Research On Seam Tracking Based On Particle Filter With Histogram Of Oriented Gradient

Automatic detection and tracking technology of welding seam is the major research subject to realize the automatic and intelligent welding process.Therefore,automatic tracking technology of welding seam has been a hot topic at home and abroad.However,due to the complexity of the actual welding environment,it is still a major challenge to realize the real-time,accurate detection and tracking of the welding seam.In this thesis,the welding seam tracking in the automatic welding process is susceptible to the influence of arc,spatter and other noise interference,which results in the decrease of tracking accuracy.Combining the directional characteristics of the laser image of the seam,a seam tracking algorithm based on the particle filter of directional gradient histogram(Hog)is proposed.The tracking accuracy and anti-interference ability of the weld seam can be promoted by the Hog feature which can describe the directionality of the weld stripe image well.The main research contents of this thesis are as follows:(1)According to the directional characteristics of the laser stripe at the inflection point of the weld seam image,the good geometric and optical deformation invariance of the directional gradient histogram is deeply investigated,which can well describe the directional characteristics of the weld seam stripe image and can suppress the interference of arc light.(2)Using the continuous predictability of the weld sequence image,the principle of particle filtering and the target feature of the direction gradient histogram are used to predict the weld state,and correct the weld position.The error curve is relatively stable,and the tracking error is small,which can meet the accuracy requirements of welding seam tracking.(3)Weld tracking algorithm based on directional gradient histogram particle filtering proposed in this thesis is applied to typical V-shaped weld tracking.Compared with the particle filter tracking method based on the image gray information and the algorithm based on KCF,the result showed that the automatic detection and tracking method proposed in this thesis can accurately detect the weld seam features and the weld position in the complex environment.The algorithm based on the feature of histogram of oriented gradient can effectively avoid the influence of noise interference in the welding process and effectively improve the tracking accuracy to 0.240mm,which showed much better accuracy and robustness.It provides a reliable and effective seam tracking algorithm in automatic welding field.