Search Space Partitioning Using Convex Hull and Concavity Features for Fast Medical Image R.pdf

SEARCH SPACE PARTITIONING USING CONVEX HULL AND CONCAVITY FEATURES FOR FAST MEDICAL IMAGE RETRIEVAL Nikolay M. Sirakov1 and Phillip A. Mlsna2 1Dept. of Mathematics and Statistics, P.O. Box 5717 2Dept. of Electrical Engineering, P.O. Box 15600 Northern Arizona University, Flagstaff, AZ 86011 Phone: 1(928) 523-6893, 2(928) 523-2112; Fax: 1(928) 523-5847 E-mail: Nikolay.Sirakov@nau.edu, Phillip.Mlsna@nau.edu ABSTRACT Classification into feature categories by which the image content can be indexed [13]; A new approach is presented for partitioning an image Image indexing by features, such as shape [1, 4]; database by classifying and indexing the convex hull Image retrieval by matching a query image or feature shapes and the number of region concavities. The result vector with images or feature vectors in the image is a significant increase in image search and retrieval database [10, 11]. speed. The convex hull is first determined using a novel For an extensive database of images, search space and efficient approach based on the geometrical heat partitioning and indexing techniques can facilitate a differential equation. A set of seven templates is significant computational decrease of the query process. It available to approximate the boundary as observed from can be accomplished by reducing the si