Анализ методов выделения контуров объектов при обработке изображений печатного узла в системах контроля качества автоматической оптической инспекции

Качество. Инновации. Образование. 2015. № 9. С. 17–24.

Veselov E. A., Artyukhova M. A., Linetskiy B.

This article describes a study of the basic tasks of image processing in the field of quality control of printed circuit assemblies during the production of electronic equipment in the automatic optical inspection systems. In particular, a system which is based on the principle of defect detection using comparative analysis of the three-dimensional model of a real circuit board and the reference model is considered  The method of generating a 3D reconstruction of the inspected printed circuit board uses inverse perspective transformation. The recovery of the points coordinates on the surface of printed circuit board components is performed according to the contours coordinates of their projections (shadows), resulting from the illumination by a point light source and video recording one camera. Accordingly, when the image processing sections of the printed circuit board is necessary to highlight the contour of objects and their projections. The main challenge of this task is to ensure a high probability of the accurate contour detection under the influence of noise. The study analyzes the main methods of contour detection, the experiment on segmentation of two images of the simulated and real areas of the printed circuit board, the data identified the most optimal method of contour detection.  

Keywords: изображение, объект, object, Цифровая обработка информации, three-dimensional model, трехмерная модель, печатный узел, contour, image, дефект, digital processing, detector, gradient, perspective projection, print node, defect, automatic optical inspection, контур, детектор, градиент, перспективная проекция, автоматическая оптическая инспекция