Taking Zhongba County, Angren County, Rikaze City located at the Upper and Middle Reaches of Yarlung Zangbo River as landscape units , this paper studied the difference of the landscape pattern under various degrees o...Taking Zhongba County, Angren County, Rikaze City located at the Upper and Middle Reaches of Yarlung Zangbo River as landscape units , this paper studied the difference of the landscape pattern under various degrees of human disturbance in the three areas. The results showed that: the three areas all reflected the same characteristic of landscape pattern in Qinghai-Tibet Plateau, the natural landscapes were in the absolute dominant position. However, from Zhangba to Rikaze, with human disturbance intensity increasing, anthropogenic features of landscapes became more and more outstanding, In Zhongba, the landscape structure appeared to be simpler with coarse grains and a less rich diversity, Conversely, in Rikaze, the landscape showed a complicated shape with finer grains and a relatively richer diversity, This reflected that the impact of human activities to natural landscape behaved a gradually-growing trend from the upper reach to the middle one of Yarlung Zangbo River Basin.展开更多
根据不同木材表面的光谱反射率差异可以对木材树种进行分类识别。在木制家具及工艺品的生产实践中,考虑到防止木材腐败、开裂,美化木制品外表及延长木制品使用寿命等原因,经常需要在木材表面涂抹某种涂饰。涂抹涂饰将导致木材表面光谱...根据不同木材表面的光谱反射率差异可以对木材树种进行分类识别。在木制家具及工艺品的生产实践中,考虑到防止木材腐败、开裂,美化木制品外表及延长木制品使用寿命等原因,经常需要在木材表面涂抹某种涂饰。涂抹涂饰将导致木材表面光谱反射率曲线产生漂移和变形,经实验验证无法使用原始木材表面的光谱反射率训练出的分类器模型对涂饰木材光谱曲线进行分类识别。相对于原始木材光谱曲线,涂饰木材光谱曲线的漂移和变形可以用非线性模型来拟合;而这种非线性拟合一般使用神经网络来实现。为了能够继续使用原始木材光谱反射率训练的分类器模型,使用全连接神经网络拟合了原始木材光谱反射率和涂饰木材光谱反射率之间的关系模型,通过该模型对涂饰木材光谱反射率进行校正,实现使用原始木材光谱所训练的分类器模型对涂饰木材进行分类识别的目的。此外,还使用卷积神经网络对光谱反射率提取卷积特征,引入表征原始木材光谱反射率和涂饰木材光谱反射率的卷积特征之间关系的隐藏层,将涂饰木材光谱反射率的卷积特征进行校正,并通过输出层输出其分类结果。为了验证所提出的校正模型的有效性,本文以20种木材样本的近红外光谱(950~1650 nm/near infrared spectra,NIR)和可见光/近红外光谱(350~1000 nm/visible and near infrared spectra,VIS/NIR)为研究对象,对比了8种不同涂饰建立的校正模型性能。实验结果表明,NIR的校正分类效果要好于VIS/NIR的校正分类效果;卷积神经网络的校正模型可以将涂抹透明涂饰木材表面的NIR分类正确率提高至70%以上;全连接网络模型可以将涂抹透明涂饰木材表面的NIR分类正确率提高至80%以上,但两种模型都无法对非透明涂饰进行校正。从模型的训练速度和识别效率上看,卷积神经网络的校正模型要好于全连接神经网络的校正模型。综上所述,通过神经网络建立起的原始木材光谱反射率和涂饰木材光谱反射率之间的非线性关系模型,可以对涂抹透明涂饰的木材光谱曲线进行校正。进而实现直接使用原始木材光谱反射率所训练出的分类器模型对涂抹透明涂饰木材光谱曲线进行分类识别,使得木材树种分类识别应用领域从原始木材扩展到涂抹透明涂饰木材,具有较好的实际应用意义和前景。展开更多
文摘Taking Zhongba County, Angren County, Rikaze City located at the Upper and Middle Reaches of Yarlung Zangbo River as landscape units , this paper studied the difference of the landscape pattern under various degrees of human disturbance in the three areas. The results showed that: the three areas all reflected the same characteristic of landscape pattern in Qinghai-Tibet Plateau, the natural landscapes were in the absolute dominant position. However, from Zhangba to Rikaze, with human disturbance intensity increasing, anthropogenic features of landscapes became more and more outstanding, In Zhongba, the landscape structure appeared to be simpler with coarse grains and a less rich diversity, Conversely, in Rikaze, the landscape showed a complicated shape with finer grains and a relatively richer diversity, This reflected that the impact of human activities to natural landscape behaved a gradually-growing trend from the upper reach to the middle one of Yarlung Zangbo River Basin.
文摘根据不同木材表面的光谱反射率差异可以对木材树种进行分类识别。在木制家具及工艺品的生产实践中,考虑到防止木材腐败、开裂,美化木制品外表及延长木制品使用寿命等原因,经常需要在木材表面涂抹某种涂饰。涂抹涂饰将导致木材表面光谱反射率曲线产生漂移和变形,经实验验证无法使用原始木材表面的光谱反射率训练出的分类器模型对涂饰木材光谱曲线进行分类识别。相对于原始木材光谱曲线,涂饰木材光谱曲线的漂移和变形可以用非线性模型来拟合;而这种非线性拟合一般使用神经网络来实现。为了能够继续使用原始木材光谱反射率训练的分类器模型,使用全连接神经网络拟合了原始木材光谱反射率和涂饰木材光谱反射率之间的关系模型,通过该模型对涂饰木材光谱反射率进行校正,实现使用原始木材光谱所训练的分类器模型对涂饰木材进行分类识别的目的。此外,还使用卷积神经网络对光谱反射率提取卷积特征,引入表征原始木材光谱反射率和涂饰木材光谱反射率的卷积特征之间关系的隐藏层,将涂饰木材光谱反射率的卷积特征进行校正,并通过输出层输出其分类结果。为了验证所提出的校正模型的有效性,本文以20种木材样本的近红外光谱(950~1650 nm/near infrared spectra,NIR)和可见光/近红外光谱(350~1000 nm/visible and near infrared spectra,VIS/NIR)为研究对象,对比了8种不同涂饰建立的校正模型性能。实验结果表明,NIR的校正分类效果要好于VIS/NIR的校正分类效果;卷积神经网络的校正模型可以将涂抹透明涂饰木材表面的NIR分类正确率提高至70%以上;全连接网络模型可以将涂抹透明涂饰木材表面的NIR分类正确率提高至80%以上,但两种模型都无法对非透明涂饰进行校正。从模型的训练速度和识别效率上看,卷积神经网络的校正模型要好于全连接神经网络的校正模型。综上所述,通过神经网络建立起的原始木材光谱反射率和涂饰木材光谱反射率之间的非线性关系模型,可以对涂抹透明涂饰的木材光谱曲线进行校正。进而实现直接使用原始木材光谱反射率所训练出的分类器模型对涂抹透明涂饰木材光谱曲线进行分类识别,使得木材树种分类识别应用领域从原始木材扩展到涂抹透明涂饰木材,具有较好的实际应用意义和前景。