Fusion Enhancement Method For Optical And SAR Images In Cloudy And Rainy Regions

With the technology of remote sensing sensor continuing to mature,optical remote sensing image in cloudy and rainy area,which effected by cloud begin to cause great attention.SAR image has more structure and texture information and can overcome cloud interference.It is an effective supplement to optical remote sensing data in this area and has important significance for image restoration,cloud data availability improve and classification.Research by summing up the remote sensing image area detection,image restoration and fusion in the field,we put forward a method in cloudy and rainy area using multi-source remote images.The main research results and conclusions are as follows:(1)Design the method of thick cloud detection based on optical image spectral characteristics;achieve the minimum cross entropy cloud shadow recognition processing.And the mathematical morphology open operator is introduced to the remote sensing image noise recognition and extraction process.The results show that the method can achieve the extraction of the cloud better and the mathematical morphology can effectively filter out the small noise and the misjudgment spots in the cloud region recognition.(2)Discuss and realize the method of removing and repairing the cloud noise of the fused optical and SAR images.Through the C# programming to study the regional Landsat8 OLI multispectral image cloud and shadow occlusion area repair.The result shows that the algorithm can integrate the advantages of multi-source data,to achieve a better repair effect.(3)Combine HSV transform with wavelet transform,an improved weighted image fusion method is proposed to realize the enhancement of image.The fusion method is evaluated by means of image mean,standard deviation,information entropy and average gradient.The results show that the fusion method can maintain the details of SAR image,enhance the texture information of multi spectral images,and improve the interpretation ability of the image.(4)By comparing the traditional and object-oriented classification method,the image classification is realized by selecting the better object-oriented method.