| Summary: | 碩士 === 南臺科技大學 === 資訊工程系 === 107 === We propose an image-based die-cutting calibration system for die-cut punching machines. Nowadays, automatic die-cut punching machines are often equipped with an object packing software module for placing objects densely in a predefined area of material, and can cut much more objects than traditional manual punching machines for saving material usages. When cutting densely-packed objects by automatic punching, the inaccurate alignment of a physical die to both (i) a rotational punch center and (ii) a virtual die template in object packing, simultaneously, is a main problem that results in overlapped object cutting. In this thesis, we thoroughly investigate this problem, which has rarely been explored in previous literature. Specifically, two system modules of offline and online die calibration have been developed in this study. The offline die calibration module provides a new virtual-physical interaction interface that projects a virtual die model to a base holder by a projector for operators to visualize the die fixation position and then to mount a die on its holder accurately. The online calibration module, on the other hand, measures a mechanical installation deviation between the fixed die center and a rotational punch center in the punching machine by matching a virtual die template to the image of a cut object captured by a camera for further deviation compensation of installed die during object punching. Experiments conducted for the proposed calibration approaches include projector calibration measurements, camera calibration measurements, punch head positioning accuracy measurements, and punching results using the offline and online calibrations. The experimental results show that offline calibration is more effective in alleviating the problem of overlapped object cutting for the reduction of cutting material wastes.
|
|---|
