An optimality principle for unsupervised learning
Advances in neural information processing systems 1
Introduction to statistical pattern recognition (2nd ed.)
Introduction to statistical pattern recognition (2nd ed.)
Introduction to the theory of neural computation
Introduction to the theory of neural computation
Illumination Planning for Object Recognition Using Parametric Eigenspaces
IEEE Transactions on Pattern Analysis and Machine Intelligence
Visual learning and recognition of 3-D objects from appearance
International Journal of Computer Vision
Using Discriminant Eigenfeatures for Image Retrieval
IEEE Transactions on Pattern Analysis and Machine Intelligence
Matrix computations (3rd ed.)
Video segmentation for video data management
The handbook of multimedia information management
Digital Picture Processing
Occam Algorithms for Computing Visual Motion
IEEE Transactions on Pattern Analysis and Machine Intelligence
Dealing with occlusions in the eigenspace approach
CVPR '96 Proceedings of the 1996 Conference on Computer Vision and Pattern Recognition (CVPR '96)
Computing Ritz Approximations of Primary Images
ICCV '98 Proceedings of the Sixth International Conference on Computer Vision
Indexing Images by Trees of Visual Content
ICCV '98 Proceedings of the Sixth International Conference on Computer Vision
Partial eigenvalue decomposition of large images using spatial temporal adaptive method
IEEE Transactions on Image Processing
Video summarization by curve simplification
MULTIMEDIA '98 Proceedings of the sixth ACM international conference on Multimedia
Computing Content-Plots for Video
ECCV '02 Proceedings of the 7th European Conference on Computer Vision-Part IV
| Hi-index | 0.00 |
Large collections of images can be indexed by their projections on afew “primary” images. The optimal primary images are theeigenvectors of a large covariance matrix. We address the problem ofcomputing primary images when access to the images is expensive. This is thecase when the images cannot be kept locally, but must be accessed throughslow communication such as the Internet, or stored in a compressed form. Adistributed algorithm that computes optimal approximations to theeigenvectors (known as Ritz vectors) in one pass through the image set isproposed. When iterated, the algorithm can recover the exact eigenvectors.The widely used SVD technique for computing the primary images of a smallimage set is a special case of the proposed algorithm. In applications toimage libraries and learning, it is necessary to compute different primaryimages for several sub-categories of the image set. The proposed algorithmcan compute these additional primary images “offline”, withoutthe image data. Similar computation by other algorithms is impractical evenwhen access to the images is inexpensive.