Abstract: X-ray images of complex workpieces often incline to low contrast and unclear feature information by high dynamic X-ray imaging system. Methods based on local variance and partial differential equations were proposed to enhance image contrast, but they failed to effectively suppress noise. To solve this problem, an image enhancement algorithm based on gradient field and local deviation was proposed to improve visual quality. Since gradient information is sensitive to the outlines and edges of an object in an image, this study built a model of contrast adaptive enhancement based on the gradient field to improve the contrast of X-ray images. In addition, local deviation was introduced to quantify the degree of fluctuation in the local information of an image, and then the gain function was constructed by combining the local deviation with the gradient field to enhance the contrast field. Specially, a perfectly flat or reference plane was fitted by the least squares method, and then by calculating the local deviation value between the actual plane and the reference plane, more pixels were taken into account to reduce the effect of noise. Finally, an energy functional was established and solved by variational method to obtain sharper images. In the experiments, the proposed technique was applied to two workpieces for image enhancement and defect detection, and the results show that our algorithm effectively improves the visual quality and contrast of X-ray images, thereby improving the accuracy and reliability of non-destructive testing.

Key words: x-ray imaging, image enhancement, gradient field, local deviation, variational method

摘要： 复杂异形工件的X射线图像由于自身结构复杂性出现对比度低或特征信息不明确的问题，因此基于局部方差和偏微分方程的方法被提出用来增强图像对比度，但它们未能有效地抑制噪声。针对该问题，提出了基于梯度场和局部偏差的图像增强算法来提高视觉质量。首先，由于梯度信息对射线图像中工件的轮廓和边缘处比较敏感，因此基于梯度场构建了对比度自适应增强模型来提高射线图像的对比度；然后，引入局部偏差来量化图像局部信息的平坦度，并将局部偏差与梯度场相结合构建了增益函数从而增强图像的对比度场，特别地，用最小二乘法拟合了一个完全平坦的参考平面，然后通过计算实际平面与参考平面之间的局部偏差值来将更多像素点考虑在内从而降低噪声的影响；最后，建立能量泛函并用变分法求解以获得更清晰的图像。实验部分对两个工件的射线图像进行图像增强和缺陷检测，结果表明所提算法能够有效地提升X射线图像的质量和对比度，从而提高无损检测的准确性和可靠性。

关键词: X射线成像, 图像增强, 梯度场, 局部偏差, 变分法