Abstract
ABSTRACT In the present study, an improved iteratively reweighted multivariate alteration detection (IR-MAD) algorithm was proposed to improve the contribution of weakly correlated bands in multi-spectral image change detection. In the proposed algorithm, each image band was given a different weight through single-band iterative weighting, improving the correlation between each pair of bands. This method was used to obtain the characteristic difference in the diagrams of the band that contain more variation information. After removing Gaussian noise from each feature-difference graph, the difference graphs of each band were fused into a change-intensity graph using the Euclidean distance formula. Finally, unsupervised fuzzy C-means (FCM) clustering was used to perform binary clustering on the fused difference graphs to obtain the change detection results. By comparing the original multivariate alteration detection (MAD) algorithm, the IR-MAD algorithm and the proposed IR-MAD algorithm, which used a mask to eliminate strong changes, the experimental results revealed that the multi-spectral change detection results of the proposed algorithm are closer to the actual value and had higher detection accuracy than the other algorithms.
Highlights
Remote sensing image change detection technology has been applied in many fields, such as environmental monitoring (Zhuang, Deng, & Fan, 2016), urban research (Zhuang, Deng, Yu, & Fan, 2017), land use (Yonezawa, 2007), sand cover monitoring (Yan-Hong, Pei, Wang, & Yun-Peng, 2010), forest monitoring (Zhuang, Deng, Fan, & Ma, 2018), agricultural investigation (Shi, Gao, & Shen, 2016), and disaster assessment (Chen & Chen, 2016)
Because the noise in a multi-spectral image is mainly additive noise, the Gaussian denoising algorithm is selected in this paper for the simple denoising of each feature-difference graph. (The comparison algorithm in this paper includes the processing of Gauss filtering.) After Gaussian noise removal of each feature-difference graph, the difference graphs of each band were fused into a change-intensity graph using the Euclidean distance formula
The objective indicators include the number of false negatives (FN), the number of false positives (FP), the overall error (OE), the percentage correct classification (PCC) (Gao, Liu, Dong, Zhong, & Jian, 2017), and the Kappa coefficient (KC) (Rosin & Ioannidis, 2003)
Summary
Remote sensing image change detection technology has been applied in many fields, such as environmental monitoring (Zhuang, Deng, & Fan, 2016), urban research (Zhuang, Deng, Yu, & Fan, 2017), land use (Yonezawa, 2007), sand cover monitoring (Yan-Hong, Pei, Wang, & Yun-Peng, 2010), forest monitoring (Zhuang, Deng, Fan, & Ma, 2018), agricultural investigation (Shi, Gao, & Shen, 2016), and disaster assessment (Chen & Chen, 2016). The change detection of remote sensing images is based on the multiple remote sensing images acquired at different time points in the same region to extract the features and process the changes in the ground objects. Because there are different structures and components between the objects, different features have different spectral characteristics, which means that the reflection spectra of different features are different. If the reflectance spectra of different objects are similar in some bands, the reflectance spectra of these objects in other bands will greatly differ. Single-band remote sensing image change detection can identify an object in a band but cannot extract the features of other wavelength change information. The multi-spectral remote sensing images of multiple wavelengths can reflect the characteristics of features under different wave bands and make good use of the spectral correlation
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
