摘要
针对传统逐像素对胡杨林进行识别的方法忽略了相邻像元之间的空间关联性及识别精度低等问题,结合不同层次的特征知识,提出一种多模态模型的胡杨林语义信息描述与识别方法。借助边缘滤波对原始图像进行降噪,保留待识别物体的边缘信息提高局域平滑度;将提取的多重空间特征作为多模态模型(CNN _ LSTM)的输入,在进行时间和空间扩展的同时进一步挖掘图像中胡杨的深层语义特征,采用SoftMax分类器实现胡杨林的正确识别。实验结果表明,该方法优于传统的胡杨林识别方法。
Because traditional per-pixel recognition method for Populus diversifolia ignores the spatial correlation between adjacent pixels ,and shows low accuracy of recognition problems,combined with the feature of different levels of knowledge,a multimodal Populus diversifolia semantic information description and recognition method was proposed. The original image was denoted by edge filtering to preserve the edge information of the object to be identified to improve the local smoothness. The extracted multidimensional feature was used as input of the multi-modal model (CNN _ LSTM),and the deep semantic features of Populus diversifolia in the image were further excavated while the time and space were extended and the SoftMax classifier was used to realize the correct identification of Populus diversifolia. Experimental results show that this method is better than the traditional method.
作者
王媛
阿里甫·库尔班
李均力
吕亚龙
阿依加马力·克然木
WANG Yuan;Alifu·Kuerban;LI Jun-li;LYU Ya-long;Ayijiamali·Keranmu(College of Information Science and Engineering,Xinjiang University,Urumqi 830046,China;College of Software,Xinjiang University,Urumqi 830008,China;College of Resource and Environment Sciences,Xinjiang University,Urumqi 830046,China;Xinjiang Key Laboratory of Oasis Ecology,Xinjiang University,Urumqi 830046,China;Xinjiang Institute of Ecology and Geography,Chinese Academy of Sciences,Urumqi 830011,China)
出处
《计算机工程与设计》
北大核心
2019年第7期1978-1983,共6页
Computer Engineering and Design
基金
国家自然科学基金项目(31570536、61163029、61562084、U1703102)
关键词
空间关联性
边缘滤波
多模态模型
深层特征
识别方法
spatial correlation
edge filter
multimodal model
deep characteristics
recognition method
