Power Plant Pipeline Defect Detection and Segmentation Based on Otsu's and Region Growing Algorithms

In this study, we consider the complex background and high interference that adversely affect infrared images of high-temperature pipelines in power plants and the requirements of image processing algorithms for inspection robot systems. We propose a high-temperature pipeline defect detection and extraction method based on an improved two-dimensional Otsu and region growth algorithms. After grayscale conversion, a 2D Otsu method was used to extract the pipeline area. Based on the grayscale histogram of the pipeline region and the average gray value of the neighborhood, automatic detection and positioning of multiple sub-points were realized. The segmentation of the defect area was accomplished using two methods. The adaptive threshold was determined based on the gray mean and standard deviation values of the growth area, while the growth criterion was improved using the gradient amplitude of the Prewitt operator. The experimental results show that the proposed algorithm can not only realize the automatic detection and positioning of various defects in high-temperature pipelines of power plants, but it additionally segments the defect regions more accurately with high accuracy and good real-time performance.