Image Processing and Pattern Recognition

Cornelius T. Leondes

• Publisher: Academic Press
• Year: 1998
• ISBN:
  9780080551449
• Topic: Technology & Engineering
• Subtopic: Signals & Signal Processing
• Index: Technology & Engineering

Image Processing and Pattern Recognition covers major applications in the field, including optical character recognition, speech classification, medical imaging, paper currency recognition, classification reliability techniques, and sensor technology. The text emphasizes algorithms and architectures for achieving practical and effective systems, and presents many examples. Practitioners, researchers, and students in computer science, electrical engineering, andradiology, as well as those working at financial institutions, will value this unique and authoritative reference to diverse applications methodologies.

Coverage includes:

• Optical character recognition
• Speech classification
• Medical imaging
• Paper currency recognition
• Classification reliability techniques
• Sensor technology

Algorithms and architectures for achieving practical and effective systems are emphasized, with many examples illustrating the text. Practitioners, researchers, and students in computer science, electrical engineering, and radiology, as wellk as those working at financial institutions, will find this volume a unique and comprehensive reference source for this diverse applications area.

Parallel Analog Image Processing: Solving Regularization Problems with Architecture Inspired by the Vertebrate Retinal Circuit

Tetsuya Yagi Kyushu Institute of Technology, 680-4 Kawazu, Iizuka-shi Fukuoka Prefecture 820 Japan

Haruo Kobayashi Department of Electronic Engineering, Gumma University, 1-5-1 Tenjin-cho Kiryu, 376 Japan

Takashi Matsumoto Department of Electrical, Electronics and Computer Engineering, Waseda University, Tokyo 169, Japan

I INTRODUCTION

Almost all digital image processors employ the same architecture for the sensor interface and data processing. A camera reads out the sensed image in a raster scan-out of pixels, and the pixels are serially digitized and stored in a frame buffer. The digital processor then reads the buffer serially or as blocks to smooth the noise in the acquired image, enhance the edges, and perhaps normalize it in other ways for pattern matching and object recognition. There have been several attempts in recent years to implement these functions in the analog domain, to attain low-power dissipation and compact hardware, or simply to construct an electrical model of these functions as they are found in biological systems. For analog implementations, we must evaluate their performance in comparison with their digital counterparts and establish systematic techniques for their design and implementation. These considerations guided the work described here.

Single-chip analog image processor chips consist primarily of resistors and transistors in an array, and sometimes memory elements. A two-dimensional image is sensed by embedded photosensors at nodes in the arra...

Table of contents

• Cover image
• Title page
• Table of Contents
• Copyright page
• Contributors
• Preface
• Pattern Recognition
• Comparison of Statistical and Neural Classifiers and Their Applications to Optical Character Recognition and Speech Classification
• Medical Imaging
• Paper Currency Recognition
• Neural Network Classification Reliability: Problems and Applications
• Parallel Analog Image Processing: Solving Regularization Problems with Architecture Inspired by the Vertebrate Retinal Circuit
• Algorithmic Techniques and Their Applications
• Learning Algorithms and Applications of Principal Component Analysis
• Learning Evaluation and Pruning Techniques
• Index