Change Detection in Ottawa City Synthetic Aperture Radar Remote Sensing Images based on DWT Fuzzy C Means Clustering

Abstract

Abstract: The radiometric, spectral, spatial and temporal resolutions of remote sensing data have a major effect on the success rate of the applications that deploy remote sensing applications. In the traditional works, the issues related to remote sensing images are usually signified for particular kinds of sensors (that is, active or passive). While deploying passive sensors in remote sensing(e.g., optical images), differentiated image is generally calculated and a suitable measure of change is provided. On the other hand, in rainy or cloudy regions, the exploitation of optical images tends to be limited. From this viewpoint, Synthetic Aperture Radar (SAR) imaging can be regarded as the most excellent substitute for remote sensing applications. In this regard, two multitemporal SAR images have been taken for this fusion process. Usually, the major need of image fusion is to extract the information from multiple images and convert them into one image with all information in individual images. To perform this image fusion multiple wavelet coefficients have been applied in Discrete Wavelet Transform (DWT) such as Daubchies wavelet coefficients (daubchies2) have been used. After performing this fusion, fused image is segmented as changed regions and unchanged regions based on Fuzzy C Means clustering (FCM). the segmented image is compared with the ground truth image the outcome is measured in terms measuring parameters such as accuracy, precision, sensitivity and FDR.