基于图像处理的轮船吃水线计重系统的研究与实现

Research and Implementation of Ship Waterline Weighing System Based on Image Processing

观测吸砂船吃水线是吸砂船计重中的重要一环,在人工观测中,吃水线计重受人的主观因素影响较大,涉及多方利益。基于此,利用图像阈值分割和霍夫直线检测确定吃水线在图像中的位置,基于R-CNN前馈神经网络,并利用PyTorch框架的深度学习算法,识别水尺字符,最终确定轮船的吃水深度,结合轮船的其他参数,进而确定轮船的载重量。偏差率仅在0.3%左右,有利于保障轮船的各方利益。

Observing the waterline of the ship is an important part of the weight measurement of the ship,in manual observation,the weight of the waterline is greatly affected by human subjective factors and involves multiple interests.This paper uses image threshold segmentation and Hough line detection to determine the position of the waterline in the image,and proposes a deep learning algorithm based on the R-CNN feedforward neural network and using the PyTorch framework to recognize water gauge characters and finally determine the draught of the ship,Combined with other parameters of the ship,and then determine the load capacity of the ship.The deviation rate is only about 0.3%,which is conducive to protecting the interests of all parties in the ship.