Shadow Detection And Removal Of Tall Buildings In Remote Sensing Images

Building shadows are common in high resolution remote sensing images.On the one hand,the presence of shadows results in a reduction in the amount of remote sensing image information,and causes the degradation of the visual effects of the image,which need to be detected and removed to meet the needs of advanced applications;on the other hand,it can be extracted through the detection of shadows.Information such as building outlines,heights,etc.extends the potential of the image.In this paper,the existing shadow detection method of dual-threshold shadow detection method and shadow detection method based on shadow index SI are studied and simulated.It is found that the above methods can perform accurate building shadow detection under specific application conditions.However,there are also the following problems:?1?The features of the water and shadows on the image are similar.Both have the characteristics of high saturation and low brightness values.When there is a large amount of blue-colored water in the non-shaded area,it is easy to confuse the shadow area with the water body.?2?Using the characteristics of the brightness of the ground features,the ground and water bodies and shadows with high threshold brightness values,and artificially determining the threshold,the experimental results are based on the assumption that the brightness values of all shadows and water bodies are lower than the ground features with high brightness values,and objective Lack of objectivity.?3?When the shadow of a building is projected onto the surface of a water body,it is difficult to detect the shadow of a building on the surface of the water body by the above method,and there is a phenomenon of missed detection.In order to solve the above problems,this paper proposes a shadow detection method based on the principal component transformation and spectral features of HSI space,combined with the method based on shadow probability constraints,to successfully detect the presence of tall building shadows on the water surface.This paper proposes a shadow detection method based on principal component transformation and spectral features of HSI space,combined with a method based on shadow probability constraints.According to the characteristics of shadow low brightness and high saturation in HSI space,combined with the difference of the first principal component?PCA₁?value after the main component transformation of different ground objects,the shadow index is constructed;the concept of shadow probability is combined with the shadow of water body and The surrounding water area is subject to the shadow probability constraint,and a complete shadow detection result is finally obtained.The shadows of the tall buildings on the surface of the water body are successfully detected.Experimental results show that the proposed method is superior to the existing dual-threshold shadow detection method and shadow detection method based on the shadow index SI,and can not only detect the shadows of tall buildings existing on the ground,but also successfully detect the Tall buildings shadows on the surface of the water.On the basis of the existing shadow detection method can better locate the shadow,the paper studies the shadow removal method.This paper presents an existing shadow removal algorithm based on histogram equalization.Based on the gamma-shift shadow-removal algorithm and the multi-scale Retinex?MSR?-based shadow-removal algorithm,the simulation results show that the above method has problems in color shift and edge processing.In this paper,an adaptive regularized shadow removal algorithm is used to perform shadow compensation in the intensity domain while keeping the information in the non-shaded region.Using a regularized energy function,it includes a fidelity term and a regularization term.The shadow region information is compensated,and the shadow edge region is regularized,and the complete remote sensing image shadow removal result is finally obtained.The regularized shadow removal method proposed in this paper increases the brightness of the shadow area of the image and increases the brightness uniformity.The shaded area's hue,brightness and texture details have been restored.Under the premise of ensuring that the information in the non-shaded areas is not changed,the information in the shadow areas is effectively restored and the shadows of the tall buildings are removed.The regular term smoothes the boundary traces,smoothing the transition between the non-shaded and shaded areas of the image.