遥感地理图像采集目标精准提取仿真

Accurate Extraction Simulation of Remote Sensing Geographic Image Acquisition Target

对遥感地理图像的采集目标进行准确提取,能够得到完整性较好的图像采集目标特征。图像采集目标的提取,需要将图像的采集目标各特征分量加权融合,合并为完整特征向量,完成遥感地理图像采集目标的提取。传统方法标记图像采集目标的轮廓像素,估计采集目标特征的单元向量权值,但忽略了合并特征向量,导致提取精度低。提出基于结构元分析的遥感地理图像采集目标精准提取方法,根据能量守恒和散度定理推导出遥感地理图像能量随时间的变化情况;构造图像各向异性扩散方程;生成图像相应的映射子图,提取采集目标的结构元和连通性特向量征,并将采集目标的各个特征分量进行加权融合,合并为采集目标的一组完整特征向量,对不同特征具有较好的区分能力。仿真证明,所提方法具有较好的去噪性能,且大大提高了遥感地理图像的提取精度。

This article puts forward a method for accurately extracting acquisition target of remote sensing geo- graphic image based on structural element analysis. According to the energy conservation and the divergence theo- rem, we derived that the energy of remote sensing geographic image changed with time. Then, we constructed anisotropic diffusion equation of image and established mapping sub graph corresponding to image to extract structure element of target and feature vector of connectivity. Moreover, we weighted and fused each characteristic component of collected target to form a set of complete feature vector, which had good distinguishing ability for different features. Simulation proves that the proposed method has good denoising performance, which greatly improves the extraction precision of remote sensing geographic image.