Classification and Quality Control of Color Pavings by Evolution Strategies

Fairouz Lekhal, mohamed El Hitmy


In this paper, we propose a supervised classification of color texture images. The purpose is to detect and identify the defected images among the initial set of images. The method constructs a discriminate attribute space of color texture images of reduced dimension. This obtains a good quality classification in a well reduced computation time.  This approach is based on the evolution strategies algorithm. It selects an optimal number of attributes which is much less than the initial number considered. We have considered the (R, G, B) color space and the intra and inter components of this space. This technique is applied to the classification and quality control of Moroccan paving decorations and has obtained good results.


Quality control; color texture; supervised classification; evolutionary strategy; attribute vector; selection of attributes

Texte intégral :

PDF (English)