الفهرس | Only 14 pages are availabe for public view |
Abstract A Cellular Neural Network (CNN) is a regular (rectangular, triangular or hexagonal) array of identical dynamical systems; called cells which satisfY two properties: (i) most interactions are local within a finite radius r, and (ii) all state values are continuous valued signals. In this thesis, modified VLSI architecture adaptation of the Cellular Neural Network (CNN) is described. It is based on a combination of MOS transistors operating in weak inversion regime. This combination has enabled a CMOS implementation of a simplified version of the original CNN model by employing a local logic and memory added into each cell that results in providing a simple dual computing structure, analog and digital. A four-quadrant analog multiplier is used as a voltage controlled current source which is fed from the weighting factors of the template elements. The main feature of the multiplier is the high value of the weight voltage range which varies between the ground voltage and the supply voltage. A simulation example for stability of a class of nonreciprocal cellular neural network with opposite-sign template is presented. from this novel compact network architecture, an implementation of an analog VLSI of a cellular neural network for Connected Component Detection (CCD) application is described. The functionality of the proposed network has been verified through SPICE simulations for I-D vectors of arbitrary black-and-white pixels. Also, in this thesis, a novel approach for designing Uncoupled Universal Cellular Neural Network cell capable of’ realizing any arbitrary Boolean functions is presented. Finally, a 10 x 10 -cell CNN chip is reported with simulation results for different image processing tasks. |