The method of preliminary medical images selection

This work investigates the use of some features computed using the Hough Transform as a criterion for image similarity evaluation based on picture retrieval. The method is based on a modified Hough Transform and used to approximate edges detected in the input image with straight lines. We consider retrieving images from a thematic database, where the semantic contents of images is limited to a specific domain (for example medical images). Proposed analysis of images does not claim that all features will be used, but special selection of features is well suited for evaluating of the visual coherence of images. In the first stage line elements are extracted from the image by using Hough Transform. In the second stage similarity coefficients are computed.