测绘学报 ›› 2019, Vol. 48 ›› Issue (5): 583-596.doi: 10.11947/j.AGCS.2019.20180122

• 摄影测量学与遥感 • 上一篇    下一篇

遥感影像要素提取的可变结构卷积神经网络方法

王华斌1,2, 韩旻1,2, 王光辉2, 李玉1   

  1. 1. 辽宁工程技术大学测绘与地理科学学院, 辽宁 阜新 123000;
    2. 自然资源部国土卫星遥感应用中心, 北京 100048
  • 收稿日期:2018-03-20 修回日期:2018-08-27 出版日期:2019-05-20 发布日期:2019-06-05
  • 通讯作者: 韩旻 E-mail:mzlapqowjf321@qq.com
  • 作者简介:王华斌(1980-),男,副研究员,研究方向为卫星遥感应用。E-mail:whb@sasmac.cn
  • 基金资助:
    国家重点研发计划(2016YFB0501403)

Surface features extraction in remote sensing images based on architecture-variant CNN

WANG Huabin1,2, HAN Min1,2, WANG Guanghui2, LI Yu1   

  1. 1. School of Geomatics, Liaoning Technical University, Fuxin 123000, China;
    2. Land Satellite Remote Sensing Application Center, MNR, Beijing 100048, China
  • Received:2018-03-20 Revised:2018-08-27 Online:2019-05-20 Published:2019-06-05
  • Supported by:
    The National Key Research and Development Program of China (No. 2016YFB0501403)

摘要: 针对利用经典卷积神经网络提取遥感影像地物要素的方法中,模型容量受既定网络固有结构的限制而难以得到良好提取效果的问题,提出利用可变结构卷积神经网络的遥感影像要素提取方法。该方法将结构搜索与权重求解过程统一,在定义卷积神经网络架构的基础上将其中的关键结构作为变量,并以要素提取精度指标作为目标函数,利用遗传算法求解网络结构,最后以该网络为模型提取遥感影像中的目标要素。相关试验表明,可变结构卷积神经网络具备灵活的模型容量,对遥感影像中目标要素的提取效果良好。

关键词: 地物要素提取, 卷积神经网络, 可变结构, 遗传算法

Abstract: To exceed limited capacity of established convolutional neural network(CNN) with fixed architecture in traditional surface feature extraction in remote sensing images, we propose a new feature extraction method based on architecture-variant convolutional neural network (AVCNN). In AVCNN, key units are variables and the performance of the unknown model become object function. That means architecture search is added before traditional weights solving. Genetic algorithm is introduced to search proper architecture and classical algorithm is used to solve unknown weights in the candidate CNN. The CNN with final architecture is used to extract the surface feature in remote sensing images. The experiment result shows that AVCNN has flexible capacity and performances well in surface features extraction in remote sensing images.

Key words: surface feature extraction, convolution neural network, variant architecture, genetic algorithm

中图分类号: