针对火箭发动机喷管钎焊缝射线检测图像灰度分布不均匀、结构与缺陷信息混叠导致缺陷检测精度低的问题，提出一种基于边缘信息和结构特征提取的图像处理和缺陷分析方法。首先通过亮度矩阵、高斯滤波与锐化等方法对原始检测图像进行增强预处理；然后利用Canny算子和霍夫变换抽取图像中的边缘信息及边缘中的线段结构信息，并通过自适应阈值处理获取类正弦结构信息；最后对所得边缘信息、线段信息和类正弦结构信息进行差分处理并进行缺陷信息判定。基于所提算法开发了相应软件，对实际发动机喷管钎焊缝射线检测图像进行分析，处理结果表明，所提方法可有效克服区域灰度不均匀、结构与缺陷信息混叠等问题，实现了射线检测图像的统一处理和缺陷信息的提取，具有较高的检测效率及准确性。

A method for image processing and defect analysis based on edge information and structural feature extraction was proposed to address the problem of low defect detection accuracy caused by uneven grayscale distribution and overlapping structural and defect information in rocket engine nozzle brazing X-ray detection images. Firstly, quality enhancement preprocessing was performed on the original detection image through methods such as brightness matrix, Gaussian filtering, and sharpening; Then, the Canny operator and Hough transform were used to extract edge information and line segment structure information from the image, and the sine like structure information was obtained through adaptive threshold processing; Finally, differential processing was performed on the obtained edge information, line segment information, and sinusoidal like structure information, and defect information was determined. Based on the proposed algorithm, corresponding software was developed to analyze the actual radiographic inspection images of the engine nozzle brazing seam. The processing results showed that the proposed method can effectively overcome the effects of uneven regional grayscale, mixed information of structures and defects, and achieve unified processing of radiographic inspection images and extraction of defect information, with high detection efficiency and accuracy.