期刊文献+

基于多尺度局部极值和边缘检测的目标性算法 被引量:4

Objectness Algorithm Based on Multi-scale Local Extremum and Edge Detection
在线阅读 下载PDF
导出
摘要 目标性作为目标检测的预处理算法,用于高效提取少量可靠的目标潜在区域,可替代针对复杂特征的多尺度滑动窗的分析方式,达到提升目标检测效率的目的。该文提出了一种基于多尺度局部极值和边缘检测的目标性算法。首先,基于原始图像的多尺度梯度特征,在不同尺度下利用均值滤波得到梯度强度的局部极值,并在原始图像上还原出初始目标潜在区域;然后,通过提取图像的边缘特征,计算初始目标潜在区域的目标性得分值;最后,对得分值进行尺度加权,并结合非极大值剔除冗余区域,最终输出少量可靠的目标潜在区域。通过PASCAL VOC和ILSVRC2014数据库的实验对比,该算法给定1000个候选区时在PASCAL VOC和ILSVRC2014分别达到97%和98%以上的召回率,同时有效地提升了首框召回率。 Aiming to promote the efficiency of object detection, the objectness was introduced to pre-analyze the potential location of objects instead of the sliding window strategy with the complex features. Based on the multi-scale local extremum and edge detection, an objectness method was proposed to leverage the efficiency. First, mean filter was used to obtain the local extremum on the multi-scale gradient feature maps. According to these local extremums, coarse object proposals are extracted on the original RBG image. Second, the objectness score of each coarse object proposal was calculated based on the edge feature. Finally, redundant proposals were removed by non-maximum suppression with the scale information and the objectness score. The comparative experiment results in the public datasets (PASCAL VOC and ILSVRC 2014) demonstrated that the recall rate of our method achieved over 97% (PASCAL) and 98% (ILSVRC 2014) with 1000 proposals respectively. Furthermore, the recall rate of the top one proposal was improved too.
出处 《信号处理》 CSCD 北大核心 2016年第8期911-921,共11页 Journal of Signal Processing
基金 国家高技术研究发展计划(863计划)(2015AA015904) 国家自然科学青年基金项目(61502187) 娄底市科技计划项目(2015ZHJHK017) 湖南人文科技学院校级青年基金项目(2015QN03) 湖南省教育厅资助科研项目(14C0599 14C0597)
关键词 目标检测 多尺度 局部极值 边缘检测 目标性 object detection multi-scale local extremum edge detection objectness
作者简介 方智文男,1983年生,湖南长沙人,湖南人文科技学院能源与机电工程学院讲师。2004年于北京航空航天大学获得电气工程及其自动化专业学士学位,2008年于北京航空航天大学获得检测自动化专业硕士学位,2012年至今于华中科技大学自动化学院攻读博士学位。目前主要研究方向为模式识别、机器学习。E-mail:fzw310@hust.edu.cn 曹治国(通信作者)男,1964年生,湖北武汉人,博士,华中科技大学自动化学院教授。1985年于电子科技大学获得电子工程专业学士学位,1990年于电子科技大学获得通信与电子系统专业硕士学位,2001年于华中科技大学获得模式识别与智能系统专业博士学位。目前主要研究方向为自动目标识别、机器学习。E—mail:zgcao@hust.edu.cn 肖阳男,1982年生,湖北武汉人,博士,华中科技大学自动化学院讲师,2004年于华中科技大学电信系与外语学院获得电信与英语双学士学位,2007年获得模式识别与智能系统硕士学位,2011年获得控制科学与工程博士学位,2012年至2014年在新加坡南洋理工大学从事博士后研究工作,目前研究方向为计算机视觉、机器学习与人机交互。E—mail:Yang_Xiao@hust.edu.cn
  • 相关文献

参考文献23

  • 1胡正平,周爽.曲率-HOG目标检测算法研究[J].信号处理,2013,29(11):1470-1475. 被引量:4
  • 2Dalal N, Triggs B. Histograms of oriented gradients for hu- man detection[ C]//Computer Vision and Pattern Recogni- tion, 2005 IEEE Conference on. IEEE 2005: 886-893.
  • 3Felzenszwalb P, MeAllester D, Ramanan D. A discrimina- tively trained, multiscale, deformable part model [ C ]// Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on. IEEE, 2008:1-8.
  • 4Yan J, Lei Z, Wen L, et al. The fastest deformable part model for object detection [ C ] //Computer Vision and Pattern Recognition (CVPR), 2014 IEEE Conference on. IEEE, 2014: 2497-2504.
  • 5Song H O, Girshiek R, Zickler S, et al. Generalized Sparselet Models for Real-Time Muhiclass Object Recog- nition[ J ]. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 2015, 37(5) : 1001-1012.
  • 6Alexe B, Deselaers T, Ferrari V. What is an object? [C]// Computer Vision and Pattern Recognition, 2010 IEEE Conference on. IEEE, 2010: 73-80.
  • 7Cheng M M, Zhang Z, Lin W Y, et al. BING: Binarized normed gradients for objectness estimation at 300fps[ C]// Computer Vision and Pattern Recognition, 2014 IEEE Conference on. IEEE, 2014: 3286-3293.
  • 8Zitnick C L, Dollar P. Edge boxes: Locating object pro-posals from edges[ C ] //European Conference on Comput- er Vision (ECCV2014). Springer International Publish- ing, 2014 : 391-405.
  • 9Lu C, Liu S, Jia J, et al. Contour Box: Rejecting Object Proposals Without Explicit Closed Contours [ C ]//J Com- puter Vision and Pattern Recognition, 2015 IEEE Confer- ence on. IEEE, 2015 : 2021-2029.
  • 10Arbelaez P, Pont-Tuset J, Barron J, et al. Muhiscale com- binatorial grouping [ C ] //Computer Vision and Pattern Recognition, 2014 IEEE Conference on. IEEE, 2014: 328 -335.

二级参考文献11

  • 1Dalai N, Triggs B. Histograms of oriented gradients for human detection[ C]. 2005 IEEE Computer Society Con- ference on Computer Vision and Pattern Recognition,2005, 1:886-893.
  • 2Kojima A, Sasano Y, Kise K. Object recognition from human actions using N-grams of HOG features [ J ]. ICIC Express Letters, 2012, 6(5):1155-1160.
  • 3Takahashi K, Kuriya Y, Morie T. Bicycle detection using pedaling movement by spatiotemporal Gabor filtering[ J]. International Journal of Innovative Computing, Informa- tion and Control, 2012, 8(6) :4059-4070.
  • 4Semeikina E V, Yurin D V, Krylov A. S, et al. Scale- space line curvature estimation for straight tine and circle detection[ J ]. Pattern Recognition and Image Analysis, 2012, 22(2) :360-370.
  • 5Fabian T, Erhardt B. Accurate eye centre localisation by means of gradients [ C ]. Proceedings of the International Conference on Computer Vision Theory and Application. Vilamoura, Algarve, Portugal, 2011 : 125-130.
  • 6Kawamura K, Ishii D, Watanabe H. Automatic scale de-tection for contour fragment based on difference of curva- ture [ J ]. IEICE Transactions on Information and Sys- tems, 2011, E94-D(10) : 1998-2005.
  • 7Ludwig J O, David D, Valter G, et al. Trainable classifi- er-fusion schemes: An application to pedestrian detection [ C ]. IEEE Conference on Intelligent Transportation Sys- tems, Proceedings, 2009:432-437.
  • 8闻帆,屈桢深,闫纪红.基于CKPCA-HOG和支持向量机的运动目标分类算法[J].哈尔滨工程大学学报,2011,32(5):643-649. 被引量:6
  • 9胡正平,杨建秀.HOG特征混合模型结合隐SVM的感兴趣目标检测定位算法[J].信号处理,2011,27(8):1206-1212. 被引量:8
  • 10姚雪琴,李晓华,周激流.基于边缘对称性和HOG的行人检测方法[J].计算机工程,2012,38(5):179-182. 被引量:22

共引文献3

同被引文献30

引证文献4

二级引证文献12

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部