A Comprehensive Method for Measuring the Dimensions of Rolling Wire Rod Cut-Off Materials Based on Machine Vision
This paper proposes a machine vision-based method for measuring the shape and size of scrap material resulting from the shearing of control rod wires.The aim is to improve the level of automation in the rod wire rolling process.Additionally,a corresponding image acquisition device has been designed.Based on image merging and the UNet neural network,criteria for determining the deformed and normal sections in the sheared waste material have been proposed,and measurement results for the length of the de-formed sections were presented.The measurement method proposed in this paper exhibits robustness to vari-ations in the detection environment and target shapes.In addition,it requires a significantly lower sample size compared to other traditional neural networks,making it more suitable for practical industrial datasets.The measurement range in the length direction can also be flexibly adjusted with the increase or decrease of the number of images.Experimental results demonstrate that the proposed measurement method is simple,effective,and highly accurate.Furthermore,due to the lightweight and pre-trainable nature of the model,this method has the potential for online monitoring deployment.
rolling wire rod cut-off materialsmachine visiondimension measurementUNet neural network