Robot Welding Seam Tracking System Research Basing On Image Identify

摘要

With the continuous development of Computer Vision and a variety of advanced seam imaging equipment, theinformation contained in the seam image is very rich. It is of great significance for industry automation system. Singleimage feature is difficult to fully express seam image content. Multi- feature fusion has become a natural way to extractthe seam image features. It can comprehensively utilize the seam image information to gain more rapid and accurateunderstanding of welding images.From low to high, information fusion can be divided into three levels. The feature-level fusion not only keeps the mostoriginal information, but also overcomes the unstable and large characteristics of original data. Fusion feature can beeffectively used in seam image recognition.Firstly, we build the JARI robot system to research the seam tracking from the image identify. Secondly, principalcomponent analysis (PCA) method based on multivariate statistical analysis is used in feature- level fusion. And it isapplied in liver B- image recognition. The recognition results are analyzed and compared. Finally, through the gantryrobot 9 degree system to verify the logic of the identify V type seam.The experimental results show that fusion feature can fully and effectively express seam image, which can bring betterrecognition results. Analyzing and comparing the feature selection results of different sample images, the results showthat feature selection is stable and effective. Comparing with the results of direct PCA fusion applications, the recognitioneffect after feature selection is better, not only improves the the average accuracy rate of recognition but also reduces thetime complexity of the recognition process. It has better performance, can be more effectively applicated in weldingimage recognition.