Effective methods for Brinell Hardness Measure based on vision

Abstract

In the Brinell Hardness Measure System based on vision, the measuring accuracy is greatly influenced by the huge interference, strong noises and non-smooth indentation edge in the indentation image amplified by object glass because of too many scratches on the surface and too small indentation of the metal block to be measured. In order to calibrate the practical pixel length, a standard gauge block and the Least Squares Method are adopted to fit the line of the block's edge. In the process of extracting the indentation diameter, the method based on region growing is utilized to segment the image and then, the circle detection method based on Existence Probability is applied to detect the indentation edge. The experiment testifies that these methods, which have better real-time performance and robustness, can better recover indentation edge and acquire indentation diameter.