Masatoshi Nishimura

Biologically inspired vision sensor for the detection of higher-level image features

The paper briefly reviews certain aspects of the biological visual system and presents a smart vision sensor for the detection of higher-level features. The visual system processes information in a hierarchical manner starting from the retina up to the visual cortex. It decomposes the image in simple features (edges, orientation, line stops, corners, etc) using spatial and temporal information. At the higher level it integrates these primitive features, resulting in the recognition of complex objects. The sensor described in the paper is loosely modeled after the visual system and incorporates pixel level, programmable elements which extract orientation, end stops, corners and junctions from a line drawing. The architecture resembles a CNN-UM that can be programmed with a 30-bit word. The 16/spl times/16 pixels array detects these higher-level features in about 54 /spl mu/s.

A compact line and edge orientation detection sensor

Abstract A new line and edge orientation detection sensor using four position-sensitive devices is presented. This sensor is composed of four oblong position-sensing devices configured in a rectangular fashion. Each device detects the center of gravity of a light stripe on the device. By combining these positions obtained from four devices, a line orientation can be calculated. The simulation result demonstrates the translation invariance as well as the linewidth invariance if the sensor dimension is chosen such that the width/length ratio is smaller than or equal to 0.1. A pn diode formed by the n-well and the p-substrate shows the best position-sensing capability among the structures tested due to its relatively high well resistance. The position-sensing capability is investigated with different biases and under different levels of background illumination. Two position-sensing devices are used to measure the line orientation, which agrees well with the actual orientation.

Pattern recognition based on orientation and linestops using an orientation sensor and multilayered neural network

A method for optical pattern recognition which is based on the human visual system and is suitable for hardware implementation is presented. The system is composed of two stages. The first stage detects local features such as line orientation, linestops, corners, and intersections to create a feature map, which represents the number of these features and hence is invariant to position, size, and slight deformation of an input pattern. The next stage is a multilayered neural network that classifies an input pattern to one of predefined categories using the feature map. We have found a method of detecting these features in analog hardware which would considerably speed up the process of pattern recognition. The decomposition of an input pattern into lines with different orientations is done by an array of two-dimensional orientation sensors. We have built an orientation sensor which is invariant to the position, size, and contrast of an input pattern. The generation of the feature map is currently being done in software which receives its inputs from the line orientation sensor. Linestops, corners and intersections are detected after a series of convolution and thresholding operations for each orientation. The convolution operation can be mapped into hardware using a resistive grid technique. The simulation with an example of character recognition showed that the proper selection of convolution kernels and thresholds can detect local features described above and demonstrated the feasibility of a full hardware implementation of a feature detector.

A silicon VLSI optical sensor for image decomposition

Abstract A silicon VLSI optical sensor has been successfully fabricated and is used as a computational front-end of a pattern-recognition system. The sensor decomposes an input image into a set of four oriented line images for further detection of features such as linestops, corners, and intersections. The sensor is composed of an array of six by six orientation sensors. Each orientation sensor classifies the local orientation of a projected image on the sensor into one of four orientations. The orientation sensor is composed of four phototransistors at the corners of the square and a two-layered analog circuit in the middle. The first layer classifies the state of the phototransistors into either ‘on’ (illuminated) or ‘off’ (not illuminated) using a winner-take-all circuit. The use of the winner-take-all circuit enables intelligent classification to take place even under background illumination without externally applying a threshold. The second layer produces four current outputs corresponding to four orientations. The presence of an output indicates the orientation. Experimental results show that the sensor can decompose several input images into four orientations. The sensor also demonstrates its functionality as an edge detector. rights reserved.