Development of vision-based system for measurement of bend angle of formed components

This thesis presents the development of a low-cost, portable, vision-based system designed for the precise measurement of bending angles in sheet metal components. The system aims to accurately measure bend angles of materials such as galvanized sheet steel, DKP (Deep Drawing Quality) black sheet steel, and MDF wood under industrial conditions. In the sheet metal processing industry, measuring bent parts is typically time-consuming and requires specialized hardware and expert personnel. The lack of standardized measurement procedures and operator dependency lead to inconsistencies between different devices or users, consequently increasing quality control costs. To address these issues, a vision acquisition setup comprising two line lasers (one vertical and one horizontal) and a webcam was designed. The vertical laser facilitates the operator's easy identification of the bend line, while the horizontal laser is used for the precise calculation of the bend angle through image processing algorithms. The measurement system was designed based on the principle of laser triangulation, and its 3D solid model was created using reverse engineering. The laser line projected onto the bent part was segmented from the background to calculate the angle value of the workpiece. For camera calibration, the checkerboard pattern method proposed by Zhang was employed. Additionally, an exponential calibration model was developed to account for the variation of light intensity with distance. The developed system was tested on three different sample types: galvanized sheet metal, DKP black sheet metal, and MDF wood, each exhibiting bending angles ranging from approximately 45° to 135°. Additionally, the performance of lasers in three distinct colors—red, blue, and green—was compared across these three sample types. Camera parameters, including exposure, gain, and saturation values, were optimized according to the color of the laser used. The experimental results demonstrated that the proposed system achieves a measurement accuracy of ±1° and meets the industrial standards applicable to the coarse assembly industry.