GIS and Remote Sensing

Biography

Biography: Volodymyr Hnatushenko

Abstract

Change detection analyses describe and quantify differences between images of the same scene at different times. Change detection is a complex phenomenon which includes different procedures such as identifying the specific change detection problem, image preprocessing and variables and algorithm selection for the computations. The widely used methods for high-resolution image change detection rely on textural/structural features. However, these spatial features always produce high-dimensional data space since they are related to a series of parameters. Moreover, the current urban change detection methods are totally reliant on visual interpretation. This article presents a new automatic change detection method of multitemporal remote sensing high-resolution images and visual interpretation of results. To detect change we apply a series of algorithms, which are independent of each other: subpixel registration of multitemporal images, spectral classification (building masks), singling-out of the most informative stripes and threshold segmenting, morphological filtering and object classification, vectorization and calculation of parameters, visualization of the changes on the map. The candidate changed areas are obtained base on spatial mask filtering, then the spectral difference, searching for spectral-temporal anomalies, morphological technique and a shadow detection method to identify the real changes. Experiments were conducted on the multitemporal Pléiades images. Experimental results show that the proposed method can effectively and quickly extract the changing urban area between the two temporal optical remote sensing images of high spatial resolution. Compared with other change detection methods, the proposed method reduces the effects of classification and segmentation on the change detection accuracy