红外技术2017，Vol.39Issue(8)：728-733,6.
基于深度卷积神经网络的红外场景理解算法
Infrared Scene Understanding Algorithm Based on Deep Convolutional Neural Network
摘要
Abstract
We adopt a deep learning method to implement a semantic infrared image scene understanding. First, we build an infrared image dataset for the semantic segmentation research, consisting of four foreground object classes and one background class. Second, we build an end-to-end infrared semantic segmentation framework based on a deep convolutional neural network connected to a conditional random field refined model. Then, we train the model. Finally, we evaluate and analyze the outputs of the algorithm framework from both the visible and infrared datasets. Qualitatively, it is feasible to adopt a deep learning method to classify infrared images on a pixel level, and the predicted accuracy is satisfactory. We can obtain the features, classes, and positions of the objects in an infrared image to understand the infrared scene semantically.关键词
红外图像/红外场景/语义分割/卷积神经网络Key words
infrared images/infrared scene/semantic segmentation/convolutional neural network分类
信息技术与安全科学引用本文复制引用
王晨,汤心溢,高思莉..基于深度卷积神经网络的红外场景理解算法[J].红外技术,2017,39(8):728-733,6.基金项目
国家"十二五"国防预研项目,上海物证重点实验室基金(2011xcwzk04),中国科学院青年创新促进会资助(2014216). （2011xcwzk04）
