應用於地點影像辨識之快速局部L-SVMs群體分類器

以作者查詢圖書館館藏

、以作者查詢臺灣博碩士

、以作者查詢全國書目

、勘誤回報

、線上人數：6

、訪客IP：18.218.38.125

姓名

王子睿(Tzu-Jui Wang) 查詢紙本館藏

畢業系所

通訊工程學系

論文名稱

應用於地點影像辨識之快速局部L-SVMs群體分類器
(Efficient Ensemble of Local L-SVMs for Visual Place Categorization)

相關論文

★ 應用於車內視訊之光線適應性視訊壓縮編碼器設計	★ 以粒子濾波法為基礎之改良式頭部追蹤系統
★ 應用於空間與CGS可調性視訊編碼器之快速模式決策演算法	★ 應用於人臉表情辨識之強健式主動外觀模型搜尋演算法
★ 結合Epipolar Geometry為基礎之視角間預測與快速畫面間預測方向決策之多視角視訊編碼	★ 基於改良式可信度傳遞於同質區域之立體視覺匹配演算法
★ 以階層式Boosting演算法為基礎之棒球軌跡辨識	★ 多視角視訊編碼之快速參考畫面方向決策
★ 以線上統計為基礎應用於CGS可調式編碼器之快速模式決策	★ 適用於唇形辨識之改良式主動形狀模型匹配演算法
★ 以運動補償模型為基礎之移動式平台物件追蹤	★ 基於匹配代價之非對稱式立體匹配遮蔽偵測
★ 以動量為基礎之快速多視角視訊編碼模式決策	★ 以高品質合成視角為導向之快速深度視訊編碼模式決策
★ 以運動補償模型為基礎之移動式相機多物件追蹤	★ 基於匹配代價曲線特徵之遮蔽偵測之研究

檔案

[Endnote RIS 格式]

[Bibtex 格式]

[相關文章]

[文章引用]

[完整記錄]

[館藏目錄]

[檢視]

[下載]

本電子論文使用權限為同意立即開放。
已達開放權限電子全文僅授權使用者為學術研究之目的，進行個人非營利性質之檢索、閱讀、列印。
請遵守中華民國著作權法之相關規定，切勿任意重製、散佈、改作、轉貼、播送，以免觸法。

摘要(中)

近年來基於影像之地點辨識(Visual place categorization)是研究人員重視的研究議題。然而，一影像辨識系統常受到物件大小、光線條件、物件遮蔽、視角變化影響。此外，地點辨識系統必須具備有效率的訓練及測試過程。相關的地點辨識研究中，其在分類影像內特徵時，大多採用非線性支撐向量機(support vector machine, SVM)，因其相較線性 SVM普遍能獲得較佳的辨識率，但其缺點是訓練階段之時間複雜度為O(N2)～O(N3)，N為訓練樣本數量，如此無法有效率地處理大規模(large-scale)樣本，因此本論文提出基於局部特性之linear SVM之群體分類器(ensemble of local linear SVMs, ELL-SVMs)，可有效降低訓練階段之時間複雜度至O(N1.5)。
　　本系統分為訓練階段及測試階段。於訓練階段，我們提出一創造群體(ensemble creation)的方法，其可在原特徵空間中找尋近乎線性可分割(linearly-separable)之子集，再利用一linear SVM分割每個子集內的樣本。在測試階段，對於影像中各個所偵測的特徵，我們利用貝式分類器及最近鄰居法則(nearest neighbor rule)個別選擇對每個特徵最佳的分類器。接著，本論文提出基於信心度之加權式一對多方案(confidence-based weighted one-against-all, CW-OAA)以結合影像中各個特徵之分類結果，以決定影像所屬地點。實驗結果顯示，本論文所提出的ELL-SVMs在訓練速度上，相似於FaLK-SVM，且優於BVM，在測試速度上，同時優於FaLK-SVM、BVM以及標準SVM。此外，本論文所提出之整合式地點影像的辨識率，皆優於FaLK-SVM、BVM以及標準SVM搭配OAA之方案。

摘要(英)

Recently visual place categorization is an important research topic due to its numerous potential applications. However, such visual categorization system is easily affected by object scale, illumination conditions, object occlusion and viewpoints. In addition, categorization system should be efficiently trained and tested with huge amount of visual cues extracted in a very short period. Relevant researches on visual place categorization rely on non-linear SVM to categorize those visual cues within each image, since non-linear SVM has always shown promising categorization results. However, its major defect is that it suffers from O(N2) to O(N3) in training complexity and O(D‧S) in test complexity, where N is the size of training data, D is the dimension of data vector, and S is the amount of support vector. Efficient training and test pro-cesses are demanding for tackling large-scale categorization problems. Therefore, this thesis proposes ensemble of local linear SVMs (ELL-SVMs), lowering training com-plexity to O(N1.5).
Our proposed scheme has training and test phases. In training phase, we propose a scheme for generating ensemble of local linear SVM (L-SVM). This idea is derived from discovering the linearly-separable partitions among training data, while such partitions are found, those partitions could be classified by linear SVM instead of non-linear one. In test phase, we impose nearest neighbour rule into Bayes decision rule to assist in identifying the best trained local L-SVM for test sample. Afterwards, we further propose confidence-based weighted one-against-all (CW-OAA) approach to fuse the categorization results of visual cues within an image, and thus to categorize the image. Empirically, the training speed of proposed ELL-SVMs is similar to that of FaLK-SVM and much lower than those of BVM and standard SVM. The test speed of ELL-SVMs is lower than FaLK-SVM, BVM and standard SVM. Moreover, the categorization ability of ELL-SVMs with CW-OAA outperforms three SVM’s variants with OAA.

關鍵字(中)

★ 決策融合
★ 支撐向量機
★ 地點辨識

關鍵字(英)

★ decision fusion
★ SVM
★ Place categorization

論文目次

摘要 I
Abstract III
致謝 VI
第一章緒論 1
1.1 前言 1
1.2 研究動機 1
1.3 研究方法 3
1.4 論文架構 4
第二章　地點辨識演算法之文獻回顧 5
2.1 典型辨識系統之架構 5
2.2 應用於地點辨識之影像特徵擷取 7
2.3 以視覺為基礎之地點辨識演算法現況 11
2.4 總結 13
第三章　Support Vector Machine (SVM)相關技術介紹 14
3.1 Support Vector Machine (SVM) 14
3.2 多類別Support Vector Machine (Multi-Class SVM) 22
3.3 Linear Support Vector Machine之快速訓練方案現況 23
3.4 Kernelized Support Vector Machine之快速訓練方案現 23
3.5 總結 25
第四章　本論文所提出之地點影像分類演算法 27
4.1 系統流程概述 27
4.2 訓練階段 28
4.2.1 貪婪式分割訓練樣本(Greedily Partitioning Training Data) 28
4.2.2 量化已分割樣本(Quantizing Partition using K-means++) 31
4.2.3 產生群體局部L-SVMs (Generating Ensemble of L-SVMs) 32
4.3 測試階段 35
4.3.1 基於貝式決策法(Bayesian Decision Rule)與最近鄰居(Nearest Neighbor Rule)之分類器選擇 35
4.3.2 近似最近鄰居距離(Approximated Nearest Neighbor 38
4.3.3 加權式決策值 (Weighted Decision Value) 39
4.3.4 基於信心度之加權式一對多決策法(Confidence-based Weighted One-against-All Scheme) 45
4.4 總結 47
第五章　實驗結果與討論 48
5.1 實驗環境及測試影像資料庫 48
5.2 分類SIFT描述子之效能分析 50
5.2.1 訓練及測試時間分析 52
5.2.2 辨識率分析 55
5.3 地點影像辨識方案分析 60
5.4 總結 64
第六章　結論及未來展望 66
參考文獻 67

參考文獻

[1] H. D. Whyte and T. Bailey, “Simultaneous localization and mapping: Part I,” IEEE Robotics & Automation Magazine, Vol. 13, No. 2, pp.99-110, Jun. 2006.
[2] B. E. Boser, I. Guyon, and V. Vapnik, “A training algorithm for optimal margin classifiers,” in Proceedings of ACM Conference on Learning Theory, Jul. 1992.
[3] I. W. Tsang, J. T. Kwok, and P.-M. Cheung, “Core vector machines: Fast SVM training on very large data sets,” Journal of Machine Learning Research, Vol. 6, pp. 363-392, Dec. 2005.
[4] I. W. Tsang, A. Kocsor, and J. T. Kwok, “Simpler core vector machines with en-closing balls,” in Proceedings of International Conference on Machine Learning, pp.911-918, Jun. 2007.
[5] N. Segata and E. Blanzieri, “Fast and scalable local kernel machines,” Journal of Machine Learning Research, Vol. 11, pp. 1883-1926, Mar. 2010.
[6] A. Oliva and A. Torralba, “Modeling the shape of the scene: A holistic represen-tation of the spatial envelope,” International Journal of Computer Vision, Vol. 42, No. 3, pp. 145-175, Jan. 2001.
[7] J. Wu, H. I. Christensen, and J.M. Rehg, “Visual place categorization: problem, dataset, and algorithm,” in Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 4763-4770, Oct. 2009.
[8] D. G. Lowe, “Distinctive image features from scale-invariant keypoints,” Interna-tional Journal of Computer Vision, Vol. 60, No. 2, pp. 91-110, Nov. 2004.
[9] Y. Ke and R. Sukthankar, “PCA-SIFT: a more distinctive representation for local image descriptors,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, Jul. 2004.
[10] J. M. Morel and G. Yu, “ASIFT: a new framework for fully affine invariant im-age comparison,” SIAM Journal on Imaging Sciences, Vol. 2, No. 2, Apr. 2009.
[11] H. Bay, A. Ess, T. Tuytelaars, and L. V. Gool, “SURF: speeded up robust fea-tures,” Journal of Computer Vision and Image Understanding, Vol. 110, No. 3, pp. 346-359, May 2008.
[12] K. Mikolajczyk and C. Schmid “A performance evaluation of local descriptors,” IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 10, No. 27, pp. 1615-1630, 2005.
[13] J. M. Morel and G. Yu, “Is SIFT Scale Invariant?” Inverse Problems and Imag-ing, Vol. 5, No. 1, Feb. 2011.
[14] D. Filliat, “A visual bag of words method for interactive qualitative localization and mapping,” in Proceedings of the IEEE/RSJ International Conference on Ro-botics and Automation, pp. 3921-3926, May 2007.
[15] P. Viola and M. J. Jones, “Robust real-time face detection,” International Journal of Computer Vision, 57(2):137-154, May. 2004.
[16] T.-J. Chin, H. Goh and J.-H. Lim, “Using densely recorded scenes for place recognition,” in Proceedings of the IEEE International Conference on Acoustics, Speech, and Signal Processing, pp.2101-2104, Mar. 2008.
[17] A. Quattoni and A. Torralba, “Recognizing indoor scenes,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, pp.413-420, Jun. 2009.
[18] A. Pronobis, B. Caputo, P. Jensfelt, and H. I. Christensen, “A discriminative ap-proach to robust visual place recognition,” in Proceedings of the IEEE/RSJ Inter-national Conference on Intelligent Robots and Systems, pp.3829-3836, Oct. 2006.
[19] A. Pronobis, O. M. Mozos, and B. Caputo, “SVM-based discriminative accumu-lation scheme for place recognition,” in Proceedings of the IEEE International Conference on Robotics and Automation, pp. 522-529, May 2008.
[20] C.-J. Hsieh, K.-W. Chang, C.-J. Lin, S. S. Keerthi, and S. Sundararajan, “A dual coordinate descent method for large-scale linear SVM,” in Proceedings of Inter-national Conference on Machine Learning, pp.408-415, Jul. 2008.
[21] D. M. J. Tax and R. P. W. Duin, “Data domain description using support vectors,” in Proceedings of European Symposium on Artificial Neural Networks, pp. 251-256, Apr. 1999.
[22] D. Arthur and S. Vassilvitskii, “K-means++: The advantage of careful seeding,” in 18th Symposium on Discrete Algorithms, 2007.
[23] H. Jegou, M. Douze, and C. Schmid, “Product quantization for nearest neighbor search,” IEEE Transaction on Pattern Analysis and Machine Intelligence, Vol. 33, No. 1, pp. 117-128, Nov. 2011.
[24] A. Pronobis and B. Caputo, “Confidence-based cue integration for visual place recognition,” in Proceedings of the IEEE/RSJ International Conference on Intel-ligent Robots and Systems, pp. 2394-2401, Oct. 2007.
[25] A. Pronobis, B. Caputo, P. Jensfelt, and H. I. Christensen, “A realistic bench-mark for visual indoor place recognition,” International Journal of Robotics and Autonomous Systems, Vol. 58, No. 1, pp. 81-96, Jan. 2010.
[26] A. Pronobis, O. M. Mozos, B. Caputo, and P. Jensfelt, “Multi-modal semantic place classification,” The International Journal of Robotics Research, Special Issue on Robotic Vision, Vol. 29, No. 2-3, pp. 298-320, Feb. 2010.
[27] R.-E. Fan, K.-W. Chang, C.-J. Hsieh, X.-R. Wang, and C.-J. Lin, “LIBLINEAR: A library for large linear classification,” Journal of Machine Learning Research, Vol. 9, pp. 1871-1874, Jun. 2008.
[28] N. Segata, “FaLKM-lib v1.0: a library for fast local kernel machines,” Technical Report DISI-09-025. 2009. Software available at http://www. disi.unitn.it/~segata/FaLKM-lib.
[29] Y. Tang, Y.-Q. Zhang, N.V. Chawla, and S. Krasser, “SVMs modeling for high-ly imbalanced classification,” IEEE Transactions on Systems, Man, and Cyber-netics, Vol. 39, No. 1, pp. 281-288, Feb. 2009.

指導教授

唐之瑋(Chih-Wei Tang)

審核日期

2011-7-25

推文