关键词: 铝型材挤压模具, 机器视觉, 平面度测量, 点云处理, 模板匹配

Abstract: The use of the three-coordinate measurement method for aluminum profile extrusion dies has problems such as cumbersome use and the measurement results are easily affected by environmental factors,and the flatness,height and other features cannot be detected by 2D visual measurement mothod. A multi-parameter collaborative measurement method based on machine vision was proposed. The geometric dimensions of the concave cavity and convex model core were quickly extracted through 2D vision,and the 3D point cloud technology was combined to complete the three-dimensional parameters detection such as flatness of the clamping surface and cavity/core depth difference. Template matching and automatic quality judgment functions were realized,and a hardware platform including Bomin SS120 sensor (field of view 200 mm×160 mm) and WinForm packaging software system were built. The experiment adopted two placement orientations. For each orientation the concave die was measured five consecutive times and compared with the actual dimensions. The relative errors in width and heigh were ≤1 %, the absolute errors in angle was ≤3°; the relative errors in depth difference was ≤0.28 %; and the flatness absolute error was ≤0.02 mm; confirming that the system has good measurement accuracy.

Key words: aluminum profile extrusion dies, machine vision, flatness measurement, point cloud processing, template matching