Research On Image Rain Streaks Based On Machine Learning

In rainy weather,the image captured by the camera on rainy days is usually blurred and less visible,which greatly interferes with the ability of the computer to detect,identify and track objects automatically.For example,in terms of the division of responsibility in traffic accidents and the tracking and location of criminals by the police,the interference of the rain streaks to the cameras in important geographical locations leads to the failure of the cameras to provide clear images as evidence,which will have adverse effects.Therefore,it is of great theoretical significance and practical application value to improve the clarity of images by removing rain streaks in rain images.In recent years,visual perception enhancement technology,which improves image quality by removing rain streaks,has become a research hotspot in the field of image processing.In this paper,the technology of removing rain streaks from a single image is comprehensively studied.The specific research contents are as follows:(1)Aiming at the single image rain streaks deviation from the vertical direction,to the renderings appear after the rain the rain streaks residual obvious or background blur phenomena,this paper implements the Angle the rain streaks translation global sparse to rain model,first of all,using median filter to filter the rain image,get a preliminary rain streaks estimation,and through the transformation,such as translation,calculate the vertical gradient strategy makes the rain streaks vertically,and then,through three sparse regularization build global sparse model is going to rain,three sparse regularization include: the sparse regular term of the rain streaks,the gradient sparseness in the main direction of the rain streaks(y direction),and the gradient sparseness in the main direction of the rain streaks(x direction).Finally,the image obtained by removing the rain streaks through the global sparse model is operated by the corresponding inverse transformation to obtain the rain-free image.The experimental results show that the algorithm can remove the rain streaks to a large extent and retain the background details of the image even if the rain streaks deviates from the vertical direction.(2)Based on the residual Res Net can strengthen the deep learning by changing the mapping range,the input is divided into high frequency detail layer and base layer,through a direct mapping of input to output to strengthen the image features such as network operating building: by frequency domain transformation to separate image makes the operation objectives further thinning,the image is divided into high frequency part and low frequency part,because of the rain streaks almost only exists in the high frequency part,so here we only to the operation of the high frequency part to remove the rain streaks.We use Squeeze-and-Excitation(SE)network layer replace batch normalized(BN),the Squeeze-and-Excitation(SE)network layer added to the residue in the network,in this way can make the operation of image target range narrowed,sparse sexual enhancement.In order to show this method is effective and feasibility,this paper made a lot of experiments on three data sets,the experimental results proved that the implementation of model to remove the rain streaks effect is good,has solved the rain streaks residual obvious or insufficient background blur,on experimental operation speed is more than a lot of single image rain streaks removal algorithms.