深度学习在高分辨率遥感图像语义分割中的算法研究

Algorithm Research of Deep Learning in High-Resolution Remote Sensing Image Semantic Segmentation

遥感图像语义分割是为遥感图像分配像素级语义标签的计算机视觉任务。随着传感器技术以及深度学习的发展,深度学习算法在精度与速度上远超传统算法。其中,基于深度学习的高分辨率遥感图像语义分割的算法成为众多学者的主要研究方向之一。本文主要针对深度学习在遥感图像语义分割中的相关算法以及网络结构进行介绍。首先介绍语义分割CNN网络,其次分别从三个方面对高分辨率遥感图像语义分割算法进行阐述:一是结合多尺度、多阶段、上下文聚合策略,二是在语义分割之后采用后处理技术,三是结合注意力机制。随后介绍经典数据集,最后对未来深度学习在高分辨率遥感图像语义分割中的算法的发展进行总结与展望。

Remote sensing image semantic segmentation is a computer vision task to assign pixel level semantic labels to remote sensing images. With the development of sensor technology and deep learning, deep learning algorithm is far superior to traditional algorithms in accuracy and speed. Among them, the algorithm of high-resolution remote sensing image semantic segmentation based on deep learning has become one of the main research directions of many scholars. This paper mainly introduces the related algorithms and network structure of deep learning in remote sensing image semantic segmentation. First, the semantic segmentation CNN network is introduced, and then the semantic segmentation algorithm of high-resolution remote sensing images is described from three directions: first, combining multi-scale, multi-stage, context aggregation strategies, second, using post-processing technology after semantic segmentation, and third, combining attention mechanism. Then we introduce the classical datasets, and finally, summarize and prospect the development of deep learning algorithm in the high-resolution remote sensing images semantic segmentation in the future.