摘要： 根据DR图像目标边缘模糊、对比度低的特点，将小波多尺度分解与基于水平集演化的CV模型相结合，由此加快主动轮廓线向缺陷边缘收敛的速度，通过改进亚像素线性插值方法，使分割精度从整像素级提高到亚像素级，从而实现缺陷的亚像素测量。仿真结果表明，该方法的测量精度优于传统方法，对实际铸件DR图像的实验结果验证了其高效性。

关键词: DR图像, 缺陷测量, 亚像素, 小波多尺度分解, CV模型

Abstract: Aiming at the problem that Digital Radiography(DR) images have blurred edges and low contrast, this paper uses a method combining wavelet multi-scale decomposition and Chan-Vese(CV) model to expedite the active contour converges to the defect edges. The edge reaches sub-pixel accuracy combining with the improved linear interpolation and the sub-pixel measurement of defects is obtained. Simulation results show that the measuring accuracy of the method is superior to the traditional methods. Experiment on real casting DR images indicates it is feasible and efficient.

Key words: Digital Radiography(DR) image, defect measuring, sub-pixel, wavelet multi-scale decomposition, Chan-Vese(CV) model

中图分类号: 

• TP391