中文摘要：

基于红外热波方法对某型固体火箭发动机壳体含脱粘缺陷试件进行了数值仿真和实验研究。为了得到直观、精确的缺陷三维图像，采用同态滤波方法对原始图像进行降噪增强处理，利用粒子群模糊聚类算法对缺陷进行了分割，并对缺陷的大小和深度进行了定量识别，在此基础上，利用体绘制重建算法对缺陷进行三维重建。结果表明：同态滤波方法信噪比高，图像增强效果好；粒子群模糊聚类算法分割效果好，缺陷保真性高；缺陷定量识别的精度较高，三维重建效果较好，为实现缺陷的自动识别奠定了基础。

英文摘要：

Debond defects between shell and insulation layer of solid rocket motor （SRM） were inspec- ted by numerical analysis and experiment based on thermal wave nondestructive testing （TWNDT） technique. In order to gain intuitionistic and accurate image of defects, the homomorphic filtering was used to enhance the image quality and the particle swarm optimization fuzzy clustering method （PSO- FCM） was applied to segment the defect. Then the size and depth were estimated quantificationally. On this basis, the defects can be 3D reconstructed by volume rendering. The results show that homo- morphic filtering has higher peak signal to noise ratio（PSNR）and can improve image quality; the PSO- FCM can achieve the better effect of image segmentation; it could be detected quantitatively and 3D reconstructed effectively for the defects, which establish basis for future research of defect auto-identi- fication.