利用红外图像处理方法检测材料缺陷

Defect test of material using infrared image processing methods

由于物体的导热特性和红外成像系统的限制,系统所成红外图像中目标与背景的对比度普遍较低,目标边缘模糊,难于直接判读缺陷,对于异形构件更是如此。针对这一问题,在热传导原理的基础上,首先通过中值滤波、帧间滤波尽量减少噪声对红外图像的影响。接着运用受限拉普拉斯算法和高提升滤波加强图像轮廓信息。然后通过灰度分段线性变换对图像中的缺陷信息进行增强,突出缺陷,改善红外图像的视觉效果,使其中的缺陷更适合于人眼观察,最后通过边缘检测的方法将缺陷边缘提取出来。仿真实验结果表明:该检测方法可得到清晰、光滑的缺陷轮廓图像,是一种实用型红外缺陷检测方法。

For the heat conduction properties of object and the restriction of infrared imaging systems,target in the infrared image has low contrast generally,its edge blur compared with background and the defects are difficult to be read directly,especially for heteromorphy components.In order to solve this problem,on the basis of the principle of heat conduction,first of all,the median and interframe filtering were used to minimize noise of infrared images.Second,the limited Laplace algorithm and high profile information filtering were used to strengthen image outline.Then,the highlight flaws were strengthened by gray-scale piecewise linear transform and the visual effect of the defects was improved,which were more suitable for human observation.Finally,the defect edge was obtained through edge extraction.Simulation experimental results show that the proposed method can be used to get clear,smooth contour image of the defect,and is a practical method of infrared defect test.