Research On Accurate Shadow Detection And Compensation Method Of UAV Remote Sensing Images

Unmanned Aerial Vehicle(UAV)remote sensing has become a powerful supplement to satellite remote sensing and manned aerial vehicle remote sensing due to its significant advantages such as flexible detection range,rich monitoring means and efficient data acquisition.At present,it is widely used in farmland monitoring and management,mineral resources exploration,natural disaster warning,urban construction planning and other fields.However,due to the compound effect of solar illumination,ground reflection and atmospheric disturbance,UAV remote sensing images generally have the problem of low authenticity of local color features and texture features due to the blurred details or information loss in shadow areas of images.Furthermore,the implementation and accuracy improvement of subsequent UAV remote sensing tasks such as urban ground feature classification,crop growth monitoring and geographic mapping are severely restricted.The application conditions of existing shadow processing methods are limited.For UAV remote sensing images with a wide variety of ground objects and various shadow types,shadow detection methods are difficult to extract irregular and fragmented shadows,and shadow compensation methods cannot simultaneously take into account color and texture compensation leading to color distortion and texture blurring in shadow areas.Aiming at the irregular and fragmented shadow regions in UAV remote sensing images,a shadow detection method based on significant enhancement of differences is designed in this paper.Furthermore,a shadow compensation method based on texture-preserving local color transfer is proposed to achieve accurate compensation of color and texture information in shadow regions of the UAV remote sensing image.The research work mainly includes the following four aspects:(1)According to the space optical remote sensing imaging mechanism and radiation transmission theory,the radiation sources and influencing factors in shadow regions of UAV images are analyzed,and the causes and types of shadow formation are clarified.The spectral characteristics and imaging characteristics of shadow region and non-shadow region of UAV image are explored,and the characteristics of typical color space are analyzed.Starting from typical shadow detection and compensation methods,the principle of shadow detection and compensation is studied.(2)Aiming at the problem that existing shadow detection algorithms are difficult to extract irregular and fragmented shadows,a shadow detection method based on significant enhancement of differences for UAV remote sensing images is proposed.Combined with the spectral characteristics of shadow/non-shadow region of UAV image,the difference shadow detection index of the visible band of UAV image is constructed.The index image is segmented by the maximum inter-class variance method to obtain the shadow detection results.The accuracy and robustness of the proposed method are verified by subjective visual evaluation and objective quantitative evaluation of shadow detection results on different regional experimental data sets.(3)In order to solve the problem of "oversaturation" of pixels on the compensated shadow boundary due to the fact that the existing methods cannot accurately describe the dynamic range of the penumbra,a pixel-by-pixel width penumbra compensation method is proposed.Based on the dynamic variation of penumbra illumination intensity in UAV remote sensing images,the penumbra range is quantitatively described and the information of penumbra region is recovered pixel-by-pixel width by combining with the texture-preserving local color transfer method.The effectiveness of the proposed method is verified by qualitative and quantitative analysis.(4)In order to maintain texture information meanwhile to accurately compensate the color information in shadow regions of an image,a shadow compensation method for UAV images based on texture-preserving local color transfer is proposed.Firstly,homogeneous regions are extracted from UAV images and homogeneous subregion segmentation is performed on the basis of homogeneous regions using the mean shift method.Secondly,in combination with the shadow mask,each shadow subregion is matched with the corresponding non-shadow subregion based on its texture features and spatial distance.Then,the matched non-shadow subregion is used as the reference region,and the color transfer based on preserving texture is performed on the shadow subregion.Finally,pixel-by-pixel width shadow compensation is applied to the penumbra region.Through qualitative and quantitative analysis,the accuracy of the proposed method in color compensation and texture detail compensation of shadow regions is verified.