Shadow Detection And Removal With Single Image

Due to the straightforward nature of the light,Shadows are created when the light source is blocked by an opaque object.It will lead to color change,brightness decrease,or even change the image texture.Therefore,shadow removal is important for enhancing image visibility,this paper focuses on the shadow detection and shadow removal of a single image.The specific work includes:1.Aiming at the problem of distinguishing between shadow area and dark area in single image shadow detection,a shadow detection method based on color space clustering is proposed.Firstly,the image is decomposed into a uniform region with sharp edges and detail layers,and the image is enhanced by enlarging the detail layer.Secondly,the input classification mean clustering is obtained by RGB pixel value training KNN classifier to obtain the initial shadow mask;finally,in order to weaken the image in the image Texture noise and dark areas are removed,and YUV fused images and color ratios are used to obtain the final test results.The method can distinguish the shadow area from the dark area,and can avoid the influence of complex texture on the detection result to some extent.Experiments show that the method can detect the shadow area more accurately.2.Aiming at the effect of the penumbra region on the removal effect and the inconsistency of the image texture after removing the shadow,the shadow removal method based on.penumbra region illumination restoration and color correction is proposed.In this paper,the shadow boundary is firstly obtained according to the shadow mask and sampled according to the curvature of the boundary.The penumbra region is obtained by smooth interpolation and extrapolation,and the pixel variation law.Secondly,the penumbra region and the umbral region are respectively restored by illumination;Finally,iteratively aligns the shadow boundaries on each scale to remove inconsistencies between the illuminated and shaded areas,resulting in a consistently shadowed image with illumination..Experiments show that the proposed method not only does not leave artificial marks on the edge portion due to the penumbra area,but also removes the shadowed image to have a consistent effect on chromaticity,brightness and texture.3.According to the above research content,a single image shadow detection and removal system is designed and implemented.The system integrates the shadow detection and removal methods and the related classical algorithms.