Research And Implementation Of Image Weather Detection And Removal

Weather has a very important impact on human daily life and production activities.From the period of Xia and Shang,people have begun to engage in agricultural production.However,many bad weather still cause many problems in our daily life.Reducing the impact of these severe weather on computer vision applications is the fundamental purpose of this study.This paper designs two novel image weather detection algorithms and image rain removal algorithm.Through experimental verification and comparison with other similar algorithms,these algorithms work well and are applied to the image rain removal and weather detection system studied in this paper.The system can recognize the weather of outdoor images,provide basic support for outdoor video processing and other computer vision applications,and second,it can perform image restoration on rainy weather images to remove the effects of the rain and make the images clearer.Inspired by the excellent performance of deep learning technology on many computer vision issues in recent years,unlike traditional image classification and image restoration methods,this paper applies deep learning technology to weather detection and recovery applications.There are three main contributions below:(1)20,000 images of different types of weather were collected,and a data set consisting of weather images of cloudy,sunny,rainy and snowy was constructed,and a deep convolutional neural network was used for weather detection.The trained network was used to extract image features.The features were input into the classifier and the images were classified into four types of cloudy,rain,and snow weather.(2)Take a rainy image as input,the corresponding clean image as output,train a deep residual network.This deep residual network model is embedded into the frame of the plugand-play alternating direction multiplier method of image restoration,(3)The above-mentioned image weather detection and rain removal algorithm is implemented,and applied to the image removal rain and weather detection system,which can efficiently and accurately provide weather detection and rain chart recovery services.