Steam piping infrared image segmentation with trend coefficients algorithm

Abstract

In the industrial production or life of mankind, the use of steam piping network brings convenience and rapidity. As we all know, steam piping are often applied to transport high-temperature materials. Excessive temperature often has potential safety hazards or causes waste of resources. This requires real-time monitoring of abnormal phenomena such as excessive local temperature in steam pipelines. Based on the characteristics of infrared images, steam network pipe images can be easily captured by infrared thermal imager. However, the complexity and diversity of the environment make it difficult for infrared images to directly distinguish the high temperature area and normal temperature area of the pipes. In order to solve this problem, this paper proposes a trend coefficient algorithm for infrared spectroscopy image. Firstly, one-dimensional single-threshold Otsu method is extended to one-dimensional multi-threshold acquisition, and then one-dimensional method is extended to two-dimensional method to form two-dimensional double-threshold Otsu segmentation algorithm. The algorithm includes the trace of between-class scatter matrix as the evaluation function, and analyzes the trend coefficient to obtain the optimal threshold. Through the simulation experiment of MATLAB, it can be seen that the method can clearly get the distribution of high temperature area of pipeline image. It not only extracts the pipeline area from the image, but also accurately segments and locates the over-temperature area on the steam pipeline image. And it also eliminates the interference of trees and shrubs in the outdoor environment to a certain extent. Under the characteristics of different steam pipeline images, the evaluation results confirm that the proposed method can locate and segment the high temperature area of pipeline accurately.