CAJ | 학술논문

针对 Bayer 真彩色遥感影像进行特征匹配时，彩色描述符的效果与适用性问题一直研究较少。结合 Bayer 真彩色遥感影像成像变化规律和彩色描述符算法特点，从理论上分析彩色描述符的不变性。提出模拟数据评价、不同类别影像评价、任务总体适用性评价等三种实验方法对彩色描述符适用性进行实验验证和分析。通过理论分析和实验评估，总体上彩色描述符中 RGBSIFT 算法效果最优，对 Bayer 真彩色航空影像特征匹配有较好的适用性，并且不同地物属性影像在特征匹配时有不同的最优彩色描述符。
침대 Bayer 진채색요감영상진행특정필배시，채색묘술부적효과여괄용성문제일직연구교소。결합 Bayer 진채색요감영상성상변화규률화채색묘술부산법특점，종이론상분석채색묘술부적불변성。제출모의수거평개、불동유별영상평개、임무총체괄용성평개등삼충실험방법대채색묘술부괄용성진행실험험증화분석。통과이론분석화실험평고，총체상채색묘술부중 RGBSIFT 산법효과최우，대 Bayer 진채색항공영상특정필배유교호적괄용성，병차불동지물속성영상재특정필배시유불동적최우채색묘술부。
It is unclear that what color descriptors best fits feature matching of aerial remote sensing Bayer real-color image.To analyze variation of imaging of the aerial image and algorithm characteristics of color descriptors,a theoretical analysis of invariant properties of color descriptors was presented.The effect and applicability of descriptors were verified and analyzed by three experimental methods which are evaluations of simulation data,different categories images and general applicability of task.The theoretical and experimental results show that RGBSIFT has best applicability for the feature matching of aerial Bayer real-color image.The usefulness of color descriptor is category-specific.