论文部分内容阅读
焊缝图像的边缘提取精度对实现焊缝视觉跟踪至关重要。根据对焊接过程中采集的图像进行特征分析,新颖的把压缩技术中的梯度边缘检测(GED)预测器模板应用到焊缝图像边缘提取中,首先对焊缝图像进行滤波、二值化,生成预处理图像,然后对预处理焊缝图像应用GED预测器模板生成预测误差图像,再通过阈值分类误差图像边缘,细化边缘。实验证明,该方法检测的焊缝图像伪边缘少、定位准确、边缘清晰,达到较好的效果。
The edge extraction accuracy of the weld image is crucial for the visual tracking of the weld. Based on the feature analysis of the images collected during the welding process, the novel gradient edge detection (GED) predictor template in the compression technique is applied to the edge extraction of the weld image. Firstly, the weld image is filtered and binarized to generate The image was preprocessed, and then the GED predictor template was used to generate the prediction error image for the preprocessed weld image. Then the edge of the error image was classified by the threshold to refine the edge. Experiments show that this method has fewer false edges, accurate positioning and clear edges, and achieves better results.