Study on the Rain Removal Algorithm of Single Image

In this paper, we propose a new method to remove raindrop and rain streak from a single image. We know that raindrop and rain streak on the image can severely hamper the visibility of a background scene and degrade an image considerably. By analysing the rainy image, we conclude that raindrop and the rain streak are the nature noise, at the same time, we find that compared to the background scene, the spectrum of the raindrops and rain streaks are mainly the high-frequency component, especially to the rain streaks. So we propose a new method to remove raindrop and rain streak from a single image. First we use the median filter to filter the salt and pepper noise in the nature noise. Secondly, we use a group of low-pass filters to filter the high-frequency component of the image. Thus, we remove a lot of raindrops and rain streaks on the image, but after filtering, there is something like mist or fog on the image, then we use the haze removal method to remove them. Finally, we use the sharpening technology to show off details of the image. In the end, we use our method to experiments, we find the result is significantly, so we can conclude our approach is effectiveness and quantitatively.