Abstract

This thesis presents a case of study of the development and performance
analysis of a surface grading application with real-time compliance.
The application focuses on the ceramic tile industry and aims to automate
the inspection process of surface grading and removing human inspectors
from this subjective and tedious task. First, an overview of surface
grading works is given. These works have been reported in recent years
in many production areas such as ceramic tile, marble, granite and
wood industries.

We then address the issue of spatial and temporal uniformity in the
acquisition system. In a surface grading application it is crucial
to ensure the uniform response of the system through time and space.
Spatial and temporal uniformity is demonstrated and two illuminating
systems (high frequency uniform fluorescents and white LED arrays)
are compared from this point of view.

All the results presented for surface grading were obtained using
real data from the ceramic tile industry. One of the aims of the
thesis has been to build an extensive image database of ceramic tile
models representing the wide range of ceramic tile surface classes.
The VxC TSG database is public and can be accessed at
miron.disca.upv.es/vision/vxctsg/.

Afterwards, we present a study of methodologies developed to obtain
a fast and accurate approach to surface grading. From this study is
extracted a method based on soft colour-texture descriptors computed
in perceptually uniform colour spaces. The method is parameterized
and the involved factors are studied using two statistical procedures;
experimental design and logistic regression. Although
it is not a new theoretical contribution, we have found and demonstrate
that a simple set of global colour and texture statistics, together
with well-known classifiers, are powerful enough to meet stringent
factory requirements for real-time and performance. Two approaches
from literature were also implemented, parameterized and statistically
studied for comparative purposes. These methods are Colour Histograms
and Centile-LBP.

Finally, we explore the method's capacity for on-line inspection in
a study of real-time compliance and parallelization based on cluster
and MPI technologies.