基于多曝光的高动态焊接图像视觉系统研究

Study on High Dynamic Welding Image Vision System Based on Multi-exposure

为了实现智能化焊接,首先需要解决的问题是能够实时获取焊接过程中的焊接区域图像。现有相机获取的图像都存在曝光不足或曝光过度的问题。为了在弧光、飞溅干扰的前提下获得完整的焊区信息,通过设置相机的不同曝光参数来获取焊接过程中多幅图像,再对多幅图像进行融合。通过基于软件触发的相机曝光参数在线修改的方法,从而实现焊接过程中多种曝光参数的图像实时采集。对于获取到的多曝光图像,先对图像进行伽马变换,再进行平均加权的方法进行图像融合。经过实验证明:本文所用的方法实现了焊接过程中高动态焊接区域图像的获取和融合,为智能化焊接提供理论基础和依据,满足焊接过程中的实时监控作用。

In order to realize intelligent welding, the first problem to be solved is to acquire real-time images of welding area during welding process. The images collected by existing cameras have the problems of under exposure or over exposure.In order to obtain complete information of welding area under the premise of arc light and spatter interference, multiple images are acquired by setting different exposure parameters of the camera during welding process, and then the multiple images are fused. This paper focuses on the method of on-line modification of camera exposure parameters based on software trigger to realize real-time image acquisition of various exposure parameters in welding process. Through the method of on-line modification of camera exposure parameters based on software trigger, the real-time image acquisition of various exposure parameters in welding process was realized. For the acquired multiple exposure images, firstly, the images were transformed by gamma transform, then the average weighted method was used for image fusion. It was proved by experiments that the method used in this paper realizes the acquisition and fusion of high dynamic welding area images in welding process. It provides a theoretical basis for intelligent welding, and satisfies the real-time monitoring role in welding process.