Feature extraction of manufactured parts in the presence of spurious surface reflections

Abstract

In the application of image processing techniques for feature selection in manufactured parts, the generation of spurious features due to strong reflections from illuminated surfaces (particularly metal parts) has presented problems in the proper analysis of features for quality control purposes. Such spurious features most frequently appear as bright regions on a dark background and when image thresholding is performed for the purpose of feature detection, such regions appear as genuine features of the surface, causing ambiguities in the feature detection. This paper proposes a simple and effective technique for eliminating such features through an efficient border-following technique followed by linear scanning f the interiors. The effectiveness and the usefulness of the technique is demonstrated by considering a piston (in particular the surface of the piston head) of the internal combustion engine of an automobile.