GWDWT-FCM: Change Detection in SAR Images Using Adaptive Discrete Wavelet Transform with Fuzzy C -Mean Clustering
- PDF / 1,485,709 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 87 Downloads / 171 Views
RESEARCH ARTICLE
GWDWT-FCM: Change Detection in SAR Images Using Adaptive Discrete Wavelet Transform with Fuzzy C-Mean Clustering Thrisul Kumar Jakka1 • Y. Mallikarjuna Reddy2 • B. Prabhakara Rao1 Received: 12 June 2018 / Accepted: 6 November 2018 / Published online: 30 November 2018 Ó Indian Society of Remote Sensing 2018
Abstract Change detection in remote sensing images turns out to play a significant role for the preceding years. Change detection in synthetic aperture radar (SAR) images comprises certain complications owing to the reality that it endures from the existence of the speckle noise. Hence, to overcome this limitation, this paper intends to develop an improved model for detecting the changes in SAR image. In this model, two SAR images captivated at varied times will be considered as the input for the change detection process. Initially, discrete wavelet transform (DWT) is employed for image fusion, where the coefficients are optimized using improved grey wolf optimization (GWO) called adaptive GWO (AGWO) algorithm. Finally, the fused images after inverse transform are clustered using fuzzy C-means (FCM) clustering technique and a similarity measure is performed among the segmented image and ground truth image. With the use of all these technologies, the proposed model is termed as adaptive grey wolf-based DWT with FCM (AGWDWT-FCM). The similarity measures analyze the relevant performance measures such as accuracy, specificity and F1 score. Moreover, the performance of the AGWDWT-FCM in change detection model is compared to other conventional models, and the improvement is noted. Keywords Synthetic aperture radar Adaptive discrete wavelet transform Filter coefficient Grey wolf optimization Fuzzy C-means clustering
Introduction SAR imagery exists as a significant data source for the applications in remote sensing. In spite of the actuality that remote sensed imagery could be portrayed more easily by a human operator, it can only perform throughout daytime, and therefore, it is subjected to various climatic conditions (Zheng et al. 2017; Cui et al. 2016; Gong et al. 2012). Consequently, an existing SAR sensor describes an & Thrisul Kumar Jakka [email protected] Y. Mallikarjuna Reddy [email protected] B. Prabhakara Rao [email protected] 1
Jawaharlal Nehru Technological University, Kakinada 533003, Andra Pradesh, India
2
Vasireddy Venkatadri Institute of Technology, Nambur, Guntur, India
obligatory source for the entire weather and 24-h imagery with a predetermined revisit cycle at insistent high resolution. The process of change detection from SAR images comprises two SAR images attained over the similar geological area with almost similar acquisition characteristics at two different times to plot the areas where alterations take place among the two acquisition stages (Yang et al. 2014; Vu et al. 2017; Jia et al. 2015). Images attained by this method can be exploited for diverse applications such as detection of objects, fundamental radar functionalities and their geological localizatio
Data Loading...
