摘要
图像语义分割是对图像中的每个像素点进行分类,将图像中的前景和背景区分并且识别出每个前景的类别。随着深度学习技术的发展,传统图像语义分割方法在分割精度和分割速度上已经彻底被超越。针对深度学习图像语义分割方法研究现状进行综述,对近年来国内外基于深度学习图像语义分割方法主要思想、优缺点进行了分析和总结。提出了该领域目前存在的问题,对将来的发展进行总结和展望。
In order to distinguish the foreground and the background in an image and identify the category of each foreground,each pixel point in an image was classified by image semantic segmentation.With the development of deep learning techniques,traditional image semantic segmentation methods have been completely surpassed in segmentation precision and speed.The research status of deep learning image semantic segmentation methods has been reviewed and the main ideas,advantages and disadvantages of deep learning image semantic segmentation methods at home and abroad in recent years have been analyzed and summarized.The existing problems in this field are put forward,and the future development is summarized and prospected.
作者
李新叶
宋维
LI Xin-ye;SONG Wei(Department of Electronic and Communication Engineering,North China Electric Power University,Baoding 071003,China)
出处
《科学技术与工程》
北大核心
2019年第33期21-27,共7页
Science Technology and Engineering
关键词
图像语义分割
像素分类
深度学习
image semantic segmentation
pixel classification
deep learning
作者简介
第一作者:李新叶(1969—),女,汉族,河北平山人,博士,副教授。研究方向:计算机视觉、模式识别、信息检索等。E-mail:lxyyjh@126.com。
