基于深度卷積神經(jīng)網(wǎng)絡(luò)和二進(jìn)制哈希學(xué)習(xí)的圖像檢索方法
doi: 10.11999/JEIT151346
-
1.
(河南工程學(xué)院計算機(jī)學(xué)院 鄭州 451191) ②(河南圖像識別工程技術(shù)中心 鄭州 450001)
基金項目:
國家自然科學(xué)基金(61301232)
Image Retrieval Based on Deep Convolutional NeuralNetworks and Binary Hashing Learning
-
1.
(Department of Computer Science and Engineering, Henan Institute of Engineering, Zhengzhou 451191, China)
-
2.
(Henan Image Recognition Engineering Center, Zhengzhou 450001, China)
Funds:
The National Natural Science Foundation of China (61301232)
-
摘要: 隨著圖像數(shù)據(jù)的迅猛增長，當(dāng)前主流的圖像檢索方法采用的視覺特征編碼步驟固定，缺少學(xué)習(xí)能力，導(dǎo)致其圖像表達(dá)能力不強(qiáng)，而且視覺特征維數(shù)較高，嚴(yán)重制約了其圖像檢索性能。針對這些問題，該文提出一種基于深度卷積神徑網(wǎng)絡(luò)學(xué)習(xí)二進(jìn)制哈希編碼的方法，用于大規(guī)模的圖像檢索。該文的基本思想是在深度學(xué)習(xí)框架中增加一個哈希層，同時學(xué)習(xí)圖像特征和哈希函數(shù)，且哈希函數(shù)滿足獨立性和量化誤差最小的約束。首先，利用卷積神經(jīng)網(wǎng)絡(luò)強(qiáng)大的學(xué)習(xí)能力挖掘訓(xùn)練圖像的內(nèi)在隱含關(guān)系，提取圖像深層特征，增強(qiáng)圖像特征的區(qū)分性和表達(dá)能力。然后，將圖像特征輸入到哈希層，學(xué)習(xí)哈希函數(shù)使得哈希層輸出的二進(jìn)制哈希碼分類誤差和量化誤差最小，且滿足獨立性約束。最后，給定輸入圖像通過該框架的哈希層得到相應(yīng)的哈希碼，從而可以在低維漢明空間中完成對大規(guī)模圖像數(shù)據(jù)的有效檢索。在3個常用數(shù)據(jù)集上的實驗結(jié)果表明，利用所提方法得到哈希碼，其圖像檢索性能優(yōu)于當(dāng)前主流方法。
-
關(guān)鍵詞:
- 圖像檢索 /
- 深度卷積神徑網(wǎng)絡(luò) /
- 二進(jìn)制哈希 /
- 量化誤差 /
- 獨立性
Abstract: With the increasing amount of image data, the image retrieval methods have several drawbacks, such as the low expression ability of visual feature, high dimension of feature, low precision of image retrieval and so on. To solve these problems, a learning method of binary hashing based on deep convolutional neural networks is proposed, which can be used for large-scale image retrieval. The basic idea is to add a hash layer into the deep learning framework and to learn simultaneously image features and hash functions should satisfy independence and quantization error minimized. First, convolutional neural network is employed to learn the intrinsic implications of training images so as to improve the distinguish ability and expression ability of visual feature. Second, the visual feature is putted into the hash layer, in which hash functions are learned. And the learned hash functions should satisfy the classification error and quantization error minimized and the independence constraint. Finally, an input image is given, hash codes are generated by the output of the hash layer of the proposed framework and large scale image retrieval can be accomplished in low-dimensional hamming space. Experimental results on the three benchmark datasets show that the binary hash codes generated by the proposed method has superior performance gains over other state-of-the-art hashing methods. -
LOWE D G. Distinctive image features from scale-invariant keypoints[J]. International Journal of Computer Vision, 2004, 60(2): 91-110. DALAL N and TRIGGS B. Histograms of oriented gradients for human detection[C]. Computer Vision and Pattern Recognition, San Diego, CA, USA, 2005: 886-893. KRIZHEVSKY A, SUTSKEVER I, and HINTON G E. ImageNet classification with deep convolutional neural networks[C]. Advances in Neural Information Processing Systems, South Lake Tahoe, Nevada, US, 2012: 1097-1105. DATAR M, IMMORLICA N, INDYK P, et al. Locality sensitive hashing scheme based on p-stable distributions[C]. Proceedings of the ACM Symposium on Computational Geometry, New York, USA, 2004: 253-262. ZHANG Lei, ZHANG Yongdong, ZHANG Dongming, et al. Distribution-aware locality sensitive hashing[C]. 19th International Conference on Multimedia Modeling, Huangshan, China, 2013: 395-406. KONG Weihao and LI Wujun. Isotropic hashing[C]. Advances in Neural Information Processing Systems, South Lake Tahoe, Nevada, US, 2012: 1646-1654. WEISS Y, TORRALBA A, and FERGUS R. Spectral hashing[C]. Advances in Neural Information Processing Systems, Vancouver, Canada, 2009: 1753-1760. GONG Yunchao, LAZEBNIK S, GORDO A, et al. Iterative quantization: a procrustean approach to learning binary codes for large-scale image retrieval[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 35(12): 2916-2929. WANG Jun, KUMAR S, and CHANG Shihfu. Semi-Supervised hashing for large scale search[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2012, 34(12): 2393-2406. KULIS B and DARRELL T. Learning to hash with binary reconstructive embeddings[C]. Advances in Neural Information Processing Systems, Vancouver, Canada, 2009: 1042-1052. LIU Wei, WANG Jun, JI Rongrong, et al. Supervised hashing with kernels[C]. Computer Vision and Pattern Recognition, Providence, RI, 2012: 2074-2081. GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. Computer Vision and Pattern Recognition, Ohio, Columbus, 2014: 580-587. OQUAB M, BOTTOU L, LAPTEV I, et al. Learning and transferring mid-level image representations using convolutional neural networks[C]. Computer Vision and Pattern Recognition, Ohio, Columbus, 2014: 1717-1724. RAZAVIAN A, AZIZPOUR H, SULLIVAN J, et al. CNN features off-the-shelf: an astounding baseline for recognition[C]. Computer Vision and Pattern Recognition, Ohio, Columbus, 2014: 806-813. XIA Rongkai, PAN Yan, LAI Hanjiang, et al. Supervised hashing for image retrieval via image representation learning[C]. Proceedings of the AAAI Conference on Artificial Intelligence, Qubec, Canada, 2014: 2156-2162. LAI Hanjiang, PAN Yan, LIU Ye, et al. Simultaneous feature learning and hash coding with deep neural networks[C]. Computer Vision and Pattern Recognition, Boston, MA, USA, 2015: 3270-3278. LIN K, YANG H F, HSIAO J H, et al. Deep learning of binary hash codes for fast image retrieval[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Boston, MA, USA, 2015: 27-35. GIONIS A, INDYK P, and MOTWANI R. Similarity search in high dimensions via hashing[C]. Proceedings of the International Conference on Very Large Data Bases, Edinburgh, Scotland, UK, 1999: 518-529. KRIZHEVSKY A and HINTON G. Learning multiple layers of features from tiny images[R]. Technical Report, University of Toronto, 2009. CHUA TatSeng, TANG Jinhui, HONG Richang, et al. NUS-WIDE: A real-world Web image database from national university of singapore[C]. Proceedings of the ACM International Conference on Image and Video Retrieval, Greece, 2009: 48. LIU Wei, WANG Jun, Kumar Sanjiv, et al. Hashing with graphs[C]. Proceedings of the 28th International Conference on Machine Learning, Bellevue, Washington, USA, 2011: 1-8. -
計量
- 文章訪問數(shù): 2230
- HTML全文瀏覽量: 217
- PDF下載量: 1324
- 被引次數(shù): 0
下載:
