期刊文献+
共找到2篇文章
< 1 >
每页显示 20 50 100
基于Mask R-CNN的柑橘主叶脉显微图像实例分割模型 被引量:7
1
作者 翁海勇 李效彬 +3 位作者 肖康松 丁若晗 贾良权 叶大鹏 《农业机械学报》 EI CAS CSCD 北大核心 2023年第7期252-258,271,共8页
针对目前植物解剖表型的测量与分析过程自动化低,难以应对复杂解剖表型的提取和识别的问题,以柑橘主叶脉为研究对象,提出了一种基于掩膜区域卷积神经网络(Mask region convolutional neural network,Mask R-CNN)的主叶脉显微图像实例分... 针对目前植物解剖表型的测量与分析过程自动化低,难以应对复杂解剖表型的提取和识别的问题,以柑橘主叶脉为研究对象,提出了一种基于掩膜区域卷积神经网络(Mask region convolutional neural network,Mask R-CNN)的主叶脉显微图像实例分割模型,以残差网络ResNet50和特征金字塔(Feature pyramid network,FPN)为主干特征提取网络,在掩膜(Mask)分支上添加一个新的感兴趣区域对齐层(Region of interest Align,RoI-Align),提升Mask分支的分割精度。结果表明,该网络架构能够精准地对柑橘主叶脉横切面中的髓部、木质部、韧皮部和皮层细胞进行识别分割。Mask R-CNN模型对髓部、木质部、韧皮部和皮层细胞的分割平均精确率(交并比(IoU)为0.50)分别为98.9%、89.8%、95.7%和97.2%,对4个组织区域的分割平均精确率均值(IoU为0.50)为95.4%。与未在Mask分支添加RoI-Align的Mask R-CNN相比,精度提升1.6个百分点。研究结果表明,Mask R-CNN模型对柑橘主叶脉各类组织区域具有良好的识别分割效果,可为柑橘微观表型研究提供技术支持与研究基础。 展开更多
关键词 柑橘主叶脉 显微图像 掩膜区域卷积神经网络 实例分割 微观表型
在线阅读 下载PDF
Modeling bidirectional reflection distribution function of microscale random rough surfaces
2
作者 王爱华 HSU P.F. 蔡九菊 《Journal of Central South University》 SCIE EI CAS 2010年第2期228-234,共7页
The radiative properties of three different materials surfaces with one-dimensional microscale random roughness were obtained with the finite difference time domain method(FDTD) and near-to-far-field transformation.Th... The radiative properties of three different materials surfaces with one-dimensional microscale random roughness were obtained with the finite difference time domain method(FDTD) and near-to-far-field transformation.The surface height conforms to the Gaussian probability density function distribution.Various computational modeling issues that affect the accuracy of the predicted properties were discussed.The results show that,for perfect electric conductor(PEC) surfaces,as the surface roughness increases,the magnitude of the spike reduces and eventually the spike disappears,and also as the ratio of root mean square roughness to the surface correlation distance increases,the retroreflection becomes evident.The predicted values of FDTD solutions are in good agreement with the ray tracing and integral equation solutions.The overall trend of bidirectional reflection distribution function(BRDF) of PEC surfaces and silicon surfaces is the same,but the silicon's is much less than the former's.The BRDF difference from two polarization modes for the gold surfaces is little for smaller wavelength,but it is much larger for the longer wavelength and the FDTD simulation results agree well with the measured data.In terms of PEC surfaces,as the incident angle increases,the reflectivity becomes more specular. 展开更多
关键词 bidirectional reflection distribution fimction random rough surfaces Maxwell equations finite difference time domain method
在线阅读 下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部