| 遥感图象地面植被的分类识别 |
| |
| 引用本文： | 李永宁,游志胜.遥感图象地面植被的分类识别[J].四川大学学报(自然科学版),1989,26(3):283-290. |
| |
| 作者姓名： | 李永宁 游志胜 |
| |
| 作者单位： | 四川大学模式识别研究室
(李永宁，游志胜，聂建荪)，四川大学模式识别研究室(王开平) |
| |
| 基金项目： | 国家自然科学基金资助项目 |
| |
| 摘 要： | 对图象自动纹理分类识别中的纹理特征和分类判别方法进行了讨论.利用三类图象纹理特征组合(二阶统计量,一阶统计量及付里叶功率谱)设计出fisher线性分类器,对广州郊区的地面航空照片进行了大量植被分类判别试验,得到较好的分类结果.使用“留出一个”的分类器测试方法,最高正确率达95％.
|
| 关 键 词： | 遥感图象 地面植被 分类 模式识别 |
COMPOSITE TEXTURE FEATURES STUDY FOR VEGETATION CLASSIFICATION |
| |
| Li Yongning You Zhisheng Nie Jiansheng Wang Kaiping.COMPOSITE TEXTURE FEATURES STUDY FOR VEGETATION CLASSIFICATION[J].Journal of Sichuan University (Natural Science Edition),1989,26(3):283-290. |
| |
| Authors: | Li Yongning You Zhisheng Nie Jiansheng Wang Kaiping |
| |
| Institution: | Pattern recognition Lab |
| |
| Abstract: | This paper describes an approach to automatic texture classification, which makes use of composite texture features based on the Fourier power spectrum,second-order gray level statistics, and first-order statistics of gray level difference. Using the Fisher Linear Discrimination and "Leave one-out" testing method,various composite feature sets were used to classify 120 samples subimages belonging to four vegetation classes. The best correct classification rate is 95 percent. These results indicate that the multi-class composite texture features have good performance for vegetation classification in remote sensing images. |
| |
| Keywords: | texture feature image classification remote sensing pattern recognition |
| 本文献已被 CNKI 维普 等数据库收录！ |
