Optical disk based neural network

Real-time target recognition and tracking of vehicles moving on a street by matched filtering that uses quantization of modulated function to complexes of a quadrupole

Vehicles moving on a street were monitored by a CCD digital TV camera and recognized in real time by a nonlinear matched filtering method, quantization of modulated function to complexes of a quadrupole, with a personal computer. In the nonlinear matched filtering method an input scene containing the vehicles is modulated in the frequency domain by matched filters, the modulated functions are quantized to complex numbers representing a quadrupole, and then quantized functions are inverse-Fourier transformed. In the frame time of an NTSC-TV system, autocorrelation signals of four vehicles were calculated and discriminated, and their peak positions were determined. Their correct loci in real time were tracked in sequential frames.

Algorithm for optimizing bipolar interconnection weights with applications in associative memories and multitarget classification

An algorithm for optimizing a bipolar interconnection weight matrix with the Hopfield network is proposed. The effectiveness of this algorithm is demonstrated by computer simulation and optical implementation. In the optical implementation of the neural network the interconnection weights are biased to yield a nonnegative weight matrix. Moreover, a threshold subchannel is added so that the system can realize, in real time, the bipolar weighted summation in a single channel. Preliminary experimental results obtained from the applications in associative memories and multitarget classification with rotation invariance are shown.

Optical learning neural network with a pockels readout optical modulator

We have constructed an optical neural-network system with learning capability by using a Pockels readout optical modulator. The system has a two-dimensional structure that permits easy optical alignment and can handle images without scanning. Learning signals are calculated optically with two liquid-crystal devices by a matrix-matrix outer-product method. The calculated learning signals are added directly to the weights memorized on the Pockels readout optical modulator. A two-layer network is implemented, and the learning and recognition of four alphabetical characters are realized according to the delta rule.

Modified signed-digit arithmetic based on redundant bit representation

Fully parallel modified signed-digit arithmetic operations are realized based on redundant bit representation of the digits proposed. A new truth-table minimizing technique is presented based on redundant-bitrepresentation coding. It is shown that only 34 minterms are enough for implementing one-step modified signed-digit addition and subtraction with this new representation. Two optical implementation schemes, correlation and matrix multiplication, are described. Experimental demonstrations of the correlation architecture are presented. Both architectures use fixed minterm masks for arbitrary-length operands, taking full advantage of the parallelism of the modified signed-digit number system and optics.

Incoherent dynamic lenslet array processor

The use of a dynamic lenslet array processor for the implementation of unipolar and bipolar analog inner product, outer product, and vector sum operations is described. Its matrix-vector operations are used as a basis for neural networks and digital circuits. Experimental results of two circuits are presented: a unipolar neural network that computes parity of a 3-bit input word and a digital 3-to-8 decoder circuit.

Optical disk based joint transform correlator

A joint transform correlation system based on optical disks is presented. The operation principle, system considerations, and processing speed are discussed.

Optical neural network with pocket-sized liquid-crystal televisions

A compact optical neural network that uses high-resolution liquid-crystal televisions has been constructed. System design considerations and an experimental demonstration of the liquid-crystal television neural network are reported.