Fine localization algorithm of automatic cuttingmachines based on machine vision
                                                                       LI Guangming, ZHAO Liangliang
                      (College of Electrical and Information Engineering, Shaanxi University ofScience & Technology, Shannxi 710021, China)

Abstract: Aiming at the problem that the requirements for highprecision products could not be satisfied due to the low cutting precision of traditional automatic cutting machines, a method of using machine vision technology to improve the cutting precision of the cutting machine was proposed by studying the structure of the cutting machine. Firstly, the improved morphological gradient filter operator was used to find the rough edge of the sheet, and then the subpixel fine positioning was performed by the gray scale method. Finally, the subpixel edge points of meeting the requirements were fitted into a straight line by least square method. Then the offset of the sheet was calculated according to the geometric relationship, and was used to guide the correction platform to compensate the deviation. The method was simulated and tested by image edge positioning experiment designed by Matlab. The results indicate that the cutting precision of the traditional automatic cutting machine can reach 0.03mm theoretically, which meets the precision requirements of most highprecision products.

Key words: cutting machine; morphological gradient; gray scale method; subpixel fine positioning; correction