Discontinuity detection on industrial parts: Real time image segmentation using Parzen's kernel

This paper describes a real time system for visual testing of textured industrial parts - cathode ray tubes. Four types of discontinuities are detected: smooth surfaces, bumps, missing material and hollow knocked surfaces. In order to distinguish zones with discontinuities and acceptable zones with minimum error, both optimal lighting parameters and most discriminative features are selected. The feature and lighting selection process is based on a probability density estimation method using a Parzen's kernel based technique. This same method is used in a multidimensional way to perform the pixel wise segmentation step, based on the Bayes rule. The nature of the application requires a real time response. The segmentation is therefore implemented as a hyperrectangle based method of feature space division. A comparison of the two classifiers' behavior is presented. The final cathode classification step is performed using the segmented image combined with a circularity testing of the part. The proposed approach achieved the quality required. The system runs in real time on a standard, midrange personal computer. Since the proposed approach does not take into account any particular features of the tested part (for example, its shape), it can be used in other visual testing tasks of textured parts.

[1]  Frederic Truchetet,et al.  Analysis of compatibility between lighting devices and descriptive features using Parzen's Kernel: application to flaw inspection by artifical vision , 2000 .

[2]  Patrick Gorria,et al.  Classification géométrique par polytopes de contraintes , 1991 .

[3]  Luc Van Gool,et al.  Automated Visual Inspection , 1990 .

[4]  Pierre Geveaux.  Conception d'un système de détection de défauts sur surfaces texturées par traitement d'images : application au contrôle haute cadence de pièces en défilement , 2000 .

[5]  Heekuck Oh,et al.  Neural Networks for Pattern Recognition , 1993, Adv. Comput..

[6]  B. Lamalle,et al.  Simulations of lighting for an optimal defect detection by artificial vision , 1997, Proceedings of IEEE/ASME International Conference on Advanced Intelligent Mechatronics.

[7]  Vladimir Vapnik,et al.  Statistical learning theory , 1998 .

[8]  Richard O. Duda,et al.  Pattern classification and scene analysis , 1974, A Wiley-Interscience publication.

[9]  M. Unser.  Local linear transforms for texture measurements , 1986 .

[10]  Fabrice Meriaudeau,et al.  Machine vision for the control of reflecting nonplane surfaces , 1997, Other Conferences.

[11]  Mark Michael,et al.  Experimental Study of Information Measure and Inter-Intra Class Distance Ratios on Feature Selection and Orderings , 1973, IEEE Trans. Syst. Man Cybern..

[12]  P. Bolon,et al.  Analyse d'images: filtrage et segmentation , 1995 .

[13]  Thomas G. Dietterich,et al.  An Experimental Comparison of the Nearest-Neighbor and Nearest-Hyperrectangle Algorithms , 1995, Machine Learning.

[14]  Miteran,et al.  4 - Classification géométrique par polytopes de contraintes. Performances et intégration , 1994 .

[15]  P. Gorria,et al.  Real time defect detection using image segmentation , 1997, ISIE '97 Proceeding of the IEEE International Symposium on Industrial Electronics.