@article{13630514f04f483e90ca0a9c4b2ddb6e,
title = "Inspection of 2-D objects using pattern matching method",
abstract = "A pattern matching scheme is developed for the inspection of objects in industrial environment. The inspection includes dimensional verification and shape matching which compares a 2-dimensional image of an object to a pattern image. The method proves to be computationally efficient and accurate for real time application.",
keywords = "Axis of least inertia, Pattern matching, Polygon approximation, Vision inspection",
author = "Han, {Min Hong} and Dongsig Jang and Joseph Foster",
note = "Funding Information: Computer vision has found a wide application in manufacturing environment for parts inspection as well as for parts identification. With the advancement in manufacturing technology, production has become more automated and speedier than ever before. Consequently, unautomated inspection tends to become the bottleneck for the whole manufacturing process. For real time application, therefore, the inspection speed and accuracy are major concerns. Suppose the three types of objects shown in Fig. 1 are inspected for dimensional as well as shape specifications. Our method is to compare the 2-D image of an object to its pattern image. We assume that the 2-D image of the pattern carries sufficient information for the inspection of the object. If no deviation is observed for the dimension and the shape from the pattern image, the object is declared to meet the required specifications. In this paper, the pattern image refers to the image of the pattern whereas the test image refers to the image of an object to be inspected. No two images taken of the same object can hardly be identical due to noise present in the images. Besides, due to roundoff errors introduced in the process of calculating translation and rotation information for the two images, no two images can be superimposed on top of the other exactly. Therefore, pattern matching for dimensional and shape verification is probabilistic in nature. The first step for the pattern matching is to hypothesize that the test image is the same as the pattern image. Then the centroid of each image is calculated and every pixel of the test image is translated to the * This work was partially supported by the Engineering Excellence Fund of Texas A&M University.",
year = "1989",
doi = "10.1016/0031-3203(89)90024-1",
language = "English",
volume = "22",
pages = "567--575",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "5",
}