Size and angle filter based rain removal in video for outdoor surveillance systems

초록

Rain presents a challenge for outdoor surveillance systems, because raindrops can cause temporal and random noise, which in turn affects the performance of vision systems. This paper presents a method to remove the effects of raindrops using two continuous frames in a video. The two adjacent frames are assumed to have the same background and the raindrops do not overlap in the two images. Thus, the rain-affected pixels can be identified by finding the changes in RGB values between the two adjacent frames. Finally, the false candidate rain-affected pixels are filtered using the properties of raindrops in images such as speed and size. The performance of the proposed algorithm on rainy videos shows that the algorithm can effectively remove the raindrops in real time for outdoor videos. © 2011 Asian Control Association.