基于优化导向滤波算法的遥感图像预处理仿真  

Remote Sensing Image Preprocessing Simulation Based on Optimized Oriented Filtering Algorithm

作　　者：么鸿原 林雪原 王海鹏 李雪腾 YAO Hong-yuan;LIN Xue-yuan;WANG Hai-peng;LI Xue-teng(Naval Aviation University,Yantai Shandong 264000,China)

机构地区：[1]海军航空大学

出　　处：《计算机仿真》2019年第9期300-303,共4页Computer Simulation

基　　金：国家自然科学基金(61531020,61471383)

摘　　要：针对无人机遥感影像拼接中因图像噪声大、光照差异大、视野中景物存在畸变、图像不清晰等因素影响而导致的图像特征点提取数量少、特征点误匹配率高、拼接效果不理想甚至拼接错误等问题,提出了一种基于优化导向滤波算法的遥感图像预处理算法。上述优化算法包括对原始图像的几何校正、辐射校正、噪声平滑与细节增强等步骤。在仿真阶段,应用上述算法对优化SIFT匹配算法进行改进后发现,改进后的匹配算法在特征提取数目上比起为原算法平均增加了约16%,且图像细节明显增强。实验表明,所提出的优化算法,能够有效减小图像噪声、校正畸变、增强图像细节,使后续的配准拼接工作更加高效准确。For the remote sensing image mosaic of UAV,due to large image noise,large illumination difference,distortion of the scene in the field of view,unclear image and other factors,the number of image feature points extracted is small,the mismatch rate of feature points is high,and the stitching effect is not ideal for even splicing errors,this paper proposes a remote sensing image preprocessing algorithm based on optimized steering filtering algorithm.The optimization algorithm includes geometric correction,radiation correction,noise smoothing and detail enhancement in the original image.In the stage,the improved algorithm was used to improve the optimized SIFT matching algorithm proposed in Reference 8.It is found that the improved matching algorithm increases the number of feature extraction by about 16%compared with the original algorithm,and the image effect is obviously enhanced.It shows that the optimization algorithm proposed in this paper can effectively reduce image noise,correct distortion,enhance image details,and make subsequent registration and splicing work more efficient and accurate.

关 键 词：无人机遥感图像 预处理 畸变校正 直方图匹配 导向滤波 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]