摘要
针对赤潮检测方法的及时性和普适性较差的问题,选取美国国家航空航天局(NASA)对地观测系统(EOS)所获取的多景MODIS遥感图像,经过光谱分析和特征选择获得所需数据集,采用数据挖掘技术训练赤潮检测的决策树分类模型,抽取赤潮分类规则。通过对2004年5到6月份一次赤潮爆发周期内获取的8景MODIS遥感影像的赤潮检测,验证了利用挖掘出的分类规则对渤海黄河入海口及邻近海域所发生的赤潮进行检测可取得较满意的结果。
Taking into account the timeliness and universal drawbacks in the red tide detection, we select several MODIS imagery obtained from NASA＇s EOS, and get the necessary data set through spectral analysis and feature selection, then we train the classification model of red tide detection through data mining method and extract the classification rules. The classification rules is used to detect the 8 Modis imagery which acquired in the period of May 2004 to June. The results show that the usage of the classification rules can get good detection performance in the detection of red tide which outbreaks in Yello River and adjacent seas.
出处
《青岛大学学报（自然科学版）》
CAS
2012年第2期47-52,共6页
Journal of Qingdao University(Natural Science Edition)
基金
国家自然科学基金重大专项(91130035)
国家自然科学基金(60974085)
国家公益性行业科研专项(200905030)
海洋沉积与环境地质国家海洋局重点实验室开放基金(MASEG200812)
国家海洋局海洋赤潮灾害立体监测技术与应用重点实验室开放基金(MATHAB20120301)
