论文部分内容阅读
针对密封装配组件无损探伤中难于进行精确图像边缘分割的情况,探讨了一种改进的边缘分割融合算法。该方法借助模糊数学形态学中的开闭算法对图像作平滑处理,然后对图像进行梯度算子计算,再将得到的梯度图像作分割处理就可得到精确的图像边缘。以某工件被检断面的二维灰度图像为例进行了仿真实验,从图像上可直观地看到目标细节的位置、形状、大小。仿真结果表明,该算法可更好地消除过分割现象,更适合在无损检验中应用。
In view of the difficulty of accurate image edge segmentation in non-destructive testing of seal assemblies, an improved edge segmentation fusion algorithm is discussed. This method uses the open-close algorithm in fuzzy mathematical morphology to smooth the image, and then performs gradient operator calculation on the image, and then obtains the accurate image edge by dividing the obtained gradient image. Taking a two-dimensional grayscale image of a workpiece being examined as an example, a simulation experiment is carried out, and the position, shape and size of the target detail can be visually observed from the image. Simulation results show that this algorithm can eliminate the over-segmentation phenomenon and is more suitable for non-destructive testing.