Authors:
G. Pabón Rodríguez
;
G. Andreu-García
;
A. Rodas-Jordá
;
J. Valiente-González
and
F. Acebrón-Linuesa
Affiliation:
Universidad Politécnica de Valencia, Spain
Keyword(s):
Surface inspection, Computer vision system, 3D defect detection, Quality control, Range images.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Biomedical Signal Processing
;
Computer Vision, Visualization and Computer Graphics
;
Data Manipulation
;
Feature Extraction
;
Features Extraction
;
Health Engineering and Technology Applications
;
Human-Computer Interaction
;
Image and Video Analysis
;
Informatics in Control, Automation and Robotics
;
Methodologies and Methods
;
Neurocomputing
;
Neurotechnology, Electronics and Informatics
;
Pattern Recognition
;
Physiological Computing Systems
;
Segmentation and Grouping
;
Sensor Networks
;
Signal Processing, Sensors, Systems Modeling and Control
;
Soft Computing
Abstract:
In this paper we propose a system to characterize 3D defects of range images, which can be combined with traditional surface inspection methods in an industrial environment for ceramic tiles inspection. Our application has the advantage of learning the geometric features of the ceramic pieces, creating a unique 3D model against which we compare the test pieces. In addition to this, the system includes a robust learning phase, which discards tiles with defects impossible to see from a human expert and a more stringent inspection in areas with low uncertainty. Experiments with real data were performed. Our data consist of tiles of different types, shapes and silk-screen of ceramic tiles. Results are promising for tiles with a straight orientation, over 99 % of defects are correctly classified.