Shadow Detection Algorithm Based On Multi-Temporal Remote Sensing Image

High-resolution remote sensing images are rich in ground spectral information,and have extensive applications in change detection,land classification,and environmental monitoring.High-resolution satellite images have created a new era in the field of remote sensing.With the increase of image resolution,shadows are commonly found in images and are usually caused by elevated objects(such as buildings and bridges),especially in urban areas.Remote sensing image feature recognition,feature extraction and image classification.Shadows,on the other hand,have their own side.For example,the shadow shape,angle,and shadow block size can be used to infer the shape of the projected object,the intensity of the light source,and the positional information.Therefore,research in the field of shadow detection is particularly important for the subsequent processing of remote sensing images.At present,many scholars are dedicated to improving the detection accuracy of shadow detection in remote sensing images.The main problem is that due to the abundant types of remote sensing imagery,there are many features with similar spectral characteristics in the shadow region,which are difficult to distinguish in the detection and exist brightness.High shadows are easily missed when detected.In view of the above research status,the main research contents of this paper are as follows:First,for the problem of missed detection in non-homogeneous shadow regions and light shadow regions,a shadow detection algorithm combining fuzzy clustering and color space is proposed.The algorithm uses FLICM clustering algorithm and can combine the gray information of the neighborhood.Non-parametric clustering is used to make effective use of the spatial luminance information,and the advantages of the feature channels of the two color spaces of HSV and RGB are combined.Finally,the shadow area is detected.The detection results show that the algorithm can effectively improve the leakage of non-homogeneous shadow areas.Check problems and interference with similar features.Second,for the consideration of the speed and degree of automation of shadow detection algorithms,the concept of Markov random field is introduced to propose a new shadow detection algorithm.First,the PCA algorithm is used to reduce the dimension of the image,enhance the contrast,and then use MRF.The algorithm performs image segmentation.The result of feature segmentation results in less noise and smoothness,and it eliminates the steps of the morphological processing of artificial parameters after the segmentation of common algorithms,improves the speed of the algorithm,and eliminates the steps of manual parameters adjustment.The degree of automation of the algorithm has also been improved.Finally,the high-resolution satellite data is used to test the proposed algorithm,and subjective and objective evaluation of the experimental results are performed.The evaluation results show that in the high-resolution remote sensing images with complex composition of ground features,the two shadow detection algorithms proposed in this paper have good detection accuracy,and the algorithms are respectively in the non-homogeneous shadow and light shadow regions.There are different optimizations in the degree of automation of the algorithm.Visually,it can be seen intuitively that the integrity of the test results is improved.On the other hand,the three evaluation indicators in the quantitative evaluation have improved in different degrees.