期刊名称：International Journal of Multimedia and Ubiquitous Engineering
印刷版ISSN：1975-0080
出版年度：2016
卷号：11
期号：8
页码：367-378
出版社：SERSC
摘要：In this paper, a dynamic algorithm is proposed that is aimed at removing raindrops within fixed background video images according to the volume of raindrops. In surveillance videos with fixed backgrounds, when the field of vision is obstructed because of falling rain or snow, obtaining picture information on the objects becomes difficult. Therefore, obstacles such as raindrops that obstruct the field of vision should be removed from pictures to obtain more specific picture information. The proposed technique obtains images on a periodical basis from a fixed background picture and assesses whether there are raindrops in the present screen among the obtained images. The proposed technique predicts the volume of raindrops by comparing images obtained periodically and removes raindrops by applying a different dynamic algorithm based on the predicted volume of raindrops. This study embodied and experimented with the proposed technique and verified that the result was excellent.
关键词：computer vision; rain detection; rain removal; video surveillance
