基于图像处理的石油管道故障检测

Pipeline Fault Detection Based on Image Threshold Segmentation

石油管道的内壁检测维护是管道长期使用的重要保障,直接影响运输安全以及经济效益。在管道的实际使用过程中,异常磨损是其影响日常使用安全的主要原因。针对现有的机器视觉检测系统的效果和准确性低、人工检测劳动强度高和错检漏检率高等问题,提出一种基于改进粒子群优化(Particle Swarm Optimization,PSO)Otsu的图像分割技术。最后进行多阈值对比试验,实验结果表明经过粒子群优化Otsu分割后的图像增强了原图像的边缘部分,并保留了图像中较为平坦的部分,更能突显出管道的缺陷特点。

Inspection and maintenance of the inner wall of oil pipelines is an important guarantee for the long-term use of pipelines,directly affecting transport safety and economic efficiency.In the actual use of the pipeline,abnormal wear and tear is the main reason for the safety of its daily use.To address the problems of low effectiveness and accuracy of existing machine vision inspection systems,high labour intensity of manual inspection and high rate of error and omission,an image segmentation technique based on improved Particle Swarm Optimization(PSO)Otsu is proposed.Finally,a multi-threshold comparison test is conducted,and the experimental results show that the image segmented by PSO Otsu enhances the edge part of the original image and retains the flatter part of the image,which can highlight the defect characteristics of the pipe.