遥感影像提取森林火灾可燃物类型的研究

Research on Classification of Forest Fuel Based on Remote Sensing Image

采用最大似然法、参数优化的神经网络法和支持向量机法对数据处理后的遥感影像进行分类,研究不同时相、影像质量和分类方法对森林火灾可燃物分类精度的影响,提出了森林火灾可燃物的粗分类标准,并实验验证了分类标准的可行性,采用神经网络法对时效性好、质量高的TM影像进行森林火灾可燃物分类的精度最高。

Processed remote sensing images were classified by maximum likelihood method, neural networks method of parameter optimization andsupport vector machine method of parameter optimization. Studies were conducted on influence of different imaged time, image quality andclassification method on the accuracy of forest fuel classification. Experiments proved the feasibility of classification criteria, that the classificationaccuracy was higher with good imaged time, higher quality of the TM images by neural network method.