面向中山坑数字地貌模拟的无人机(UAV)影像增强对比试验研究  

Comparative ExperimentalStudy of UAV Image Enhancementfor Digital Geomorphology Simulation of Zhongshan Crate

作　　者：杨明龙[1,2] 熊国来 高莎 马石林 唐秀娟[1,4] YANG Mingong;XIONG Guoai;GAO Sha;MA Shiin;TANG Xiujuan(Faculty of Land and Resources Engineering,Kunming University of Science and Technology,Kunming 650093,China;Application Engineering Research Center of Spatial Information Surveying and Mapping Technology in Plateau and Mountainous Areas Set by Universities in Yunnan Province,Kunming 650093,China;Yunnan Institute of Land Resources Planning and Design,Kunming 650216,China;Kunming Surveying and Mapping Institute,Kunming 650091,China)

机构地区：[1]昆明理工大学国土资源工程学院,云南昆明650093 [2]云南省高校高原山区空间信息测绘技术应用工程研究中心,云南昆明650093 [3]云南省国土资源规划设计研究院,云南昆明650216 [4]昆明市测绘研究院,云南昆明650091

出　　处：《昆明理工大学学报（自然科学版）》2024年第4期78-88,共11页Journal of Kunming University of Science and Technology(Natural Science)

基　　金：国家自然科学基金项目(41861054,62266026).

摘　　要：无人机(UAV)影像数据采集时,受拍摄设备、环境条件和自然光照等因素的影响,UAV获取的影像往往存在着噪声、模糊和对比度不足等问题,制约了数字地貌模拟的精度和可信度.面向中山坑数字地貌模拟的UAV影像增强对比试验研究,针对无人机影像预处理技术,确定综合影像质量评价指标,利用多种算法进行影像增强处理对比分析,实验优选出适用于无人机影像预处理的Retinex算法增强技术.对比直接灰度变换、Mask法、Wallis滤波器和基于Retinex四种影像增强处理方法,采用信息熵、方差、平均梯度等统计影像参数进行影像质量定量评价,并结合目视解译定性结果进行检查与核实,再用SIFT算法对增强后影像的地物边缘特征提取精度进行了综合比较.实证对比得出了Retinex算法在阴影去除和亮度校正方面表现出显著增强效果,且利用Retinex算法增强后影像特征提取结果客观评价指标对比值均有提高,整体精度值提高了4.1012%,卡帕系数值提高了0.0635.With the rapid advancement of low-altitude UAV technology,small consumer UAVs carrying optical sensors show the advantage of fast and flexible acquisition of high-resolution image data.However,due to the limitations of shooting equipment,environmental conditions and natural illumination,the image acquired by UAV often has problems such as noise,blurring and lack of contrast,which restricts the accuracy and reliability of digital geomorphology simulation.To further explore ways to improve the processing of UAV image data quality,this study studied the UAV image enhancement comparative experiment on the digital landform simulation of Zhongshan crater,carried out a comprehensive image quality evaluation index determination for the UAV image preprocessing technology,and conducted a comparative analysis of image enhancement processing using a variety of algorithms.The Retinex algorithm enhancement technology suitable for UAV image preprocessing in this study is optimized by experiments.Direct gray transform,Mask method,Wallis filter and four image enhancement processing methods based on Retinex were studied and compared.Statistical image parameters such as information entropy,variance and average gradient were used for quantitative image quality evaluation,and the qualitative results of visual interpretation were checked and verified.Therefore,the SIFT algorithm is further used to comprehensively compare the extraction accuracy of the edge features of the enhanced image.The empirical comparative study shows that the Retinex algorithm can significantly enhance shadow removal and brightness correction,and the objective evaluation index of image feature extraction results enhanced by the Retinex algorithm can improve the ratio.Among them,the OA value increased by 4.1012%and the Kappa value increased by 0.0635.

关 键 词：无人机 误差分析 质量评价 影像增强 

分 类 号：P23[天文地球—摄影测量与遥感] TP391.41[天文地球—测绘科学与技术]