基于证据理论的水稻遥感信息提取方法研究

A extraction method of rice remote sensing information based on evidence theory

水稻播种面积是农业管理部门关心的重要问题之一。本文介绍了证据理论在遥感图像分类上的应用,并以汉川市为示范区,采用2002年TM遥感影像,在GIS技术支持下,通过建模运算,剔除不可能是水稻的像元。然后将证据理论用于影像分类,将影像初次分类结果,与参考图对照,将不满足要求的区域提取出来,再次进行第二次分类。将第二次分类结果与参考图对照,显示分类效果满足要求;若不满足要求,可继续进行再次分类,直到分类效果满意为止。将两次分类结果中的水稻信息进行叠加,提取出水稻遥感信息,经检验精度达到91.39%。

Rice planting area is one of the most concerned problems in agricultural administration department. Having introduced the application of evidence theory in remote sensing image classification, this paper takes Hanchuan city as a study area, adopts 2002 TM remote sensing image and GIS technique and operation of modeling, rejects non -rice pixels. Then evidence theory is used for image classification. The first classified result, the comparison with reference map, picks up the area that does not satisfy the requirement. Next, new image carries on the second classification. The second classified result is also compared with the reference map. If it does not satisfy the requirement, the classification will be continued until the classification result is satisfied. The rice remote sensing information is gotton from overlaying of two＇classified results. Its clssification accuracy is up to 91. 39% through examining.