Design, implementation and evaluation of a PC-based image analysis system

Date

6-1992

Degree

Bachelor of Science in Applied Physics

College

College of Arts and Sciences (CAS)

Adviser/Committee Chair

Nelio C. Altoveros

Abstract

A PC-based image analysis system capable of object counting, area and perimeter measurement and shape recognition was designed in the study. The main components of the image analysis system are the Frame Grabber board, CCD camera, video camera, AT-type period computer and the image analysis software. The software for the image analysis was implemented using Turbo C. The different features of the image analysis system were tested in terms of their accuracy and execution time. the accuracy of the measurement process was evaluated by comparing the measurement made by the system against the actual values of the parameter of the objects. The result of the measurement process showed that the minimum mean error in area measurement is 0.014 centimeter2 and the maximum mean error is 0.057 centimeter2 The mean error for the perimeter measurement ranges from 0.026 centimeter to 0.064 centimeter. In terms of percent error, the result of the measurement showed that the deviation of measured values from the actual values is less than 2%. Two object counting algorithms were designed, the first one is for counting objects which do not contain holes while the second is a general algorithm which can be used to objects regardless of their structure. The advantage of the first algorithm is that is faster than the second. The execution time of the different routines determined and the result showed that some routines have image dependent execution time while some have constant execution time. The first and second object counting algorithm takes 9 and 27 seconds respectively to complete a counting process. The routines for shape recognition and perimeter measurement is completed in less than a second. The execution time for area measurement is not constant and is linearly proportional to the size of the object in pixels.

Language

English

Location

UPLB Main Library Special Collections Section (USCS)

Call Number

Thesis

Document Type

Thesis

This document is currently not available here.

Share

COinS