Change Detection In SAR Images Based On Novel Non-local Means And Clustering Methods

Change detection of synthetic aperture radar (SAR) images of the same scene taken atdifferent times is of widespread interest in the change detection field. In this paper, adeeply study is carried out for the SAR image change detection technique based on thegeneration of difference image and the analysis of difference image.First, a novel change detection approach based on non-local means algorithm isproposed for synthetic aperture radar (SAR) images. Non-local means technique isintroduced to generate a difference image by using information from the whole pair ofimages. In order to take the characteristics of SAR image into account, we propose anew relativity measurement between two speckled SAR image patches based on a ratiodistance, which is valid for SAR images. The probability density function of the ratiodistance is used to map the distance into a relativity value. Furthermore, the ratiodistance and the probability density function are both parameter-free. The new non-localmeans technique is successfully applied to extend the classical mean ratio detector forSAR image detection. The experimental results on real SAR data confirm theeffectiveness of the proposed algorithm.A novel SAR image change detection approach based on an improved fuzzyc-means clustering (FCM) with the introduction of PCA non-local means. In this paper,we introduce a PCA non-local means which use principle component analysis to reducethe dimensionality of non-local means to improve the fast generalized fuzzy c-meansclustering (FGFCM). PCA non-local means image is used to replace the image of localgray and spatial information of FGFCM algorithm. It is used to reduce the effect ofspeckle noise and improve cluster performance. Experimental results on real SAR dataconfirm the effectiveness of the proposed algorithm.