Corner detection of a quadrilateral for strain analysis in sheet metal forming by image processing

Abstract

ABSTRACT Circle Grid Analysis (CGA) and Square Grid Analysis (SGA) are commonly used methods to measure the strain in sheet metal forming operations. In these methods, small diameter circles or squares are printed on a flat sheet and deformed to the required shape. The strain at any location of the formed part is evaluated by measuring the deformed circle (ellipse) or square (quadrilateral). Herein, an attempt has been made to use an image processing technique to detect the corners of a quadrilateral for automatic strain measurement in SGA. For this, different sizes of quadrilaterals are printed on a flat sheet, and images are captured using a USB microscope. These images have dense noise due to the capturing method as well as inherent surface defects on the sheet metal. Different noise reduction algorithms such as Hough Transform (HT), Gaussian filter of varying sizes, and Biorthogonal wavelets have been used to assess their efficiency in noise reduction. Among these, the HT method seems to be efficient in noise reduction. Further, the images are processed with the Harris corner detection method to detect quadrilateral corners. Finally, projective transformations have been used to fit the maximum size ellipse in the deformed quadrilateral for strain measurement.