Research On Shadow Detection And Compensation Algorithm For High Resolution Remote Sensing Image

In recent years,with the rapid development of high-resolution satellite industry,the application of high-resolution remote sensing images in geographical conditions monitoring,urban planning,land resource management,and watershed ecological environment monitoring has been continuously developed and deepened.However,although the solar elevation angle and season are constantly changing,shadows of high-resolution satellite images are common due to the satellite orbital parameter settings.The existence of shadows has seriously affected the automation,intelligent identification and application of images.Therefore,in recent years,researchers have done a lot of work on shadow detection and compensation in high-resolution remote sensing images,and have achieved a series of results.However,through literature and experimental analysis,it is found that the existing shadow detection algorithms are mostly for single or specific image types,lacking universality,and it is difficult to extract the shadow transition region(penumbra)information and effectively remove the mis-extracted water information.In view of the above problems,this paper proposes two universal automatic shadow detection algorithms.(1)An automatic shadow detection method for high resolution remote sensing imagery based on polynomial fittingBy using polynomial fitting instead of Otsu for the near-infrared band histogram to determine the shadow segmentation threshold,the method can get a more reasonable threshold,improve the extraction accuracy.In our method,the normalized water index(NDWI)is applied to extract water bodies from remotely sensed images.Instead of utilizing the logical AND operation to remove the water bodies that are falsely detected as shadows,the new method applies a scan line seed filling algorithm,which enables complete and accurate removal of bodies of water.Therefore,first,the histogram of the whole image is fitted by the fourth and fifth degree polynomials according to the histogram difference of the near-infrared bands of different shadow areas in the remotely sensed image.Second,the shadow area is preliminarily extracted based on the relationships between the shadow features of the remote sensing image and the intersections of the fourth and fifth degree polynomials.Then,the NDWI is applied to extract the water bodies.Finally,to obtain the shaded area,the scanning line seed filling algorithm is applied to remove the water bodies falsely detected as shadows in the preliminary shadow detection result.(2)Automatic expansion extraction algorithm of remote sensing imagesBy analyzing the pixel value features of the shadow boundary of various images in the near-infrared band,it is found that the fluctuation rate of the pixel value of the shadow locality is small and the peak value of the pixel value of the boundary has a peak.By using this feature,the shadow boundary judgment criterion is established.Each shadow is expand by the criterion from the inside-out,which not only can take into account a single shadow area,but also is no longer limited to the global image features or local features of remote sensing images,so that shadow is extracted more complete and the transition area is also extracted.Finally,the shadow compensation algorithm is explored,and the current conventional shadow compensation algorithms are introduced.The defects of most shadow compensation algorithms and the causes are analyzed in detail.At the same time,the shadow compensation algorithm is prospected.The proposed algorithm is evaluated using various high-resolution images including GF-1,GF-2,QuickBird2,and ZY-3.On the one hand: compared with multi-elements extraction algorithm,multi-band detection algorithm,spectral correlation algorithm based on spectral features and C3 component algorithm,the proposed shadow detection algorithm based on polynomial fitting has the lowest false detection rate,and has high accuracy and good performance.And it can completely remove water bodies with 98.90% water removal rate.On the other hand,compared with the multi-peak histogram threshold extraction algorithm and artificial experience detection results,the proposed shadow expansion algorithm has lower missed detection rate and high accuracy,and the ability to extract shadow transition areas.