CNN特征与BOF相融合的水下目标识别算法

doi:10.6040/j.issn.1672-3961.0.2017.385

摘要/Abstract

摘要：

为了改善作为低级表示的尺度不变特征变换(scale invariant feature transform, SIFT)匹配常出现的没有足够特征来防止假匹配的问题,提出在传统方法“词袋”(bag of features, BOF)算法中融合具有较好语义分割能力的卷积神经网络(convolution neural network, CNN)特征来提高识别率的方法。利用ImageCLEF网站的LifeCLEF鱼类视频,制作目标图像数据库。在caffe平台的Alexnet模型进行卷积神经网络的训练,提取图像库和查询图像的特征。利用训练好的CNN特征在Matlab软件进行识别试验验证,计算汉明距离来验证匹配效果。改变参数值来观察不同汉明距离阈值对水下目标识别结果的影响。自制图像库的试验表明,融合深度学习的特征可以有效提高BOF算法的水下目标识别率,对汉明距离阈值的选择需要根据实际情况选择合适的参数。

关键词: 水下目标识别, BOF, SIFT匹配, 卷积神经网络, 汉明距离

Abstract:

In order to prevent false matching problems of scale invariant feature transform (SIFT) matching as a low-level representation for lack of sufficient features, an improved bag of features (BOF) algorithm method combined with the convolution neural network (CNN) features was proposed, which had better semantic segmentation ability to enhance the recognition rates. The LifeCLEF fish video on ImageCLEF website was used to create our own target image databases. Convolution neural network was trained in the Alexnet architecture of caffe, and the features of image databases and query images were extracted. The trained CNN features were simulated in Matlab, and the hamming distance was calculated to verify the matching effect. In addition, the parameter values were changed to test the effect of different Hamming distance thresholds on target recognition results. The experiment of self-made image databases showed that the fusion of depth learning features could effectively improve the underwater target recognition rates of BOF algorithm, and the selection of Hamming distance thresholds required selecting the appropriate parameters according to the actual situation.

Key words: underwater target recognition, bag of features, scale invariant feature transform matching, convolution neural networks, Hamming distance

中图分类号:

TP391

权稳稳,林明星. CNN特征与BOF相融合的水下目标识别算法[J]. 山东大学学报 (工学版), 2019, 49(1): 107-113.

Wenwen QUAN,Mingxing LIN. Algorithm of underwater target recognition based on CNN features with BOF[J]. Journal of Shandong University(Engineering Science), 2019, 49(1): 107-113.

图/表 6

图1

图2

表1

表2

图3

图4

参考文献 22

1	JAFFE J S , MOORE K D , MCLEAN J , et al. Underwater optical imaging: status and prospects[J]. Oceanography, 2001, 14 (3): 66- 76.
2	王士龙, 徐玉如, 万磊, 等. 基于边界矩和改进FCM聚类的水下目标识别[J]. 系统工程理论与实践, 2012, 32 (12): 2809- 2815.
	WANG Shilong , XU Yuru , WANG Lei , et al. Underwater targets recognition based on contour moment and modified FCM algorithm[J]. System Engineering Theory and Practice, 2012, 32 (12): 2809- 2815.
3	FATAN M , DALIRI M R , MOHAMMAD S A , et al. Underwater cable detection in the images using edge classification based on texture information[J]. Measurement, 2016, 91, 309- 317. doi: 10.1016/j.measurement.2016.05.030
4	乔曦.基于水下机器视觉的海参实时识别研究[D].北京:中国农业大学, 2017.
	QIAO Xi. Sea cucumber identification in real-time based on underwater machine vision technique[D]. Beijing: China Agricultural University, 2017.
5	LOWE D G . Distinctive image features from scale-invariant key points[J]. International Journal of Computer Vision, 2004, 60 (2): 91- 110. doi: 10.1023/B:VISI.0000029664.99615.94
6	BAY H, TUYTELAARS T, VAN GOOL L. SURF: speeded up robust features[C]//Computer Vision-ECCV 2006. Graz, Austria: Springer Berlin Heidelberg, 2006: 404-417.
7	ZHENG Z Z , YUN Z , YAN L X . Global and local exploitation for saliency using bag-of-words[J]. IET Computer Vision, 2014, 8 (4): 299- 304. doi: 10.1049/iet-cvi.2013.0132
8	KRIZHEVSKY A, SUTSKEVER I, HINTON G E. Imagenet classification with deep convolutional neural networks[C]//Proceedings of the Conference on Neural Information Processing Systems. Lake Tahoe, Spain: IEEE, 2012: 1097-1105.
9	REN S , HE K , GIRSHICK R , et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39 (6): 1137- 1149. doi: 10.1109/TPAMI.2016.2577031
10	LECUN Y , BOTTOU L , BENGIO Y , et al. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86 (11): 2278- 2324. doi: 10.1109/5.726791
11	SZEGEDY C, LIU W, JIA Y, et al. Going deeper with convolutions[C]//Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston, USA: IEEE, 2015: 1-9.
12	HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition[C]//CVPR 16: Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Los Vegas, USA: IEEE, 2016: 770-778.
13	LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[C]//Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Boston, USA: IEEE, 2015: 1337-1342.
14	EKANAYAKE J, PALLICKARE S. Map reduce for data intensive scientific analysis[C]//IEEE Science. Piscatway, USA: IEEE, 2008: 277-284.
15	LAZEBNIK S, SCHMID C, PONCE J. Video google: a text retrieval approach to object matching in videos[C]//2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. New York, USA: IEEE, 2006: 2169-2178.
16	SIVIC J, ZISSERMAN A. Beyond bags of features: spatial pyramid matching for recognizing natural scene categories[C]//Proceedings of Ninth IEEE International Conference on Computer Vision. Nice, France: IEEE, 2003: 1470-1477.
17	JEGOU H , DOYZE M , SCHMID C . Improving bag of features for large scale image search[J]. International Journal of Computer Vision, 2010, 87 (3): 316- 336. doi: 10.1007/s11263-009-0285-2
18	ZHANG G X , ZENG Z , ZHANG S W , et al. SIFT matching with CNN evidences for particular object retrieval[J]. Neurocomputing, 2017, 238, 399- 409. doi: 10.1016/j.neucom.2017.01.081
19	RUSSAKOVSKY O , DENG J , SU H , et al. Imagenet large scale visual recognition challenge[J]. International Journal of Computer Vision, 2014, 115 (3): 211- 252.
20	CAMPBELL A T , MEER H G D , KOUNAVIS M E , et al. A survey of programmable networks[J]. ACM Computer Communications Review, 1999, 29 (2): 7- 23. doi: 10.1145/505733
21	LI X, SHANG M, QIN H W, et al. Fast Accurate Fish detection and recognition of Underwater Images with Fast R-CNN[C]//Oceans. Washington, USA: IEEE, 2015: 1-5.
22	JIA Y, SHELHAMER E, DONAHUE J, et al. Caffe: convolutional architecture for fast feature embedding[C]//Proceedings of the 22nd ACM international conference on Multimedia. Orlando, USA: arXiv, 2014: 675-678.

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed

类别	训练	验证	测试	训练验证总计
短身光鳃雀鲷弓月蝴蝶鱼黑带椒雀鲷三带圆雀鲷网纹圆雀鲷小高鳍刺尾鱼	2 310 926 2 300 1 516 2 303 650	289 116 288 190 288 82	288 116 288 189 288 82	2 599 1 042 2 588 1 706 2 591 732
总计	10 005	1 253	1 251	11 258

%
方法	BOF	CNN	BOF+Conv5	BOF+Fc6	BOF+Fc7
短身光鳃雀鲷弓月蝴蝶鱼黑带椒雀鲷三带圆雀鲷网纹圆雀鲷小高鳍刺尾鱼	74.0 70.7 71.6 76.5 69.7 69.5	74.3 73.8 72.6 75.8 74.2 72.4	86.8 85.5 83.1 87.3 82.8 82.1	86.7 85.3 86.2 87.0 84.5 84.8	76.0 72.8 74.6 76.8 71.3 74.2
平均识别率	72.2	73.9	84.7	85.9	74.3

[1]	梁蒙蒙,周涛,夏勇,张飞飞,杨健. 基于PSO-ConvK卷积神经网络的肺部肿瘤图像识别[J]. 山东大学学报 (工学版), 2018, 48(5): 77-84.
[2]	张璞,刘畅,王永. 基于特征融合和集成学习的建议语句分类模型[J]. 山东大学学报 (工学版), 2018, 48(5): 47-54.
[3]	何正义,曾宪华,郭姜. 一种集成卷积神经网络和深信网的步态识别与模拟方法[J]. 山东大学学报(工学版), 2018, 48(3): 88-95.
[4]	谢志峰,吴佳萍,马利庄. 基于卷积神经网络的中文财经新闻分类方法[J]. 山东大学学报(工学版), 2018, 48(3): 34-39.
[5]	赵彦霞, 王熙照. 基于SVD和DCNN的彩色图像多功能零水印算法[J]. 山东大学学报(工学版), 2018, 48(3): 25-33.
[6]	徐姗姗,刘应安*,徐昇. 基于卷积神经网络的木材缺陷识别[J]. 山东大学学报(工学版), 2013, 43(2): 23-28.