Automated keypad inspection by image registration using hough transform based perimeter extraction and multilevel pyramids

摘要

Manufacturing and Quality inspection using Machine Vision is diversely increasing in various processes and in various industries. However there are various aspects which need to be considered in order to provide the best quality output like illumination conditions, alignment of objects, efficiency of the machine, time required to detect the flaws etc. There exists a need of a robust and autonomous system which will inspect the keypad labels for quality and discard the defective samples. The sample (keypad label in this case) is taken by the user and given to a machine which automatically inspects and detects the flaw present in the sample. To identify and classify a faulty area of an image, image subtraction is carried out which is possible only when the two objects are spatially aligned. Hence the process of Image Registration (IR) is demanded. The ultimate goal of this paper is to register the multi temporal images in order to minimize the misalignment and compute a high correlation factor by Hough Transform based Perimeter Extraction (HTPE) and Image Registration using Multilevel Pyramids (IRMP). The processes of defect inspection are computationally simple, fast and yield 90% of result accuracy.