Rain removal using guided image filtering for surveillance videos

The Intelligent Transportation System includes management of vehicle transportation. Video from surveillance cameras can be used for monitoring the number of vehicles and speed using digital image processing. However, in rainy conditions the monitoring process will be less accurate. So, removing rain noise is a solution that can be used. The method that will be presented in our paper is the method of rain removal using the low frequency and high frequency parts of the video frame. For each frame, low frequency as an image non-rain. Then, high frequency part as input from the guided filter, so we get a non-rain component from the high frequency part, and add a low frequency part is the restored frames. we also restore the nature of the background edge. The experiment will generate rain noise from videos without rain. The type of rain is heavy rain, moderate rain and light rain. The result of rain removal will be calculated image quality with a peak signal to noise ratio (PSNR). In addition, proper selection of radius and regulation parameters will reduce the rain effect more optimally. To be more detailed, we also test the performance of method on videos with static and moving foreground. The application of guided filter algorithm for three times, contained in preprocessing, filtering and recovering, can improve image quality.