基于多级深度特征与随机游走的显著性检测  

Saliency Detection Based on Multi-Level Deep Features and Random Walk

作　　者：崔冬[1,2] 王明 李刚 顾广华[1,2] 李海涛 CUI Dong;WANG Ming;LI Gang;GU Guanghua;LI Haitao(School of Information Science and Engineering,Yanshan University,Qinhuangdao 066004,Hebei,China;Hebei Provincial Key Laboratory of Information Transmission and Signal Processing,Qinhuangdao 066004,Hebei,China)

机构地区：[1]燕山大学信息科学与工程学院,河北秦皇岛066004 [2]河北省信息传输与信号处理重点实验室,河北秦皇岛066004

出　　处：《华南理工大学学报（自然科学版）》2020年第8期49-55,共7页Journal of South China University of Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(61303128);河北省自然科学基金资助项目(F2017203169,F2018203239);河北省科技计划项目(18210336)。

摘　　要：为了解决图像显著性检测中传统方法特征学习不全面、复杂场景下显著区域凸出不明显的问题,提出了一种基于多级深度特征和随机游走的显著性检测算法。首先,利用全卷积神经网络,结合深层和浅层卷积特征信息对图像进行多级卷积深度特征提取;然后,对图像进行超像素分割,将提取的深度卷积特征分配给相应的超像素,构建特征矩阵;最后,通过正则化随机游走排序模型生成最终的显著图。在ECSSD和DUT-OMRON数据库上的实验结果表明,与6种具有代表性的显著性检测算法相比,文中算法的准确性和F值具有一定的优势。In order to solve the problems of incomplete feature learning and unobvious salient regions in complex scenes for the traditional saliency detection methods,a saliency detection algorithm based on multi-level deep features and random walk was proposed.Firstly,the fully convolutional networks(FCN)was used to perform multi-level convolution deep feature extraction on the image by combining the deep and shallow feature information.Se-condly,superpixels segmentation on the image was carried out and the extracted deep convolution features were assigned to the corresponding superpixels to construct a feature matrix.Finally,the final saliency map was generated by regularizing the random walk ranking model.The experimental results on the ECSSD and DUT-OMRON databases show that the proposed method has certain advantages in accuracy and F value,compared with the six representative saliency detection algorithms.

关 键 词：显著性检测 多级深度特征 特征提取 随机游走 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]