A study on the modified algorithm for image processing in Tracking Seam Welding

For robot path planning the weld seam positions need to be known in advance as the industrial robot generally work in teach and playback mode. Since the welding of the pipe is not done completely on the straight line (the nature of the pipe) and the test tube under the machine is moving, the symmetry of the two probes in relation to the welding site is very important during the test and quick tracking is required to set the probes. The use of image processing and machine vision techniques is very efficient in optimizing seamless welding radia. In designing the algorithms used, an attempt has been made to reduce the environmental conditions and unstable industrial situation well in order to track the weld seam with an acceptable speed. New approach has been used to access the central line of the weld seam area. The algorithm is designed to be implemented in a real environment and has very good results. One of the advantages of this method is the reduction of measurement error and the elimination of mechanical and electrical sensors in non-destructive tests.

Keywords:

Morphological filter , Welding robot , Non-destructive tests , Feature Extraction , ultrasonic , image processing