МЕТОДИКА ОБНАРУЖЕНИЯ ДВИЖЕНИЯ И ОПРЕДЕЛЕНИЯ СКОРОСТИ ОБЪЕКТА ПО ВИДЕОФРАГМЕНТАМ

Kirill Gennadievich Denisov, Laysan Renatovna Panteleeva


Аннотация


The article presents new basis for identification and tracking moving objects in the sequence of images. Identification is based on finding similar objects in different shots and comparison of the distance from them to the objects of shooting. The objects are caught by the method of finding object’s boundaries and further combining them into closed 4- or 5-rib contours. After reduction of dimensions, the system looks for special points inside the contours which guarantee unique features by which the objects can be found within a shot. Special point is a combination of the brightness gradients’ values around this point. Such point will be invariant in regard to the size of the object, its lighting and the turn angle. In order to identify the speed of object’s motion by the size of caught rectangular the distance from the camera to the object is measured. Then the speed is assessed and fixed on the image by the difference between two shots. The same system can be also used in the automobile safety system for finding moving objects and avoiding possible accidents.

Ключевые слова


ВИДЕОПОСЛЕДОВАТЕЛЬНОСТЬ; УСТРОЙСТВА ВИДЕОФИКСАЦИИ; ОПЕРАТОР СОБЕЛЯ; АЛГОРИТМ КЭННИ; АЛГОРИТМ ПРИБЛИЖЕНИЯ ДУГЛАСА-ПЕКЕРА; МАТРИЦА ГЕССЕ

Литература


Гребенников Р. В. Решение задачи об оптимальном поведении толпы с использованием метода оптимизации роя частиц//Вестник ВГУ: Системный анализ и информационные технологии. 2009. 2. C. 107 -111.

Прэтт У. Цифровая обработка изображений. М.: Мир, 1982. 784 с.

Canny J.F. Finding edges and lines in images. Master’s thesis. Cambridge: MIT, 1983.

Smith S.M., Brady J.M. SUSAN -a new approach to Low Level Image Processing. DRA Technical Report TR95SMMS1b. 1995. 57 p.




DOI: https://doi.org/10.12731/wsd-2014-2.1-20

Ссылки

  • На текущий момент ссылки отсутствуют.




(c) 2016 В мире научных открытий



ISSN 2658-6649 (print)

ISSN 2658-6657 (online)

HotLog Яндекс цитирования