深度卷积神经网络的显著性检测

被引:54
作者
李岳云
许悦雷
马时平
史鹤欢
机构
[1] 空军工程大学航空航天工程学院
关键词
显著性检测; 超像素; 卷积神经网络; 条件随机场;
D O I
暂无
中图分类号
TP391.41 [];
学科分类号
080203 ;
摘要
目的显著性检测问题是近年来的研究热点之一,针对许多传统方法都存在着特征学习不足和鲁棒检测效果不好等问题,提出一种新的基于深度卷积神经网络的显著性检测模型。方法首先,利用超像素的方法聚类相似特征的像素点,仿人脑视皮层细胞提取目标边缘,得到区域和边缘特征。然后,通过深度卷积神经网络学习图像的区域与边缘特征,获取相应的目标检测显著度置信图。最后,将深度卷积神经网络输出的置信度融入到条件随机场,求取能量最小化,实现显著性与非显著性判别,完成显著性检测任务。结果在两个常用的视觉检测数据库上进行实验,本文算法的检测精度与当前最好的方法相比,在MSAR数据库上检测精度相对提升大约1.5%,在Berkeley数据库上提升效果更加明显,达到了5%。此外,无论是自然场景还是人工建筑场景、大目标与小目标,检测的效果都是最好的。结论本文融合多特征的深度学习方法与单一浅层人工特征的方法相比更有优势,它避免了手工标定特征所带来的不确定性,具有更好的鲁棒性与普适性,从主观视觉愉悦度和客观检测准确度两方面说明了算法的有效性。
引用
收藏
页码:53 / 59
页数:7
相关论文
共 14 条
  • [1] Visual cortex inspired vertex and corner detection. A. B. Csapo,A. Roka,P. Baranyi. 2006 IEEE International Conference on Mechatronics . 2006
  • [2] Saliency Detection via Absorbing Markov Chain. Jiang B,Zhang L H,Lu H C. International Conference on Computer Vision . 2013
  • [3] Visual saliency detection based on Bayesian model. XIE Y,LU H. Proceedings of IEEE International Conference on Image Processing . 2011
  • [4] Saliency detection based on integration of boundary and soft-segmentation. SUN J,LU H,LI S. Proceedings of IEEE International Conference on Image Processing . 2012
  • [5] SLIC superpixels compared to state-of-the-art superpixel methods. Achanta, Radhakrishna,Shaji, Appu,Smith, Kevin,Lucchi, Aurelien,Fua, Pascal,Süsstrunk, Sabine. IEEE Transactions on Pattern Analysis and Machine Intelligence . 2012
  • [6] Model of saliency-based visual attention for rapid scene analysis. Itti, Laurent,Koch, Christof,Niebur, Ernst. IEEE Transactions on Pattern Analysis and Machine Intelligence . 1998
  • [7] Bayesian saliency via low and mid level cues. Xie Y L,Lu H C,Yang M H. Image Processing Europe . 2013
  • [8] Saliency detection based on feature learning using deep boltzmann machines. Wen S F,Han J W,Zhang D W,et al. Proceedings of IEEE International Conference on Multimedia and Expo . 2014
  • [9] Vehicle detection in satellite images by parallel deep convolutional neural networks. Chen X Y,Xiang S M,Liu C L,et al. Proceedings of the 2nd IAPR Asian Conference on Pattern Recognition . 2013
  • [10] Salient object contour detection based on boundary similar region. Zhan J,Hu B. Proceedings of the 4th International Conference on Digital Home . 2012