Image retrieval and recognition based on generalized local distance functions

被引：0
|
作者：
Gu H. [1 ]
Zhao G.-Z. [1 ]
机构：
[1] College of Electrical Engineering, Zhejiang University
来源：
Zhejiang Daxue Xuebao (Gongxue Ban)/Journal of Zhejiang University (Engineering Science) | 2011年 / 45卷 / 04期
关键词：
Adaboost ensemble; Distance metric learning; Generalized local image distance; Image categorization;
D O I：
10.3785/j.issn.1008-973X.2011.04.002
中图分类号：
学科分类号：
摘要：
A metric distance of images called generalized local distance function was proposed for the image distances computation aiming at the most commonly local feature descriptors. The distance had two parts, the feature-to-image distance and the image-to-image distance. The image-to-image distance was defined as a combination of feature-to-image distances, where the feature-to-image distance was computed according to the k-nearest neighbor distances of that feature. The feature-to-image distance was represented in several ways with different constraint assumptions. The distance function can be learned by a quadratic optimization problem based on relative comparisons. Then Adaboost was used to ensemble the learned distance functions to obtain a final image classifier. The generalized local distance overcomes the shortcoming of original linear local distance function, for which most of the statistical information is lost. Experimental results show that the method significantly improves the image categorization performance.
引用
收藏
页码：596 / 601
页数：5
相关论文
共 19 条
  • [1] Datta R., Joshi D., Li J., Et al., Image retrieval: ideas, influences, and trends of the new age, ACM Computing Surveys, 40, 2, pp. 1-60, (2008)
  • [2] Lowe D.G., Distinctive image features from scale-invariant keypoints, International Journal of Computer Vision, 60, 2, pp. 91-110, (2004)
  • [3] Dalai N., Triggs B., Rhone-Alps I., Et al., Histograms of oriented gradients for human detection, Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 1063-6919, (2005)
  • [4] Berg A.C., Berg T.L., Malik J., Shape matching and object recognition using low distortion correspondences, Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, pp. 26-33, (2005)
  • [5] Sivic J., Zisserman A., Video google: a text retrieval approach to object matching in videos, Proceedings of the 9th IEEE International Conference on Computer Vision, pp. 1470-1477, (2003)
  • [6] Grauman K., Darrell T., The pyramid match kernel: discriminative classification with sets of image features, Proceedings of the 10th IEEE International Conference on Computer Vision, pp. 1288-1296, (2005)
  • [7] Xu L., Zhao G.-Z., Gu H., Novel one-vs-rest classifier based on SVM and multi-spheres, Journal of Zhejiang University: Engineering Science, 43, 2, pp. 303-308, (2009)
  • [8] Xu L., Zhoao G.-Z., Gu H., Recurive training algorithm for one-class support vector machine based on active set method, Journal of Zhejiang University: Engineering Science, 43, 1, pp. 42-46, (2009)
  • [9] Fei-Fei L., Fergus R., Perona P., Learning generative visual models from few training examples: an incremental bayesian approach tested on 101 object categories, Computer Vision and Image Understanding, 106, 1, pp. 59-70, (2007)
  • [10] Frome A., Singer Y., Sha F., Et al., Learning globally-consistent local distance functions for shape-based image retrieval and classification, Proceedings of the 11th IEEE International Conference on Computer Vision, pp. 1255-1263, (2007)
12