Images analysis for automatic volcano visibility estimation
Автор: Kamaev Alexandr Nikolaevich, Urmanov Igor Pavlovich, Sorokin Aleksei Anatolyevich, Karmanov Dmitry Alexandrovich, Korolev Sergey Pavlovich
Журнал: Компьютерная оптика @computer-optics
Рубрика: Обработка изображений, распознавание образов
Статья в выпуске: 1 т.42, 2018 года.
Бесплатный доступ
In this paper, a method for estimating the volcano visibility in the images is presented. This method includes algorithms for analyzing parametric edges of objects under observation and frequency characteristics of the images. Procedures for constructing parametric edges of a volcano and their comparison are considered. An algorithm is proposed for identifying the most persistent edges for a group of several reference images. The visibility of a volcano is estimated by comparing these edges to those of the image under analysis. The visibility estimation is maximized with respect to a planar shift and rotation of the camera to eliminate their influence on the estimation. If the image quality is low, making it hardly suitable for further visibility analysis, the estimation is corrected using an algorithm for analyzing the image frequency response represented as a vector of the octave frequency contribution to the image luminance. A comparison of the reference frequency characteristics and the characteristics of the analyzed image allows us to estimate the contribution of different frequencies to the formation of volcano images. We discuss results of the verification of the proposed algorithms performed using the archive of a video observation system of Kamchatka volcanoes. The estimates obtained corroborate the effectiveness of the proposed methods, enabling the non-informative imagery to be automatically filtered off while monitoring the volcanic activity.
Image analysis, algorithms, edge detection, parametric edges, volcano, video surveillance, visibility analysis, edge matching
Короткий адрес: https://sciup.org/140228698
IDR: 140228698 | DOI: 10.18287/2412-6179-2018-42-1-128-140