Title

Viewpoint and illumination properties extraction for recognizing surfaces, VIPERS

Added Title

VIPERS

Date of Publication

2007

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Computer Science

Subject Categories

Computer Sciences

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Adviser

Jocelyn W. Cu

Defense Panel Member

Clement Y. Ong
Rigan Ap-apid
Jesus E. Gonzalez

Abstract/Summary

The technology in autonomous systems has greatly evolved with the development of vision for mobile robotic systems. Recognizing the environment and sensing the pathways for effective navigation needs planning the most acceptable route from a starting point to the destination. Texture recognition serves as a steppingstone for autonomous vision-based mobile robots. It allows interpretation of images based on the quality of its features, using texture segmentation, classification and feature extraction. Problems occur in segmentation when textures are extracted from a non-homogenous image. In classification, discrepancies happen in processing due to lighting and viewpoint variations. Viewpoint and Illumination Properties Extraction for Recognizing Surfaces (VIPERS), is developed for identification and differentiation of textures from a given image despite fighting and viewpoint variations. A noise removal filter enhances the image before the texture segmentation block separates the image into different textures. A texton labeling algorithm is used for the elimination of features such as lighting and viewpoint. The Chi-Square Probability Function then matches the image to the textures from the database.92% recognition rate for synthetic images is achieved and 60.53% for real world images. The indifference in performance is that the texton vocabulary is encoding general features instead of retaining material-specific information. Viewpoint variances affect the recognition rate more than illumination variations.

Abstract Format

html

Language

English

Format

Print

Accession Number

TU15367

Shelf Location

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

Physical Description

1 v. (various foliations), illustrations, 28 cm.

Keywords

Biosensors; Mobile robots; Robots--Control systems

Embargo Period

4-5-2021

This document is currently not available here.

Share

COinS