City Research Online

Automatic recognition of three dimensional planar objects by Hough transform type operations

Purbenyamin Gharachlou, A. (1992). Automatic recognition of three dimensional planar objects by Hough transform type operations. (Unpublished Doctoral thesis, City, University of London)

Abstract

This thesis describes an investigation into the recognition from range data of three dimensional objects with plane surfaces. In it a Hough transform type operation is used to identify objects. This is adapted for three dimensions and uses a voting scheme to identify objects.

First, all available edges of the object present in the scene are extracted. Then, two edges of the object and two lines of a model are taken at a time. These are pruned and potential matching lines are selected. Next, geometric transformations necessary to take them into a fixed position in space are calculated. Matrices resulting from successful matches are computed and stored. The presence of an object similar to a model results in the generation of the same matrices. Recognition is achieved by choosing the model with the highest occurring matrix.

In order to extract edges a vision system is designed and set up. In it a stripe of light generated from the projector together with a camera is employed. A procedure to calibrate the system and extract three dimensional information is devised. Then objects are scanned and from the images taken, coordinates of edge points are computed. Next, edge points are linked and edges of the object are extracted and a recognition algorithm is applied.

The system is tested on objects with varying complexity. Recognition is performed in two different categories. First objects are placed on a specific face. Then they are recognised in arbitrary position and orientation. For each object the results and implications of the recognition algorithm, are investigated. A modified version of the recognition algorithm with two and three connected lines is tested and compared with previous experiments.

Publication Type: Thesis (Doctoral)
Subjects: Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Departments: School of Science & Technology > Computer Science
School of Science & Technology > School of Science & Technology Doctoral Theses
Doctoral Theses
[img]
Preview
Text - Accepted Version
Download (17MB) | Preview

Export

Downloads

Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login