摘要
从无人机影像的特点出发,结合SIFT和CenSurE算子的特点及常用的一些匹配策略进行实验.首先分析了两种算子在特征提取有效性、特征提取速度方面的优劣,然后运用相应的匹配策略进行影像的粗匹配和精匹配(RANSAC);最后根据实验所得的匹配正确率、配准精度和速度、配准效果评价指标,对两种算子在无人机影像配准的优劣做出评价,实验得出SIFT具有速度上的劣势、实时性不佳,但综合性能好,更加适合无人机影像的配准.
Beginning with characteristics of UAV image and matching algorithm, includes SIFT and CenSurE, the article applies some formal matching strategies on making experiment. First of all, both SIFT and CenSurE＇s efficiency and speed on feature extraction are analyzed;Second,using appropriate matching strategy to making roughness matching and accuracy matching; Lastly, making an conclusion about matching choice of two algorithms based on matching accuracy, matching precision and speed, matching effect evaluation indicator. The result shows that for comprehensiveness, SIFT is more appropriate than CenSurE on UAV image matching, except speed or realtime.
出处
《河南科学》
2013年第9期1417-1423,共7页
Henan Science
