论文部分内容阅读
利用Creator建立目标模型,Vega的TMM工具进行纹理材质映射,基于视景仿真技术建立了多波段多极化SAR图像数据库。设计了融合遗传算法和二值粒子群的混合智能优化算法,对SAR图像的波段极化组合方式进行优化;基于未矫正和矫正后的图像分别提取Zernike矩、Gabor小波系数等构成候选特征序列,进行了多波段多极化SAR图像特征选择实验。实验结果表明,采用仿真技术建立SAR图像数据库是进行多波段多极化SAR图像识别的一种有效手段;采用优化后的特征集合能够提高多波段多极化SAR图像的识别率。
Using Creator to establish the target model, Vega’s TMM tool performs texture material mapping and establishes a multi-band multi-polarization SAR image database based on the scene simulation technology. A hybrid intelligent optimization algorithm based on fusion genetic algorithm and binary particle swarm optimization is designed to optimize the combination of band polarizations in SAR images. Zernike moments and Gabor wavelet coefficients are extracted from the uncorrected and corrected images to form candidate feature sequences, Multi-band multi-polarization SAR image feature selection experiments. Experimental results show that the establishment of SAR image database using simulation technology is an effective method for multi-band multi-polarization SAR image recognition. Using the optimized feature set can improve the recognition rate of multi-band multi-polarization SAR images.