计算机应用 ›› 2017, Vol. 37 ›› Issue (7): 2071-2077.DOI: 10.11772/j.issn.1001-9081.2017.07.2071

• 计算机视觉与虚拟现实 • 上一篇    下一篇

基于倒数函数谱残差的显著对象探测和提取方法

陈文兵, 鞠虎, 陈允杰   

  1. 南京信息工程大学 数学与统计学院, 南京 210044
  • 收稿日期:2016-12-16 修回日期:2017-03-02 出版日期:2017-07-10 发布日期:2017-07-18
  • 通讯作者: 鞠虎
  • 作者简介:陈文兵(1964-),男,安徽东至人,副教授,硕士,主要研究方向:计算数学、模式识别、图像处理;鞠虎(1992-),男,江苏泰兴人,硕士研究生,主要研究方向:图像处理;陈允杰(1980-),男,江苏南京人,副教授,博士,主要研究方向:计算数学、模式识别、图像处理。
  • 基金资助:
    国家自然科学基金资助项目(61672291);北极阁基金资助项目(BJG201504)。

Salient object detection and extraction method based on reciprocal function and spectral residual

CHEN Wenbing, JU Hu, CHEN Yunjie   

  1. School of Mathematics and Statistics, Nanjing University of Information Science & Technology, Nanjing Jiangsu 210044, China
  • Received:2016-12-16 Revised:2017-03-02 Online:2017-07-10 Published:2017-07-18
  • Supported by:
    This work is partially supported by the National Natural Science Foundation of China (61672291), the Beijige Foundation (BJG201504).

摘要: 针对"中心-周围"的显著对象探测方法频繁出现探测或提取对象不完整、边界不平滑以及其9级金字塔下采样的冗余问题,提出一种基于倒数函数-谱残差(RFSR)的显著对象探测方法。首先,利用灰度图像与其对应的高斯低通滤波的差代替"中心-周围"方法中灰度图像标准化,并减少高斯金字塔至6级以降低冗余;其次,利用倒数函数滤波器代替Gabor滤波器提取局部方向信息;接着,利用谱残差方法提取图像的谱特征;最后,将这三个特征经过适当融合生成最终显著图。在两个常用基准数据集上的实验结果表明,所提方法在准确率(precision)、召回率(recall)及F-measure等指标上均比"中心-周围"及谱残差模型有明显提高,其为进一步图像分析、对象识别及基于显著视觉关注的图像检索等理论及应用研究奠定了基础。

关键词: 显著对象, 显著性区域, 特征提取, 倒数函数, 显著图

Abstract: To solve the problems of "center-surround" salient object detection and extraction method, such as incomplete object detected or extracted, not smooth boundary and redundancy caused by down-sampling 9-level pyramid, a salient object detection method based on Reciprocal Function and Spectral Residual (RFSR) was proposed. Firstly, the difference between the intensity image and its corresponding Gaussian low-pass one was used to substitute the normalization of the intensity image under "center-surround" model, meanwhile the level of Gaussian pyramid was further reduced to 6 to avoid redundancy. Secondly, a reciprocal function filter was used to extract local orientation information instead of Gabor filter. Thirdly, spectral residual algorithm was used to extract spectral feature. Finally, three extracted features were properly combined to generate the final saliency map. The experimental results on two mostly common benchmark datasets show that compared with "center-surround" and spectral residual models, the proposed method significantly improves the precision, recall and F-measure, furthermore lays a foundation for subsequent image analysis, object recognition, visual-attention-based image retrieval and so on.

Key words: salient object, saliency region, feature extraction, reciprocal function, saliency map

中图分类号: