Research on Flexible Material Identification and Positioning Based on Machine Vision

The cutting of flexible materials (such as garment fabrics, leather, and packaging) is a preliminary process in product manufacturing. The accuracy of cutting determines the product production quality. The core challenge is the recognition and positioning accuracy of the marker information. Based on the positioning information, the cutter head position is initialized, establishing the correspondence between the physical coordinates of the robot and the pixel coordinates of the data image. Traditional shape-based template matching algorithms are prone to local optima and overall shifts when dealing with size scaling and deformation of flexible materials, resulting in decreased positioning accuracy. This study proposes an optimization positioning algorithm based on a standard deviation assessment to reduce the overall shift in shape matching, further improving the positioning accuracy on the original basis. Without the need to establish a scaling matrix for the template, marker information can still be located on the scaled and deformed fabrics, significantly reducing the algorithm complexity. Finally, this machine vision system is applied to equipment at a CNC cutting device manufacturing enterprise in Ningbo. Using a resolution of $640\times 480$ pixels with a black-and-white industrial camera and a lens at a distance of 9 cm from the material, the positioning accuracy was controlled within three pixels. The accuracy of the equipment was close to that of its well-known domestic and international counterparts.