Study On SAR Image Change Detection Based On Double Difference Image

As a branch of remote sensing image understanding,Remote sensing image change detection is a process in which changes in a geographical surface are determined by analyzing the images acquired at different times in the same geographical location.Synthetic aperture radar(SAR)image change detection has become an important method with the development and application of remote sensing technology.SAR can obtain image data under all illumination and weather conditions.So SAR image change detection has incomparable advantages over traditional optical images.It plays an increasingly important role in disaster assessment,land cover change,man-made target detection and urban change detection.With the acceleration of China's urbanization process,China's cities and rural areas are changed rapidly.Timely detection of ground changes is very important such as in urban management,mobile communication services and so on.In the process of imaging,the remote sensor produces speckles randomly distributed bright or dark spots in remote image,which are mixed with objects.Especially,SAR image is disturbed by speckle noise,which destroys the texture structure of the image and brings difficulties to the change detection of remote sensing image.How to reduce the influence of noise and improve the final detection accuracy is a challenging task.The main purpose of this paper is how to use the characteristics of different image,select appropriate clustering methods,and use preprocessing methods to reduce the impact of noise on change detection,and improve the accuracy of change detection.The three methods are summarized as follows:(1)A SAR image change detection algorithm based on double difference image and mathematical morphology is proposed.First,the multiplicative noise of the SAR image is transformed to additive noise by a logarithmic transformation.Compared to multiplicative noise,additive noise is easier to process.Second,the SAR images are denoised with a morphological filter.Third,the difference image is obtained based on a mean ratio operator and subtraction operator.Then,the final difference image is filtered by median filtering.Finally,the changed area is obtained by the K-means clustering algorithm.(2)A SAR image change detection method based on double difference map and PCA-FCM clustering algorithm is proposed.First,two difference images in the same region were obtained by the log-ratio method and the mean-ratio method.The final difference image is obtained by the log-ratio difference image and mean-ratio image through a fusion rule.Second,the final difference image is divided into nonoverlapping blocks,and Principal Component Analysis(PCA)is used to extract eigenvector of each block.Each pixel in the final difference map is represented by a vector that is projected into the eigenvector space.Finally,the vector of each pixel is classified into change pixel and unchanged pixel using fuzzy C-means clustering.(3)A SAR image change detection algorithm based on gamma enhancement and guided by saliency map with double difference image is proposed.Firstly,the powerlaw enhancement is used to process the two multitemporal SAR images.Then,using the mean ratio operation to obtain the mean ratio difference image,the FLICM is used to obtain the initial detection map.Then,using the salient map of the logarithmic ratio difference image,and the initial detection map is compared with salient map to obtain the final detection result.