Edge Detection Of CO₂ Welding Metal Transfer Image Based On Wavelet Transform

For the high quality seam, applicability and high efficiency, CO₂ welding is widely used as an important welding method in the modern industry. The analysis of droplet image helps to prove the theory in the metal transfer, and provides a more direct and exact way in observing the welding process, it can assistant the dynamic control of welding machine. In this thesis, wavelet transform is used to detect the edge of the droplet image in CO₂ welding.The image of droplet is gotten by the originally designed micron high-speed photography. This system is set under the theory of back light photography. It is composed by back light, UV lens (Ultra Violet), ND lens, interference filter, micron shot, C-Mount. It can filter out the major arc and get the relatively clear images of CO₂ welding.According to the character of images, the image extraction, medium filtering and image hence are performed to decrease the process time, improve images and outstand the region of interest. In the process of image hence, the whole period of short transfer is originally divided into three moments in order to perform with different parameters.The edge detection based on the wavelet multilevel transform is used on the droplet. Considered as the edge criterion, the wavelet transform local maximum parameter adds the argument direction as the comparison direction to raise the edge accuracy. The result shows that the improved way can get far better edge effect than classical edge detection operators. Furthermore, the cubic spline interpolation and polynomial fitting are used to restore the image with the incomplete edge.The clear and complete edge is the premise to show the change of...