flexible and adaptive matching capabilities of the circuit were demonstrated.

[1]  Tadashi Shibata,et al.  A functional MOS transistor featuring gate-level weighted sum and threshold operations , 1992 .

[2]  Tadashi Shibata,et al.  A hardware-friendly soft-computing algorithm for image recognition , 2000, 2000 10th European Signal Processing Conference.

[3]  John C. Platt,et al.  An Analog VLSI Chip for Radial Basis Functions , 1992, NIPS.

[4]  Luke Theogarajan,et al.  A multi-dimensional analog Gaussian radial basis circuit , 1996, 1996 IEEE International Symposium on Circuits and Systems. Circuits and Systems Connecting the World. ISCAS 96.

[5]  Tadashi Shibata,et al.  An analog similarity evaluation circuit featuring variable functional forms , 2001, ISCAS 2001. The 2001 IEEE International Symposium on Circuits and Systems (Cat. No.01CH37196).