液晶与显示2016，Vol.31Issue(9)：913-920,8.DOI:10.3788/YJYXS20163109.0913
基于空频域结合的显著目标检测
Saliency detection based on frequency domain combined with spatial domain
摘要
Abstract
In order to detect saliency goals more accurately,saliency detection method for single do-main is easy to cause the disadvantage of useful information loss.So a model is proposed by combining global information of the frequency and local information of the spatial to obtain more comprehensive saliency target information.The model is necessary.Through fusing the useful information of two do-mains,the final saliency detection combines the detail of the local information with the integrity of the global information.This model can not only strengthen saliency regions,but also can suppress non-significant parts in a complex background.Saliency detection which is obtained by the new model are clear and complete,and it can inhibit the background information to highlight the saliency regions. Experiments showed that the proposed model relative to the other single domain model can extract significant target area better.Compared with single domain model,the new model improved 9.5%relatively to spatial domain model,and improved 6.3% relatively to frequency domain model.关键词
计算机视觉/显著性目标检测/空频域结合/局部信息/全局信息/融合显著图Key words
computer vision/saliency detection/j oint spatial-frequency/local/global/fusion分类
信息技术与安全科学引用本文复制引用
杜慧,张涛,张叶,穆绍硕..基于空频域结合的显著目标检测[J].液晶与显示,2016,31(9):913-920,8.基金项目
吉林省重大科技攻关项目(No．11DGG001) （No．11DGG001）
国家自然科学基金青年资助项目(No．60902067)Supported by the National Natural Science Foundation of China (No．60902067) （No．60902067）
Major Scientific and Technology Projects in Jilin Province (No．11DGG001) （No．11DGG001）
