Because X-ray images of weld contain uncertain noise and also the defects inside them have low contrast to their background, it is difficult to detect weld defects of X-ray images. The goal of this paper is to locate and segment the line defects in X-ray images. Firstly, we present an approach to extract features of X-ray images with multiple thresholds. Then, use the support vector machine (SVM) technique to classify the defect and non-defect features to obtain a coarse defect region. Furthermore, perform the Hough transform to remove the noisy pixels in the coarse defect region. Then the defect is located and segmented. The experimental results show that the proposed approach is effective and feasible to segment and locate defects in noisy and low contrasted X-ray images of weld.

Detection of line weld defects based on multiple thresholds and support vector machine

EMD-based pulsed TIG welding process porosity defect detection and defect diagnosis using GA-SVM

Real-time defect detection on highly reflective curved surfaces

A method based on Kalman filtering is proposed for weld defect detection in real-time radiographic NDT of spiral pipes. The existence of the image noises and the inhomogeneity of the background contrast induce numerous false alarms. In this paper, the trajectory continuity of the defects in the image sequence is detected by Kalman filtering for the identification of true defects. Potential defect regions without continuous motion are considered as false alarms and are eliminated. Experiments are performed to demonstrate the adaptability of the proposed method. The robustness of the method is also verified under unstable detection velocity.

Automatic weld defect detection method based on Kalman filtering for real-time radiographic inspection of spiral pipe

This paper is about the detection of tire defects in multi-textural radiographic images. We consider the tire defects characterization problem in ways of local regularity analysis and scale characteristic. Optimal scale and threshold parameters are selected using a defect edge measurement model to frame defect edge detection. This framework distinguishes the defects from the background textures. Finally, a novel method for detection of tire defects is proposed based on wavelet multiscale analysis. We provide examples with a consistent dataset of 400 images selected over 3700 industrial images in order to illustrate and validate the obtained results which demonstrate substantial improvement over the state of the art.

Automatic Detection of Defects in Tire Radiographic Images

In order to detect weld defects in steel tube, a real-time imaging and detecting system is setup. In the extracted weld seam, based on spatial characteristics near defects—variance and contrast, defects such as slags, blowholes and incomplete penetration are automatically detected using the method of fuzzy pattern recognition, and the system will automatically alarm if the defect exceeds the national standard. Compared with other methods, it is simple and fast, and has fewer misinterpretations. It can detect weld defects in real-time.

Peng Bai

Papers

Real-time automatic detection of weld defects in steel pipe