Please use this identifier to cite or link to this item:
標題: 以支持向量訓練模糊系統做彩色直方圖為基礎之物體偵測
Color Histogram-based Object Detection by Support Vector trained Fuzzy Systems
作者: 孫文楷
Sun, Wen-Kai
關鍵字: object detection
color histogram
splitting K-means clustering
fuzzy system
出版社: 電機工程學系所
引用: References [1] Z. Suna , G. Bebisa, R. Millerb, “Object detection using feature subset selection”, Pattern Recognition 37 (2004), pp. 2165- 2176 [2] H. Zhang, W. Gao, X. Chen, D. Zhao, “Object detection using spatial histogram features”, Image and Vision Computing 24 (2006), pp. 327-341 [3] Kumar S. Ray, J. Ghoshal, “Neuro-fuzzy reasoning for occluded object recognition”, Fuzzy sets and systems 94 (1998), pp. 1-28 [4] T. kawanishi, T. Kurozumi, K. Kashino, S. Takagi, ”A fast template matching algorithm with adaptive skipping using inner-subtemplate's distances”, IEEE CNF. Pattern Recognition (ICPR'04), vol. 3, pp. 654-657, Aug. 2004. [5] D. I. Barnea and H. F. Silverman. “A class of algorithms for fast digital image registration”. IEEE Trans. on Comput., C-21(2):179-186, 1972. [6] T. Kawanishi, T. Kurozumi, S. Takagi, and K. Kashino. “Skipping template matching guaranteeing same accuracy as exhaustive search”. In Proc. of the Fifth Intl. Conf. of Advances on Pattern Recognition, pp. 209-212, 2003. [7] V. V. VINOD, H. MURASE, ”Focused color intersection with efficient searching for object extraction”, Pattern Recognition, Vol. 30, No. 10, pp.1787-1797, 1997. [8] Schiele, B. Crowley, J.L “Object recognition using multidimensional receptive field histograms”, IEEE CNF. Digital Object Identifier, Vol. 2, pp. 50-54, Aug, 1996. [9] Swain, M.J. and D.H. Ballard, “Color indexing”, International Journal of Computer Vision 7(1), pp. 11-32, 1991. [10] H. A. Rowley, S. Baluja, T. Kanade, “Neural Network-Based Face Detection”, IEEE Transactions on pattern analysis and machine intelligence. Vol. 20, pp. 22-38, January, 1998 [11] N. Cristianini and J. S.-Taylor, An Introduction to Support Vector Machines And Other Kernel-based Learning Methods, Cambridge University Press, 2000. [12] 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. 1-12, Feb. 2004. [13] 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. [14] C. T. Lin, C. M. Yeh, and C. F. Hsu, “Fuzzy neural network classification using support vector machine,” Prof. IEEE Symp. Circuits and Systems, pp. 724-727, 2004. [15] 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. [16] T. Joachims, N. Cristianini, and J. Shawe-Taylor, “Composite kernels for hypertext categorization,” Proc. Int. Conf. Machine Learning, 2001. [17] J. S. Roger Jang, and C. T. Sun, “Functional Equivalence Between Radial Basis Function Networks and Fuzzy Inference System, IEEE Trans. Neural Networks, vol. 4, no. 1, pp. 156-159, January 1993. [18] S. H. Chiu, Skin Color Image Segmentation by Support Vector Machine-aided Self Organizing Fuzzy Network, Master Thesis, Department of Electrical Engineering, National Chung Hsing University, Taiwan, July, 2005. [19] C. F. Juang, and C. T. Lin, “An online self-constructing neural fuzzy inference network and its applications,” IEEE Trans. Fuzzy Systems, vol. 6, pp. 12-32, February 1998.
摘要: 摘 要 本論文提出透過SOTFN-SV偵測物體的新方法。SOTFN-SV是一種由模糊群組與支持向量機(SVM)混合建構下的模糊系統。SOTFN-SV的前件部是透過輸入數據的模糊分群而產生的,然後支持向量機(SVM)用來調整接續部分的參數以提供網路較好的綜合成效。物體偵測是基於彩色資訊,而一件物體在色澤及飽和度的顏色空間下呈現的彩色直方圖可用來當作偵測特徵。為了儘可能精確的表達直方圖的彩色資料,因此提出了色澤及飽和度空間的不均勻分割。基於直方圖的SOTFN-SV全部偵測過程包含以下三階段:在第一階段中,輸入的影像藉由一個因子不斷重覆的再取樣,結果產生出一個金字塔影像。用已界定的視窗大小掃描所有的縮小影像,在此每個視窗的直方圖被當作輸入值而餵入SOTFN-SV分類噐中。許多候選物體皆在此階段被偵測出來。在第二階段中,分裂的K-means分群法可被應用於偵測,以致使鄰近位置的偵測被分入同一群組裡。根據群組變異數的分群方法,藉此群組數量會自動產生。大於門檻偵測值的群組會被保留,而他們的中心皆為所偵測物體的位置。在最後一個階段中,決定出每一個被偵測物體的大小。為了要驗證所提出方法的成效,因此執行偵測杯子的實驗。為了能加以比較,其他三種偵測,分別為直方圖交點、SONFIN及SVM,皆會被應用到此相同的問題。基於上述比較後,我們發現SOTFN-SV得到最好的偵測結果。
Abstract A new method for object detection by a Self-Organizing TS-type Fuzzy Network with Support Vector learning (SOTFN-SV) is proposed in this thesis. SOTFN-SV is a fuzzy system constructed by the hybridization of fuzzy clustering and SVM. The antecedent part of SOTFN-SV is generated via fuzzy clustering of the input data, and then SVM is used to tune the consequent part parameters to give the network better generalization performance. Object detection is based on color information and color histograms of an object on the Hue and Saturation (HS) color space are used as detection features. To represent color information by histograms as accurately as possible, a non-uniform partition of HS space is proposed. The whole histogram-based SOTFN-SV detection process consists of three stages. In the first stage, the input image is repeatedly sub-sampled by a factor, resulting in a pyramid of images. Scanning on all of the scaled images with a pre-defined window size is performed, where histograms of each window are fed as inputs to a SOTFN-SV classifier. Many candidate objects are detected in this stage. In the second stage, splitting K-means clustering is applied to the detections so that detections with nearby positions are grouped into the same cluster. The number of clusters is generated automatically by the clustering method according to cluster variances. The clusters containing detection numbers larger than a threshold are preserved and their centers are the positions of finally detected objects. In the final stage, size of each detected object is determined. To verify performance of the proposed method, experiments on cup detection are performed. For comparison, other three detectors, including histogram intersection, Self-cOnstructing Neural Fuzzy Inference Network (SONFIN), and Support Vector Machine (SVM), are applied to the same problem. From comparisons, we find that SOTFN-SV achieves the best detection result.
其他識別: U0005-2007200620244500
Appears in Collections:電機工程學系所



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.