中国海洋大学学报（自然科学版）2017，Vol.47Issue(7)：142-148,7.DOI:10.16441/j.cnki.hdxb.20150080
基于自适应幅度谱分析的显著目标检测
Salient Object Detection Based on Adaptive Amplitude Spectrum Analysis
摘要
Abstract
Since HFT model may fail to highlight the salient objects uniformly and lose some meaningful saliency information,this paper proposes a salient object detection method via adaptive amplitude spectrum analysis,which combines with the scale-space analysis.In order to pop out the salient region more uniformly,the algorithm smoothes the spikes in the amplitude spectrum using the specific relation between the size of salient region and the scale of filter in frequency domain.We also introduce the Gaussian model of center bias to combine different saliency maps with more meaningful saliency information.The performance evaluation on two popular benchmark data sets validates that our method gets higher precision and better recall and so outperforms the existing spectral saliency models.关键词
显著性检测/显著目标检测/幅度谱/最优尺度Key words
saliency detection/salient object detection/amplitude spectrum analysis/optimal scale分类
信息技术与安全科学引用本文复制引用
于芝涛,姬婷婷,程孝龙,赵红苗,姬光荣,郑海永..基于自适应幅度谱分析的显著目标检测[J].中国海洋大学学报（自然科学版）,2017,47(7):142-148,7.基金项目
国家自然科学基金项目(61301240、61271406)资助 Supported by the National Natural Science Foundation of China (61301240,61271406) （61301240、61271406）
