论文部分内容阅读
针对目前土地覆盖变化检测常用的方法存在不同程度的误差累积,夸大了变化区域,提出模糊支持向量机(FSVM)和变化矢量分析(CVA)相结合的土地覆盖检测方法。以某矿区2004年和2008年两期的CBERS遥感影像进行了试验。结果表明,植被大幅减少,其他地类都有不同程度的增加,主要是由于开采规模和产量提升所致。通过与常规的其他两类方法比较发现,本文方法的总体精度、Kappa系数、漏检误差和虚检误差分别为92.67%、0.892 7%、5.79%、7.31%,比其他两种方法有较大提高,能够提供较全面的变化类别和准确信息,可以有效地应用于矿区土地覆盖动态监测。
There are different degrees of error accumulation for the commonly used methods of land cover change detection, which exaggerate the changing area. A land cover detection method based on Fuzzy Support Vector Machine (FSVM) and Variable Vector Analysis (CVA) is proposed. A CBERS remote sensing image of a mine in 2004 and 2008 was tested. The results showed that the vegetation decreased drastically and that of other types of land increased in varying degrees, mainly due to the increase of mining scale and output. Compared with the other two conventional methods, the overall accuracy, Kappa coefficient, missed detection error and false detection error of the proposed method are 92.67%, 0.892 7%, 5.79% and 7.31% respectively, which are larger than the other two methods Improve and provide more comprehensive change categories and accurate information, which can be effectively applied to the dynamic monitoring of land cover in mining areas.