Application Research Of Shadow Removal Of UAV Image Based On Retinex Method

Due to light,terrain ups and downs and so on other reasons,the acquired image of UAV aerial survey is able to appear shadow,making the image contrast lower and surface information blur,reduce the quality of image while also reducing the quality of DOM and the accuracy of visual interpretation.In order to effectively recover the important surface information under the shadow in UAV aerial survey image,this paper is based on the analysis of the research progress of aerial image shadow elimination at home and abroad,uses the obtained image existing shadow of the UAV aerial survey in the Longnan mountain land to determine the right project as data sources,selects 10 images as the representative datas and based on the nature and characteristics of the shadow in the experimental area image and uses traditional processing methods and Retinex methods to deal with the image existing shadow and carry out comparative analysis research,the specific research conclusions are as follows:(1)It takes advantage of the traditional image enhancement methods such as algebraic operation,histogram equalization,homomorphic filtering and uniform light coloring to deal with the image existing shadow to evaluate the experimental results synthetically from the subjective effect and the image quality index statistical value such as standard deviation,information entropy and mean gradient.Results show:relative to the original image,using traditional methods to process images can increase the image quality indicator value,shadow area surface information has been somewhat restored.(2)It takes advantage of the Retinex algorithm based on RGB color space to deal with the image exsiting shadow,specificly implements Retinex algorithm of single-scale(SSR),multi-scale(MSR)and with color recovery factor(MSRCR)and conducts experiment,through comparing the experimental results of different values of Gaussian fuzzy scale parameters,to explore a relatively reasonable value.Results show:the MSRCR algorithm is applied to the shadow removal of UAV images,although some of the images after processing have a certain color deviation,it have a significant effect in the aspects of the obvious nature of processing effect and the degree of detail recovery,at the same time than the original image,add up the average value of the image quality indexes value,the average gradient,standard deviation,and information entropy raising multiple are as follows:4.5,1.5,0.2.(3)Aiming at the problem of color deviation in the application of Retinex algorithm in RGB color space,it takes advantage of the Retinex algorithm based on HSV color space to deal with the image exsiting shadow,and improves the pixel value quantification method in the algorithm,finally,verifies the feasibility of the algorithm through the feature matching experiment.Combined with the subjective effect and image quality evaluation indicators carries out comprehensive evaluation,results show:it solves the problem of color deviation with a certain extent,effectively recovers the surface information under the cover of the shadow area,at the same time than the original image,add up the average value of the image quality indexes value,the average gradient,standard deviation,and information entropy raising multiple are as follows:16,3.4,0.3.Through the above study,it is shown that the Retinex method can effectively restore the surface information under shadow covering,that is to say,this method applied to the shadow removal of UAV images is feasible.