Edward A. Parrish

A camera model for natural scene processing

Abstract A camera model for perspective transformations used in processing three-dimensional natural scenes is presented. An efficient method to determine the required camera offset parameters and a relatively fast calibration procedure are given.

On the Improvement of Position-Detection Accuracy Using Signal Perturbation Theory

It is shown that a triangular perturbation signal provides increased position-detection accuracy over that provided by a sinusoidal perturbation signal [1]. Triangular perturbation is applied to a two-element photosensor receptor, resulting in perfect position detection within the dimensions of the receptor. It is also shown that perfect position detection can be obtained over as large a range as desired by using a many-element zero-deadband one-dimensional receptor subjected to triangular perturbation.

Boundary Location from an Initial Plan: The Bead Chain Algorithm

A new boundary finding method is described and applied to a problem in morphometric cytology. The method is shown to offer several advantages over existing boundary locating techniques. Some experimental results are presented.

A Scheduler for Real-Time Task Control in Microcomputers

A special-purpose task scheduler is described which is suitable for many dedicated, real-time minicomputer and microcomputer applications. The scheduling algorithm is presented in detail. A brief discussion of a basic interrupt-driven real-time executive (RTE) is also included. Its memory size is estimated at about 1 kbyte, using the Intel 8080 instruction set.

Signal-Perturbation Model for Achievement of Optimal Position Transfer Characteristic

Improved position detection involving a one-dimensional photosensor array has been shown to be possible by applying a high-frequency perturbation signal to a normalized target image. A general analytical model which describes the receptor system with respect to receptor element threshold, perturbation amplitude (peak), and the deadspace between elements is presented. Functional-approximation theory is applied to determine necessary and sufficient conditions on the system parameters that optimize (Chebyshev) the system position transfer characteristic with respect to one that is linear.

Optimum Design of a Position Detection System with a Sinusoidal Perturbation Signal

In a paper by McVey and Chen [1] the effect of the amplitude of a sinusoidal perturbtion signal on the accuracy of a position detection system was presented. However, the amplitudes considered were chosen arbitrarily, and no optimum value was specified. This paper presents a method for determining an optimum amplitude for a sinusoidal or, for that matter, any other perturbation signal. In addition, the method may be used to eliminate certain functional forms of a perturbation signal from consideration for a given receptor geometry.

A Semicustom VLSI System Design Course Supported by the General Electric Microelectronics Center

This paper presents an approach to VLSI system design and fabrication using gate arrays which has been implemented at the University of Virginia in conjunction with the General Electric Microelectronics Center (GE-MEC). Material initially developed by GE-MEC was modified and incorporated into a new graduate course to supplement that which already existed in the area of VLSI system design. A software tool set which includes functional and fault simulators for the TEGASTM1 hardware description language, a testability analyzer, and packages for cell placement, channel routing, and system delay extraction was used for chip design. Chip fabrication of successful design projects is the responsibility of GE-MEC.

Signal Perturbation Theory Applied to Boundary Detection

This paper is concerned with accurate detection of the location of boundaries of images projected on a discrete array of photosensors or scanned by a television camera. A model is developed for the detection system and it is shown that proper application of a triangular perturbation signal can yield a linear transfer characteristic between the actual boundary location and the detector output.

Hardware Implementation of Signal Perturbation Theory

Signal perturbation theory has been shown theoretically to reduce the quantization errors in position and velocity detection systems [1-9]. In the above work, some of the ideal constraints on the system were relaxed, but most of the practical problems associated with actual hardware were circumvented. This paper presents an experimental system and compares the experimental results obtained from the system with those predicted by the models.

Signal Perturbation Technique for Achievement of Linear Position Transfer Characteristic

Improved position detection using a one-dimensional photosensor array has been shown [1], [2] to be possible by applying a high-frequency triangular perturbation signal to a normalized target image. This paper develops an adequate model for describing the receptor system with respect to the perturbation amplitude, the receptor element threshold, and the deadband between elements. From the model, it is shown how the parameters can be adjusted to achieve a linear position transfer characteristic when deadband exists between elements.