固体发动机ICT图像缺陷检测中的插值方法研究

Study on Interpolation Algorithm in Solid Motor ICT Image Defect Test

固体发动机的缺陷主要有药柱内气孔、裂纹和多界面脱粘,这些缺陷在ICT检测生成的断层图像中,和整幅断层图像相比只占据少量像素,给缺陷分析带来困难。采用图像插值技术放大含有缺陷的局部断层图像,可获得更丰富的缺陷信息。文章将图像插值技术分为基于图像插值方法和基于对象插值方法两类,分别进行分析、比较,并给出含有典型缺陷的固体发动机断层局部图像放大4×4倍后的图像。经过对比,基于对象的插值方法效果明显优于基于图像的插值放大方法,但其计算量大、耗时多。在固体发动机断层缺陷图像放大分析中,今后研究重点应是简化现有的或提出新的基于对象的插值方法,保证插值精度基础上提高插值速度。

Defect in solid rocket motor is classified into pore, crack and multi-interface debonding. These defects just hold a little pixel in the whole ICT image which created by industry computed tomography. So the analysis of defect is difficult. Image interpolation technology is used to magnify the local ICT image which contains defects. Much more defect information is obtained through amplification. There are two kinds of interpolation is introduced in this paper. One is scene-based interpolation method, and another is object-based interpolation method. The image which contains representative solid rocket motor defects is magnified to 4 × 4 times are described. Through comparison, the object-based interpolation method is better than the scene-based interpolation method, but its computation is complex. The research in solid rocket motor ICT defect image analysis should focus on predigest existing or bring forward a new objectbased interpolation method for the future, so as to improve interpolation speed on the base of interpolation precision.