Analysis, Diagnosis and Correction of Rain Streaks
Sumit Kumar, Akshita Gupta, Rajib Kumar Jha · 2019
In any unique and dynamic climate, for example, rain and snow, it causes varieties in local intensities of the rainy pictures. Such changes can remarkably influence the frameworks that rely upon highlights of the picture for object recognition, tracking, object detection, segmentation etc. Accordingly, there is a requirement for quick and precise detection [1]-[3] and evacuation of rain streaks or the marks of snow. Here, we build up a strategy for diagnosis and dismissal of rain components. Rain components have a higher intensity compare to the non-rain component of the image. This property has been utilized to recognize the rain components in a picture. To decrease the false alarm cases, we use the attributes of raindrops, (for example, major and minor axis) and recognize the identified rain and non-rain component. Correction of the color is utilized to reestablish shading of rain-component with the color of the background. We exhibit that our proposed technique performs well for rain-component identification with more prominent precision compare to other state-of-the-art methods.