Accession Number : AD0716690
Title : Man Versus Computer in the Measurement of Size Distribution of Nerve Fibers,
Corporate Author : NATIONAL RESEARCH COUNCIL OF CANADA OTTAWA (ONTARIO) DIV OF MECHANICAL ENGINEERING
Personal Author(s) : Kasvand,T. ; Isonor,D. ; Romero-Sierra,C.
Report Date : JAN 1970
Pagination or Media Count : 223
Abstract : In the study of nervous systems, details of the diameters and size distributions of nerve fibers are frequently required. The various fiber diameters are measured and the size distribution plotted in the form of a histogram. At the present time there are two commonly used methods of producing the desired histograms. One is a purely manual method in which an enlarged photograph is used to facilitiate direct measurement of the diameters or the areas of the fibers. The second is a semi-automatic method, in which an iris is superimposed on the image of the fibers. Using pattern recognition techniques it is possible to completely automate the process of fiber identification. Since the human visual system is adapted to high speed pattern recognition and appears simple, the first Chapter of this report is concerned with the recognition process and why it is so difficult to mechanize. The second Chapter describes the recognition techniques and measurements made on the fibers. Pattern structures leading to mistakes in the recognition of fibers are discussed and methods of correction outlined. The last Chapter gives the results of the comparisons between manual and computerized measurements.
Descriptors : (*NERVE FIBERS, MEASUREMENT), (*PATTERN RECOGNITION, NERVE FIBERS), NERVOUS SYSTEM, IDENTIFICATION, AUTOMATION, COMPUTERS
Subject Categories : Anatomy and Physiology
Distribution Statement : APPROVED FOR PUBLIC RELEASE