Multi-scale multi-intensity defect detection in ray image of weld bead

焊道射线图像中的多尺度多强度缺陷检测

Abstract

The radiographic test used to inspect the internal defects is an important non-destructive testing technique for welding.The weld bead radiographs of different detection scenes and different types of defects are very different, which limits the versatility of the automatic detection algorithm.

用于检查内部缺陷的射线探伤是一种重要的焊接无损检测技术。不同检测场景、不同类型缺陷的焊道射线照片差异很大，限制了自动检测算法的通用性。

This paper solves this problem by analyzing the radiographic images at different scales and intensities. Firstly, a multi-scale multi-intensity parameter space is established, and the preprocessed images corresponding to the parameters ensure that defects are not missed.

本文通过分析不同尺度和强度的射线照相图像解决了这个问题。首先，建立多尺度多强度参数空间，参数对应的预处理图像保证不漏检。

Then, through the detection standard of the weld and the properties of the radiographic image, the value range of the parameters is automatically limited and the preprocessed image is optimized.

然后，通过焊缝的检测标准和射线图像的特性，自动限定参数的取值范围，优化预处理图像。

Finally, algorithms for screening and merging defects in different preprocessed images are designed to reduce false detections and describe precise defect boundaries.

最后，用于筛选和合并不同预处理图像中的缺陷的算法旨在减少错误检测并描述精确的缺陷边界。

1. Introduction

The radiographic test (RT) used to inspect the internal defects of the weld is the critical non-destructive testing (NDT) technique for welding.Therefore, the weld area is often extracted as the re- gion of interest to avoid the interference of the weld edge and improve the detection efficiency [8–9].

用于检查焊缝内部缺陷的射线照相检测 (RT) 是焊接的关键无损检测 (NDT) 技术。因此，焊缝区域通常被提取为感兴趣的区域，以避免焊缝边缘的干扰，提高检测效率[8-9]。

Different welding defects often show different visual characteristics in shape, size, texture, contrast and position [12], so different detection algorithms are usually applied according to different types of defects.

不同的焊接缺陷往往在形状、尺寸、纹理、对比度和位置上表现出不同的视觉特征[12]，因此通常根据不同类型的缺陷应用不同的检测算法。

Under different welding seam inspection conditions, the visual dif- ference between radiographic images is greater, so different algorithms are usually designed according to different inspection or welding con- ditions.

在不同的焊缝检测条件下，射线照相图像之间的视觉差异较大，因此通常根据不同的检测或焊接条件设计不同的算法。

Therefore, this paper analyzes the weld seam radiographs at different scales and intensities to ensure this detection algorithm can be applied to different types of defects in different detection scenarios.

因此，本文对不同尺度和强度的焊缝射线照片进行分析，以确保该检测算法能够适用于不同检测场景下的不同类型缺陷。

Compared with traditional methods [17–19], deep learning methods show some advantages, such as auto- matic feature extraction [20], strong generalization ability and so on.However, it also has some disadvantages: deep learning methods usually require a lot of class-balanced defect samples [21–22], especially labeled data, which is difficult to collect (rare defects) [23].

与传统方法[17-19]相比，深度学习方法具有自动特征提取[20]、泛化能力强等优势。然而，它也有一些缺点：深度学习方法通常需要大量的类平衡缺陷样本[21-22]，尤其是标记数据，难以收集（罕见缺陷）[23]。

6. Conclusion

This paper proposes a multi-scale and multi-intensity defect detec- tion algorithm. Firstly, in order to adapt to different weld bead detection scenarios and various types of defects, the algorithm analyzes the defect features at different scales and intensities in the parameter space.

提出了一种多尺度、多强度的缺陷检测算法。首先，为了适应不同的焊缝检测场景和不同类型的缺陷，该算法在参数空间分析了不同尺度和强度的缺陷特征。

And then it automatically selects the appropriate scale and intensity based on the RT standard of the weld and the attributes of the radiographic image.

然后根据焊缝的RT标准和射线图像的属性自动选择合适的尺度和强度。

Finally, this algorithm eliminates false alarms and corrects the detection results based on the scale and intensity features.

最后，该算法消除虚警，并根据尺度和强度特征对检测结果进行校正。