Practical use of receiver operating characteristic analysis to assess the performances of defect detection algorithms
Résumé
Defects detection on images is a current task in quality control and is often integrated in partially or fully automated systems. Assessing the performances of defects detection algorithms is thus of great interest. However, being application and context dependent, it remains a difficult task. This paper describes a methodology to measure the performances of such algorithms on large size images in a semi-automated defect inspection situation. Considering standard problems occuring on real cases, a comparison of typical performance evaluation methods is made. This analysis leads to the construction of a simple and practical ROC-based method. This algorithm extends the pixel-level ROC analysis to an object-based approach by dilating the ground-truth and the set of detected pixels before calculating true positive and false positive rates. These dilations are computed thanks to the a priori knowledge of a human defined ground-truth and gives to true positive and false positive rates more consistent values in the semi-automated inspection context. Moreover, dilation process is designed to be automatically suited to the objects shape in order to be applied on all types of defects.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...