Paper Infomation
Retrieval of Optical Remote Sensing Image Content Based on Convolution Neural Network
Full Text(PDF, 822KB)
Author: Tengfei Ji, Yameng Zhao, Tao Yu, Bing Zhou, Xiangzhi Huang
Abstract: The insufficiency of image detection method based on bottom feature extraction is analyzed and an optical remote sensing image content retrieval method based on convolution neural network is proposed. Firstly, convolutional neural network is used to train the rsscn7-master remote sensing image data set of the training sample. Image features are extracted from the sample to build the image feature library. Then, Sofemax classifier is used to classify the feature map to achieve precise classification and improve the accuracy of the model. Then, the model is tested with the help of the sample map or the actual optical remote sensing image map, and good experimental results are obtained. Experimental results show that this method is effective in the content retrieval of optical remote sensing images and has high retrieval accuracy.
Keywords: Remote Sensing Image Content Retrieval, Image Classification, Convolution Neural Network, Dropout, Sofemax Classifier, RSSCN7 - master Data Set
References:
[1] 张鑫，温显斌，等.基于颜色特征的图像检索方法研究［Ｊ］. 计算机科学，2012，39（11）：243-245
[2] 郭庆文，王国胤，等.多粒度的图像检索方法研究［Ｊ］.计算机科学，2013，40（1）：298-301
[3] 孙君顶，崔江涛，等.基于颜色和形状特征的彩色图像检索方法［Ｊ］.中国图象图形学报，2004，9（7）：820-827
[4] 宋琳琳，王相海. 小波域颜色和纹理特征提取及图像检索方法研究［Ｊ］.小型微型计算机系统，2009，30（10）：2078-2084
[5] 温超，耿国华. 基于内容图像检索中的“语义鸿沟”问题［Ｊ］. 西北大学学报(自然科学版)，2005，35（5）：536-540
[6] 晁安娜，刘坤．基于卷积神经网络的遥感图像飞机目标识别［Ｊ］.微型机与应用，2017，36（22）：66-69，73．
[7] 李亚飞，董红斌.基于卷积神经网络的遥感图像分类研究［Ｊ］.智能系统学报，DOI：10.11992/tis.201706078.
[8] 黄洁，姜志国，等. 基于卷积神经网络的遥感图像舰船目标检测［Ｊ］.北京航空航天大学学报，2017，43（9）：1841-1848