Machine vision-based disease pipeline defect classification library building and identification method

Abstract

The invention provides a machine vision-based disease pipeline defect classification library building and identification method. The method is characterized by comprising the steps of classification library building and identification. The classification library building method comprises the steps of image acquisition, labeling, preprocessing and model training in sequence, and the identificationmethod comprises the steps of video acquisition, image processing, classification positioning and report generation in sequence. The method can replace manual classification and identification limitedby technical level and energy, greatly improves the identification speed, precision and length of pipeline disease defects, reduces misjudgment, can achieve online and offline identification, is convenient to use, and is good in effect.

Description

technical field [0001] The invention relates to a method for classifying and identifying pipeline diseases, in particular to a method for classifying and identifying diseased pipeline defects based on machine vision. Background technique [0002] With the development of cities, the construction of my country's underground pipe network has become more and more complex and diverse, and the scale of the pipe network has been expanding. By the end of 2016, the total length of my country's urban drainage pipes and channels had reached 580,000 kilometers, and it was growing at a rate of 7% per year. . Before the 1980s, the length of drainage pipelines in my country was 35,900 kilometers, accounting for 6.22% of the existing pipelines. Due to the long history of laying underground pipe networks in some old urban areas in my country, there is an intricate situation with the later laid pipe networks. At the same time, due to various reasons such as the pipeline laying process, the d...

AI Technical Summary

Problems solved by technology

At present, domestic and foreign drainage pipeline detection mainly collects pipeline information through CCTV, sonar, laser and other methods, and then transmits the image back to the ground through the transmission line. Manual review by engineers is also required, highly professional

With the increasing demand for pipeline inspection, the speed of manual identification can no longer meet the speed requirements, and the defects such as cracks are extremely small compared to the background information. It is very easy to miss detection by naked eye observation, and high-intensity labor is also likely to cause staff fatigue and missed inspection. Increased chance of detection

In addition, manual recognition depends on the experience of engineers, and the judgment results are highly subjective. Different engineers may have different judgment results, which affects the consistency of judgment structures and relatively increases the probability of misjudgment.

Images