Rain detection and removal of sequential images
Ming Zhou, Zhichao Zhu, Rong Xin Deng, Shuai Fang · 2011
Detection and removal of rain in image is a difficult and crucial problem due to the complexity of rain and its negative effects on image. The spatio-temporal property and the chromatic property of rain are comprehensively analyzed. Using the two properties, a simple but effective algorithm is proposed to detect and remove the rain of sequential images. Firstly time complexity of k-means is reduced by our optimized algorithm. Then each pixel is classified by the optimized k-means according to the spatio-temporal property. Next the chromatic constraint is used to identify accurately the rain. Finally the blending parameter α is computed and the rain in each frame is removed. The method can handle both light and heavy rain in image. Experimental results prove that our algorithm performs simply but effectively.