臺灣博碩士論文加值系統

本論文提出透過一種由模糊類神經網路與主分量為基礎的支持向量學習(FNN-PCSV) 之即時物體偵測的新方法。FNN-PCSV是一種TSK型式模糊規則所組成的模糊系統。FNN-PCSV的前件部是透過輸入數據的模糊分群而產生的，在FNN-PCSV的後件部首先是利用PCA來減少維度，然後在主分量空間利用線性的支持向量機調整後件部的參數，以提供網路較好的綜合成效。物體偵測系統包含兩階段。第一階段利用一個物體當作偵測特徵的全域色彩出現的彩色直方圖送入一個FNN-PCSV分類器。要能夠準確的描述彩色資訊利用直方圖是困難的，因此提出了色彩空間的非均勻切割。一個有效率的方法針對直方圖的擷取在影像的掃瞄過程中提出了即時系統的應用。第二階段利用幾何相依的區域色彩出現的色彩特徵送入另一個FNN-PCSV分類器中。候選區產生在第一階段，而這階段是減少錯誤個數的產生。為了要驗證所提出方法的成效，因此做了兩個特定物體的實驗。為了能加以比較，其他型式的偵測方法和分類器皆被應用在相同的問題。結果發現我們提出的FNN-PCSV 為基礎偵測的方法和其他方法比較有最好偵測結果。

A new method for real-time object detection by a Fuzzy Neural Network with Principal Component-based Support Vector learning (FNN-PCSV) is proposed in this thesis. FNN-PCSV is a fuzzy system that consists of Takagi-Sugeno-Kang (TSK) type fuzzy rules. The antecedent part of FNN-PCSV is generated via fuzzy clustering of the input data. The dimension of free parameter vector in the consequent part of FNN-PCSV is first reduced by the PCA. A linear support vector machine is then used to tune the consequent parameters on the principal component space to give the network better generalization performance. The object detection system consists of two stages. The first stage uses color histogram of the global color appearance of an object as detection feature for a FNN-PCSV classifier. To represent color information by histograms as accurately as possible, a non-uniform partition of color space is proposed. An efficient method for histogram extraction during the image scanning process is proposed for real-time implementation. The second stage uses geometry-dependent local color appearance as color feature for another FNN-PCSV classifier. Candidates generated in stage one are filtered in this stage to reduce the number of false alarms. To verify performance of the proposed method, experiments on detection of two specific objects are performed. For comparison, other types of detection methods and classifiers are also applied to the same detection task. Results show the proposed FNN-PCSV-based detection system achieves better results than compared methods.

Contents

Abstract (in Chinese)……………………………………………………………………………i
Abstract (in English)………………………………………………ii
Contents………………………………………………………………iii
List of Figures………………………………………………………v
List of Tables………………………………………………………vii

Chapter 1：Introduction……………………………………………1
1.1 Survey and Literature Review…………………………1
1.2 Organization of the Thesis……………………………4

Chapter 2：Fuzzy Neural Network with Principal Component-based Support Vector Learning (FNN-PCSV)………………………6
2.1 Structure of FNN-PCSV……………………………………6
2.2 FNN-PCSV Structure Learning……………………………9
2.3 FNN-PCSV Parameter Learning…………………………10

Chapter 3：Stage One: Global Color-based FNN-PCSV Detection………………………………………………………………15
3.1 Color Space………………………………………………15
3.2 Color Space Partition and Histogram………………18
3.3 Multi-Resolution Structure with Fast Histogram Extraction Method……………………………………………………23
3.4 Object Candidates Generation…………………………27

Chapter 4：Stage two：Local Color-based FNN-PCSV Detection………………………………………………………………30
4.1 Geometry dependent local color feature extraction……………………………………………………………30

Chapter 5：Experiments……………………………………………34
5.1 Off-Line Experiments and Results……………………34
5.2 Comparisons with Other Detection Methods…………43
5.3 Real Time Object Detection System…………………45

Chapter 6：Conclusions……………………………………………48

References……………………………………………………………49

List of Figures

Figure 1.1 Detection flow of the proposed system…………5
Figure 2.1 Structure of FNN-PCSV………………………………7
Figure 3.1 RGB color space……………………………………16
Figure 3.2 HSV color space……………………………………17
Figure 3.3 Non-uniform partition of HS space……………19
Figure 3.4 Different views of the detected can…………21
Figure 3.5 Distributions of training pixels on HS space and the non-uniform partition results of can ………21
Figure 3.6 Different views of the detected cup………22
Figure 3.7 Distributions of training pixels on HS space and the non-uniform partition results of cup………22
Figure 3.8 Input image pyramid for focus region extraction, where α is a resizing factor……………………23
Figure 3.9
(a) The horizontal scan of the features extraction………26
(b) The vertical scan of the features extraction…………26
Figure 3.10
(a) Input image pyramid and the detected objects …………28
(b) Positions of detection objects in difference scales…28
Figure 3.11
(a) Original image…………………………………………………29
(b) Centers of object candidate regions………………………29
(c) Morphological opening operation results…………………29
(d) The MERs…………………………………………………………29
(e) Detected objects after stage one…………………………29

Figure 4.1
(a) the MR within a contour………………………………………31
(b) Distributions of the five regions in a MR………………31
Figure 4.2
(a) Original image …………………………………………………32
(b) The MER and MR of two candidate regions from stage one………………………………………………………………………32
(c) Distribution of the five regions in each MR……………32
Figure 4.3 The finally detected position and size of an object after stage two……………………………………………33
Figure 5.1 Different views of the can in different backgrounds……………………………………………………………35
Figure 5.2 Detection results by the proposed FNN-PCSV based method in example 1…………………………………………39
Figure 5.3 Detection results by the proposed FNN-PCSV based method in example 2…………………………………………40
Figure 5.4 False alarms results for the tests images FNN-PCSV based method in example 1…………………………………41
Figure 5.5 False alarms results for the tests images FNN-PCSV based method in example 2…………………………………41
Figure 5.6 Sequence of the search position………………47
Figure 5.7 The result of the real time object detection system…………………………………………………………………47

List of Tables

Table 5.1 Detection Results Using FNN-PCSV And Other Different Classifiers in Proposed Two-Stage Global-Local Detection Method in Example 1……………………………………42
Table 5.2 Detection Results Using FNN-PCSV And Other Different Classifiers in Proposed Two-Stage Global-Local Detection Method in Example 2……………………………………43
Table 5.3 Detection Results of Different Methods In Example 1………………………………………………………………44
Table 5.4 Detection Results of Different Methods In Example 2………………………………………………………………44
Table 5.5 Interface of the Hardware and Software………………………………………………………………46

[1] W. E. L. Grimson, “Object Recognition by Computer: The Role of Geometric Constraints,” MIT Press, Cambridge, Massachusetts, 1990.
[2] B. Bhanu and J. Peng, “Adaptive integrated image segmentation and object recognition,” IEEE Trans. Syst., Man, and Cyber.,- Part C: Applications and Reviews, 30(4) (2000) 427-441.
[3] A. C. Berg, T. L. Berg, and J. Malik, “Shape matching and object recognition using low distoration correspondences,” in: Proc. IEEE Conf. Computer Vision and Pattern Recognition (CVPR), 2005, pp. 26-33.
[4] D. I. Barnea and H. F. Silverman, “A class of algorithms for fast digital image registration,” IEEE Trans. on Comput. 21(2) (1972) 179–186.
[5] T. Kawanishi, T. Kurozumi, K. Kashino, and S. Takagi, “A fast template matching algorithm with adaptive skipping using inner-subtemplate’s distances,” in: Proc. IEEE Int. Conf. Pattern Recognition, 2004, pp. 654-657.
[6] E. Osuna, R. Freund and F. Girosi, “Training support vector machines: an application to face detection,” in: Proc. IEEE Int. Conf. Computer Vision and Pattern Recognition, 1997, pp. 130-136.
[7] H. A. Rowley, S. Baluja, and T. Kanade, “Neural network-based face detection,” IEEE Transactions on pattern analysis and machine intelligence. 20 (1998) 22-38.
[8] C. Garica and M. Delakis, “Convolutional face finder: a neural architecture of fast and robust face detection,” IEEE Trans. Pattern Analysis and Machine Intelligence, 26(11) (2004) 1408-1423.
[9] Z. Suna , G. Bebisa, and R. Millerb, “Object detection using feature subset selection,” Pattern Recognition, 37(11) (2004) 2165– 2176.
[10] R. Lienhart and J. Maydt, “An extended set of Haar-like features for rapid object detection,” in: Proc. of IEEE Int. Conf. Image Processing, 2002, pp. 900-903.

[11] D. G. Lowe, “Distinctive Image Features from Scale-Invariant Keypoints,” International Journal of Computer Vision, 60, 2, pp. 91-110, 2004.
[12] V. V. Vinod and H. Murase, “Focused color intersection with efficient searching for object extraction,” Pattern Recognition, 30(10) (1997) 1787-1797.
[13] P. Chang and J. Krumm, “Object recognition with color cooccurrence histograms,” in: Proc. IEEE Conf. Computer Vision and Pattern Recognition, 2 (1999) 498-504.
[14] T. Kawanishi, T. Kurozumi, K. Kashino, and S. Takagi, “Dynamic active search for quick object detection with pan-tilt-zoom camera,” in: Proc. IEEE Int. Conf. Image Processing, (2001) 716-719.
[15] K. K. Sung and T. Poggio, “Example-based learning for view-based human face detection,” IEEE Trans. Pattern Analysis and Machine Intelligence. 20 (1) (1998) 39-51.
[16] H. Zhang, W. Gao, X. Chen, and D. Zhao, “Object detection using spatial histogram features,” Image and Vision Computing, 24 (2006) 327–341.
[17] B. Schiele and J. L. Crowley, “Object recognition using multidimensional receptive field histograms,” in: Proc. IEEE Int. Conf. Pattern Recognition, 2 (1996) 50-54.
[18] D. Crandall and J. Luo, “Robust color object detection using spatial-color joint probability functions,” in: Proc. IEEE Conf. Computer Vision and Pattern Recognition, 2004, pp. 379-385.
[19] L. W. Lan and A. Y. Kuo, “Development of a fuzzy neural network color image vehicular detection (FNNCIVD) system,” in: Proc. IEEE 5th Conf. Intelligent Transportation Systems, 2002, pp. 88 – 93.
[20] C. F. Juang and S. J. Shiu, “Using self-organizing fuzzy network with support vector learning for face detection in color images,” Neurocomputing, in press, [available online], 2008.
[21] C. F. Juang, S. H. Chiu, and S. W. Chang, “A self-organizing TS-type fuzzy network with support vector learning and its application to classification problems,” IEEE Trans. Fuzzy Systems. 15(5) (2007) 998-1008.
[22] C. T. Lin and C. S. G. Lee, “Neural Fuzzy Systems: A Neural-Fuzzy Synergism to Intelligent Systems,” Prentice Hall, May, 1996.
[23] V. Vapnik, “The Nature of Statistical Learning Theory,” New York : Springer - Verlag, 1995.
[24] N. Cristianini and J. S.-Taylor, “An Introduction to Support Vector Machines And Other Kernel-based Learning Methods,” Cambridge University Press, 2000.
[25] J. H. Chiang and P. Y. Hao, “Support vector learning mechanism for fuzzy rule-based modeling: a new approach,” IEEE Trans. Fuzzy Systems, vol. 12, no. 1, pp. 1-11, 2004.
[26] C. T. Lin, C. M. Yeh, S. F. Liang, J. F. Chung, and N. Kumar, “Support-vector-based fuzzy neural network for pattern classification,” IEEE Trans. Fuzzy Systems, vol. 14, no. 1, pp. 31-41, Feb. 2006.
[27] Y. Chen, and J. Z. Wang, “Support vector learning for fuzzy rule-based classification systems,” IEEE Trans. Fuzzy Systems, vol. 11, no. 6, pp. 716-728, Dec. 2003.
[28] C. F. Juang and C. T. Lin, “An on-line self-constructing neural fuzzy inference network and its applications,” IEEE Trans. Fuzzy Systems, vol. 6, no.1, pp. 12-32, 1998.
[29] G. Bradski, A. Kaehler, and V. Pisarevsky, “Learning-based computer vision with Intel’s open source computer vision library,” Intel Technology Journal. 9(2) (2005) 119-130.
[30] Open Source Computer Vision Library:
http://www.intel.com/software/products/perflib