论文部分内容阅读
在航空发动机涡轮叶片的缺陷检测线上,缺陷尺寸的自动测定是实现在线检测的关键。为了解决这一问题,分析了叶片DR(DigitalRadiography)图像的纹理特点,通过区域分割点将图像分成不同纹理区,提取各个区域的灰度分布函数以减少纹理对缺陷的干扰。消隐纹理后的图像灰度服从正态分布,遵循背景和信号(缺陷)在正态分布曲线上的取值特点,设置灰度阈值,将灰度值并入集合和数组,最后对航空发动机涡轮叶片的缺陷进行了正确的尺寸和形状测定。
In the aero-engine turbine blade defect detection line, the automatic determination of the defect size is the key to achieve online detection. In order to solve this problem, the texture features of DR images are analyzed. The image is divided into different texture regions by region segmentation points, and the gray distribution function of each region is extracted to reduce the interference of texture to the defects. The grayscale of the hidden image obeys the normal distribution, follows the background and the value of the signal (defect) on the normal distribution curve, sets the gray threshold and integrates the gray value into the set and the array. Finally, Turbine blade defects were correctly dimensioned and shaped.