PC-based two-dimensional binocular stereo imaging system (BSIS)

Date of Publication

1995

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Computer Science with Specialization in Computer Technology

College

College of Computer Studies

Department/Unit

Computer Technology

Abstract/Summary

Computer vision is an image processing technique with the use of computers as windows towards the real world. This technique is of great significance in the field of Robotics and Artificial Intelligence, breaking their barriers by adding a variety of functions to the robot eye.

The main goal of this study is to develop an experimental system called BSIS with functions for range detection of objects using stereo vision techniques, for size approximation of objects, and for classification of objects based on aspect ratio. Two images are captured with the use of a video camera which is connected to a video grabber. An auto-lighting circuit was constructed to be able to provide proper lighting condition. The resulting digitized left and right images are used as inputs to the software of BSIS.

The software manipulates the bitmap images and calculates the range of the objects. To achieve the said goal, a study on digital image processing techniques and concepts such as thresholding and segmentation, region labelling, feature extraction, feature matching, and disparity generation must be versed to come up with a software system that would detect range.

Results of this study show the advantages and disadvantages of different image processing techniques. The proponents were able to develop additional algorithms that showed good results in program implementation. BSIS gives an 80-100 percent range detection accuracy. Although some limitations were discovered along the way, the objectives of the study were met.

Major recommendations include the use of parallel processing to make the system work in real time, automatic positioning of camera to lessen errors, object recognition, and detection of depressions, and overlapping objects.

Abstract Format

html

Language

English

Format

Print

Accession Number

TU09120

Shelf Location

Archives, The Learning Commons, 12F, Henry Sy Sr. Hall

Physical Description

1 v. (various pagings)

This document is currently not available here.

Share

COinS