基于DCNN的图像语义分割综述

发布时间：2018-01-17 14:22

本文关键词：基于DCNN的图像语义分割综述　出处：《北京交通大学学报》2016年04期 　论文类型：期刊论文

【摘要】：图像的语义分割是计算机视觉中重要的基本问题之一,其目标是对图像的每个像素点进行分类,将图像分割为若干个视觉上有意义的或感兴趣的区域,以利于后续的图像分析和视觉理解.近年来,深度卷积神经网络(Deep Convolutional Neural Network,DCNN)的出现,极大地推动了语义分割的发展.本文从语义分割的基本定义出发,对语义分割中存在的困难和挑战进行了分析和描述.总结了目前用于评测语义分割算法的典型数据库,并以PASCAL VOC数据库为主线对近年来基于DCNN的语义分割算法进行了梳理和总结.最后对语义分割未来的研究重点进行了探讨和预测.
[Abstract]:Image semantic segmentation is one of the important basic problems in computer vision. Its goal is to classify each pixel of the image and divide the image into several visually meaningful or interesting areas. In order to facilitate subsequent image analysis and visual understanding, in recent years, deep convolution neural network deep Convolutional Neural Network. The emergence of DCNN has greatly promoted the development of semantic segmentation. This paper starts from the basic definition of semantic segmentation. The difficulties and challenges in semantic segmentation are analyzed and described, and the typical databases used to evaluate semantic segmentation algorithms are summarized. Taking PASCAL VOC database as the main line, the semantic segmentation algorithm based on DCNN in recent years is combed and summarized. Finally, the future research focus of semantic segmentation is discussed and forecasted.
【作者单位】：北京交通大学计算机与信息技术学院;
【基金】：国家科技重大专项资金资助(2016YFB0800404)
【分类号】：TP391.41
【正文快照】： 图像分类、物体检测和图像语义分割是计算机视觉的三大核心研究问题.其中,图像语义分割任务最具有挑战性.图像语义分割融合了传统的图像分割和目标识别两个任务,其目的是将图像分割成几组具有某种特定语义含义的像素区域,并识别出每个区域的类别,最终获得一幅具有像素语义标注

【相似文献】