Розпізнавання зображень на основі незалежного хешування структурних описів

Abstract

The work is devoted to the analysis of the effectiveness of image recognition methods based on random hashing for a database of reference descriptions. The research focuses on the development and modeling of a method that converts high-dimensional visual descriptors of key image points into compact binary codes. By applying random projection matrices to the description as a set of descriptors, the proposed system achieves significant acceleration in matching input images with stored database templates. The simulation results demonstrate that the proposed hashing approach maintains high recognition accuracy while significantly reducing computational costs compared to traditional exhaustive search methods.