A Vision-Based Algorithm for Deformation of Steel Pipe Column Detection

Abstract

It has great significance for assessing the safety of engineering structures or the bridge by detecting Deformation of steel pipe column. The manual contact measurement is one of the traditional methods to detect the deformation of the steel pipe column. However, in the actual operation environment, the manual contact measurement is difficult and there are some hidden safety problems in the measurement process, thus it is particularly important to study a non-contact measurement method of the steel pipe column. Image processing is one of the common methods of non-contact measurement. In this paper, with the same steel pipe deformation image, respectively, using different edge detection algorithm, comparative analysis of the result, it come to a conclusion that the most suitable algorithm is Canny algorithm, and it is a classic algorithm of edge detection which has been applied in image processing. However, the traditional Canny algorithm has some defects, such as difficulty in threshold setting, low efficiency. Based on traditional Canny algorithm to improve it, this paper proposes an improved custom threshold Canny algorithm. The results show that the method proposed in this paper, the method of which is more than the traditional detection method to save at least 0.8s. Basically meet the requirements of real-time detection of stable and reliable, fast, high accuracy, and has a certain practicality.