Text

theses

In modern computer vision applications where datasets are large and updates with new data may be ongoing, methods of online clustering are extremely important. Online clustering algorithms incrementally cluster the data points, use a fraction of the dataset memory, and update the clustering decisions when new data comes in. In this thesis we adapt a classic online clustering algorithm called Balanced Iterative Reducing and Clustering using Hierarchies (BIRCH) to incrementally cluster large datasets of features commonly used in computer vision, e.g., 840K color SIFT descriptors, 1.09 million color patches, 60K outlier corrupted grayscale patches, and 700K grayscale SIFT descriptors. We use the algorithm to cluster datasets consisting of non-convex clusters, e.g., Hopkins 155 3-D motion segmentation dataset. We call the adapted version modified-BIRCH (m-BIRCH). BIRCH was originally developed by the database management community. Modifications made in m-BIRCH enable data driven parameter selection and effectively handle varying density regions in the feature space. Data driven parameter selection automatically controls the level of coarseness of the data summarization. Effective handling of varying density regions is necessary to well represent the different density regions in the data summarization. Our implementation of the algorithm provides a useful clustering tool and is made publicly available. In the second part of the thesis, we present a micro-level feature based approach to register time-lapse skin images acquired over an extended period of time and multimodal skin images acquired in quick succession. Misregistration between the images makes it difficult to perform quantitative analysis and track the progression of skin disease. We demonstrate the utility of both types of registration, by employing the results for two quantitative dermatology tasks: 1) automatic detection of acne-like regions, and 2) separation of surface and subsurface reflection. Additionally, we have created a time-lapse video showing the registered time-lapse images, which clearly brings out the evolution of acne lesions with time.

ETD_4527

Ph.D.

Includes bibliographical references

by Siddharth K. Madan

http://hdl.rutgers.edu/1782.1/rucore10001600001.ETD.000068911

doi:10.7282/T3XW4HDJ

ETD doctoral

The author owns the copyright to this work.