Image Analysis and Information Fusion Based Defect Detection in Particleboards

Abstract

This paper is concerned with the problem of image analysis based detection of local defects embedded in particleboard surfaces. To increase the defect detection reliability, results obtained from both the global and local image analysis are combined. The Global analysis exploits the discrete probability distribution (the histogram) of the standard deviation of image intensity values. The local analysis is accomplished by thresholding the “standard deviation image” and inspecting the obtained result. The focus of the work is on defect detection in textured images, rather than classification of defects into various types. A 100% correct classification accuracy was obtained when testing the technique proposed on a small set of images. Ill. 5, bibl. 17 (in English; summaries in English, Russian and Lithuanian).