| 热红外港口图像特征的评估器提取与选择 |
| |
| 引用本文： | 马兰,陈筱勇,邓国臣.热红外港口图像特征的评估器提取与选择[J].测绘科学,2017,42(5). |
| |
| 作者姓名： | 马兰 陈筱勇 邓国臣 |
| |
| 作者单位： | 1. 信息工程大学,郑州 450000;国际关系学院,南京 210039;2. 国际关系学院,南京,210039;3. 中国测绘科学研究院,北京,100830 |
| |
| 摘 要： | 针对低对比度、条纹噪声、低空间分辨率等特点而导致的热红外图像识别效果不佳问题,提出了一种港口目标热红外遥感图像特征提取与选择方法,实现了一定情况下港口目标的高精度分类。采用纹理、几何等29个特征,通过评估器选择最佳特征组合,并根据识别精度选择最佳分类器,能生成热红外图像港口目标22个最佳分类特征,且具有一定的鲁棒性。经过参数优化后的libSVM(一种支持向量机)分类器分类精度较高;白天图像比夜间图像分类精度更高;像素值、灰度直方图相关的一维和二维统计特征、局部二进制模式特征、边缘方向直方图特征等与灰度和纹理相关的特征对港口目标热红外图像识别影响较大。
|
| 关 键 词： | 特征提取 特征选择 热红外图像 港口目标 评估器 分类器 热红外遥感 |
Evaluator-based extraction and selection of thermal infrared harbor image features |
| |
| MA Lan,CHEN Xiaoyong,DENG Guochen.Evaluator-based extraction and selection of thermal infrared harbor image features[J].Science of Surveying and Mapping,2017,42(5). |
| |
| Authors: | MA Lan CHEN Xiaoyong DENG Guochen |
| |
| Abstract: | To improve the inadequate effectiveness of thermal infrared (TIR) image recognition due to low contrast,striping noise and low spatial resolution,we have conducted research with respect to harbor targets and developed a method for extracting and selecting features of TIR remote sensing images of such targets.We have adopted 29 sample features like texture,geometry and so on,selected the best combination of image features with an attribute selection evaluator,and chosen the best classifier and the best classifier parameters according to the accuracy of recognition.This process can generate the best 22 features for the classification of TIR images of harbor targets with a certain degree of robustness.The results of the experiment indicate that for TIR images,the libSVM (a support vector machine),after parameter optimization,is the most accurate classifier;daytime images can be more accurately classified than nighttime ones;and pixel value,the one-dimensional and two-dimensional statistical features of gray level histogram,edge direction histogram and local binary patterns which reated to gray and texture are more meaningful for the recognition of TIR remote sensing images of harbor targets.we have developed a method for extracting and selecting features of TIR remote sensing images of harbor targets,and achieved higher classification accuracy for harbor targets under certain conditions. |
| |
| Keywords: | feature extraction feature selection thermal infrared image harbor target evaluator classifier thermal infrared remote sensing |
| 本文献已被 CNKI 万方数据 等数据库收录！ |
|
