The study of dimensionality reduction methods in the task of browsing of digital image collections
Автор: Myasnikov Evgeniy Valerievich
Журнал: Компьютерная оптика @computer-optics
Рубрика: Обработка изображений: Восстановление изображений, выявление признаков, распознавание образов
Статья в выпуске: 3 т.32, 2008 года.
Бесплатный доступ
Some dimensionality reduction methods are studied in this paper. Methods are applied to the task of browsing of digital image collections in accordance to image visual characteristics. Methods are compared to each other by evaluating the Sammon stress of mapping from multidimensional space to 2D-space and the time required to obtain a decision. A survey of methods used for construction of systems for digital image collections browsing is given. Experiment was carried out on the set of color images of wide range. The results of experimental study are present in the paper. The recommendations on methods usage are given.
Dimensionality reduction, sammon mapping, digital image collection
Короткий адрес: https://sciup.org/14058833
IDR: 14058833