多对象图像数据集建立及显著性检测算法评估

doi:10.11772/j.issn.1001-9081.2015.09.2624

计算机应用 ›› 2015, Vol. 35 ›› Issue (9): 2624-2628.DOI: 10.11772/j.issn.1001-9081.2015.09.2624

多对象图像数据集建立及显著性检测算法评估

郑斌¹, 牛玉贞^1,2, 柯玲玲¹

1. 福州大学数学与计算机科学学院, 福州 350116;
2. 福建省网络计算与智能信息处理重点实验室(福州大学), 福州 350116

收稿日期:2015-04-30 修回日期:2015-06-29 出版日期:2015-09-10 发布日期:2015-09-17
通讯作者: 牛玉贞(1982-),女,山东济南人,教授,博士,CCF会员,主要研究方向:图像和视频处理、计算机视觉,yuzhenniu@gmail.com
作者简介:郑斌(1990-),男,福建厦门人,硕士研究生,主要研究方向:图像和视频处理、计算机视觉;柯玲玲(1991-),女,福建莆田人,硕士研究生,主要研究方向:图像和视频处理、计算机视觉。
基金资助:
国家自然科学基金资助项目(61300102);福建省自然科学基金(杰青)资助项目(2015J06014);福建省自然科学基金(面上)资助项目(2014J01233)。

New multi-object image dataset construction and evaluation of visual saliency analysis algorithm

ZHENG Bin¹, NIU Yuzhen^1,2, KE Lingling¹

1. College of Mathematics and Computer Science, Fuzhou University, Fuzhou Fujian 350116, China;
2. Fujian Key Laboratory of Network Computing and Intelligent Information Processing (Fuzhou University), Fuzhou Fujian 350116, China

Received:2015-04-30 Revised:2015-06-29 Online:2015-09-10 Published:2015-09-17

摘要/Abstract

摘要： 图像视觉显著性检测算法在已有数据集上已经取得很好的结果,但是目前的多个数据集存在两个严重的问题:首先,数据集中的图像以只包含一个显著对象的图像为主;其次,在建立显著对象标注结果的过程中,忽略了用户对同一幅图像中包含的多个显著对象的不同认知。上述问题导致了在已有数据集上对显著性检测算法进行评估,不能体现算法在实际应用中的真实效果。为此,提出体现用户认知的多显著对象图像标注方法,首先设计并实现辅助软件,收集用户对各显著对象的重要程度的认知情况,包括显著区域与相应的重要程度;然后融合收集的多用户数据,绘制出以灰度图为表现形式的显著对象标注结果,并通过灰度值体现多用户对于每个显著对象的认知情况。基于改进的显著对象标注方法,建立了一个包含1000幅多显著对象图像的数据集,并为每幅图像提供了体现用户认知的显著对象标注结果。对10种具有代表性的显著性检测算法在已有数据集和建立的数据集上的性能进行了比较。实验结果表明,这些显著性检测算法在建立的数据集上的性能有大幅度的降低,例如受试者工作特征曲线下面积(ROC-AUC)评估参数的最大降幅超过了0.5,这证实了已有数据集存在的问题及建立新数据集的需求,同时指出显著性检测算法在处理包含多显著对象的复杂图像上存在的不足。

关键词: 视觉显著性检测, 多对象图像, 数据集, 用户认知, 算法评估

Abstract: Image visual saliency analysis algorithms have achieved satisfactory performance on existing datasets, but these datasets have two major problems. Firstly, most of the images contain only one salient object. Secondly, users' cognition of multiple salient objects in the same image was ignored during building salient objects' ground truth. The above problems result in that the performance of saliency analysis algorithms used in the real applications cannot be reflected by the evaluation on the existing datasets. So in this paper, a novel method of labeling the ground truth of salient objects was proposed. Firstly, a software to collect users' cognition of the important values of multiple salient objects in each image was designed and implemented. Then, according to the collected data from each user, the ground truth map represented as a gray scale image was created by manually labeling the regions covered by the salient objects. The pixel value of each region equals to the collected saliency in the first step. Based on the improved ground truth labeling method, a salient object dataset contains 1000 multi-object images was built. A ground truth map for each image was created to record users' cognition of the objects' saliencies. Then 10 state-of-the-art saliency analysis algorithms on existing datasets and the established dataset were compared. The experimental results show that these algorithms' performances are greatly reduced on the established dataset, such as the Area Under Curve of Receiver-Operating Characteristic (ROC-AUC) has a greatest decline of more than 0.5. The results prove the problems of existing datasets and the demand of building a new dataset, and point out the insufficiency of saliency analysis algorithms on complex images with multiple salient objects.

Key words: visual saliency analysis, multi-object image, dataset, users cognition, algorithms evaluation

中图分类号:

TP391.4

郑斌, 牛玉贞, 柯玲玲. 多对象图像数据集建立及显著性检测算法评估[J]. 计算机应用, 2015, 35(9): 2624-2628.

ZHENG Bin, NIU Yuzhen, KE Lingling. New multi-object image dataset construction and evaluation of visual saliency analysis algorithm[J]. Journal of Computer Applications, 2015, 35(9): 2624-2628.

参考文献

[1] GAO D, HAN S, VASCONCELOS N. Discriminant saliency, the detection of suspicious coincidences, and applications to visual recognition [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2009,31(6):989-1005.
[2] LIANG Z, FU H, CHI Z, et al. Image pre-classification based on saliency map for image retrieval [C]//Proceedings of the 2009 7th International Conference on Information, Communications and Signal Processing. Piscataway: IEEE, 2009:1-5.
[3] HADIZADEH H, BAJIC I V. Saliency-aware video compression [J]. IEEE Transactions on Image Processing, 2014,23(1):19-33.
[4] YAN Q, XU L, SHI J, et al. Hierarchical saliency detection [C]//Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2013:1155-1162.
[5] CHENG M M, ZHANG G, MITRA N J, et al. Global contrast based salient region detection [EB/OL]. [2015-01-02]. http://www.doc88.com/p-9942984819833.html.
[6] ACHANTA R, HEMAMI S, ESTRADA F, et al. Frequency-tuned salient region detection [C]//Proceedings of the 2009 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2009:1597-1604.
[7] ZHU W, LIANG S, WEI Y, et al. Saliency optimization from robust background detection [C]//Proceedings of the 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2014:2814-2821.
[8] JIANG B, ZHANG L, LU H, et al. Saliency detection via absorbing Markov chain [EB/OL]. [2015-01-04]. http://www.doc88.com/p-1932201551743.html.
[9] YANG C, ZHANG L, LU H, et al. Saliency detection via graph-based manifold ranking [C]//Proceedings of the 2013 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2013:3166-3173.
[10] WEI Y, WEN F, ZHU W, et al. Geodesic saliency using background priors [M]//FITZGIBBON A, LAZEBNIK S, PERONA P, et al. Computer Vision-ECCV 2012, LNCS 7574. Berlin: Springer, 2012:29-42.
[11] KIM C, MILANFAR P. Visual saliency in noisy images [J]. Journal of Vision, 2013,13(4):103-104.
[12] HOU X, ZHANG L. Dynamic visual attention: searching for coding length increments [EB/OL]. [2015-01-03]. http://www.researchgate.net/publication/221618957_Dynamic_Visual_Attention_Searching_for_coding_length_increments.
[13] ERDEM E, ERDEM A. Visual saliency estimation by nonlinearly integrating features using region covariances [J]. Journal of Vision, 2013,13(4):103-104.
[14] ZOU W, KPALMA K, LIU Z, et al. Segmentation driven low-rank matrix recovery for saliency detection [C]//Proceedings of the 24th British Machine Vision Conference. [S.l.]: BMVC Press, 2013:1-13.

[1]	肖振远, 王逸涵, 罗建桥, 熊鹰, 李柏林. 基于部分加权损失函数的RefineDet[J]. 计算机应用, 2021, 41(7): 1928-1932.
[2]	廖胜兰, 殷实, 陈小平, 张波, 欧阳昱, 张衡. 面向电力业务对话系统的意图识别数据集[J]. 计算机应用, 2020, 40(9): 2549-2554.
[3]	傅魁, 梁少晴, 李冰. 基于改进的深度Q网络结构的商品推荐模型[J]. 计算机应用, 2020, 40(9): 2613-2621.
[4]	崔鑫, 徐华, 宿晨. 面向不均衡数据集的过抽样算法[J]. 计算机应用, 2020, 40(6): 1662-1667.
[5]	田臣, 周丽娟. 基于带多数类权重的少数类过采样技术和随机森林的信用评估方法[J]. 计算机应用, 2019, 39(6): 1707-1712.
[6]	王莉, 陈红梅, 王生武. 新的基于代价敏感集成学习的非平衡数据集分类方法NIBoost[J]. 计算机应用, 2019, 39(3): 629-633.
[7]	卫鑫, 武淑红, 王耀力. 基于深度卷积长短期记忆网络的森林火灾烟雾检测模型[J]. 计算机应用, 2019, 39(10): 2883-2887.
[8]	甘岚, 郭子涵, 王瑶. 基于径向变换和改进AlexNet的胃肿瘤细胞图像识别方法[J]. 计算机应用, 2019, 39(10): 2923-2929.
[9]	周于皓, 张红玲, 李芳菲, 祁鹏. 局部关注支持向量机算法[J]. 计算机应用, 2018, 38(4): 945-948.
[10]	王林, 张鹤鹤. Faster R-CNN模型在车辆检测中的应用[J]. 计算机应用, 2018, 38(3): 666-670.
[11]	卜令正, 王洪栋, 朱美强, 代伟. 基于改进卷积神经网络的多源数字识别算法[J]. 计算机应用, 2018, 38(12): 3403-3408.
[12]	付眸, 杨贺昆, 吴唐美, 何润, 冯朝胜, 康胜. 基于Spark Streaming的快速视频转码方法[J]. 计算机应用, 2018, 38(12): 3500-3508.
[13]	徐琳琳, 张树美, 赵俊莉. 基于图像的面部表情识别方法综述[J]. 计算机应用, 2017, 37(12): 3509-3516.
[14]	王文朋, 毛文涛, 何建樑, 窦智. 基于深度迁移学习的烟雾识别方法[J]. 计算机应用, 2017, 37(11): 3176-3181.
[15]	靳燕, 彭新光. 多子域隔离学习组合决策用于不均衡样本[J]. 计算机应用, 2016, 36(9): 2475-2480.

多对象图像数据集建立及显著性检测算法评估

New multi-object image dataset construction and evaluation of visual saliency analysis algorithm

PDF

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 15

编辑推荐

Metrics