核电焊缝DR图像多尺度对比度增强方法

A Multi-Scale Contrast Enhancement Method for DR Image of Nuclear Power Weld

射线DR技术是核电焊缝最新射线检测手段。针对核电焊缝部分DR图像对比度差,细节容易漏检的问题,本研究采用拉普拉斯金字塔进行16位DR图像多尺度分解,结合高斯平滑滤波与减影算法提取图像细节信息,运用s型函数进行细节映射增强,再将细节增强结果加到近似图像上,重构出增强后的图像,最后实现了一种16位DR图像多尺度对比度增强算法。对含线型缺陷或圆形缺陷的DR图像进行了多尺度对比度增强实验,参数L=10、p=0.8时,算法能有效增强细节对比度,有助于提高缺陷识别能力。算法具有很好的工程应用前景。

Digital radiography(DR)is the latest X-ray inspection method in nuclear power welds.Aiming at the problems of poor contrast of nuclear power welding seam DR images and easy missing of details,a multi-scale contrast enhancement algorithm for 16 bit DR image is proposed.First,laplacian pyramid is employed for decomposing 16 bit DR image,and then gaussian smoothing filter and subtraction algorithm are used to extract the image detail,and a s-type mapping function is used to enhance detail,and finally the enhanced detail is added to approximate image for DR image reconstruction.The multi-scale contrast enhancement experiments are performed on DR images containing linear defects or circular defects.When the parameters L=10 and p=0.8,the developed algorithm can effectively enhance the detail contrast and improve the ability of detect recognition.The developed algorithm has good prospects of engineering application.