Unsupervised Satellite Change Detection Using Particle Swarm Optimisation in Spherical Coordinates
UŞAKLI, Ali Bülent
MetadataShow full item record
This paper proposes a new unsupervised satellite change de- tection method, which is invariant to shadow and shading e ects. To achieve this, rstly, the RGB satellite images are transformed into spher- ical colour space to remove illumination artifacts. Then, a new unsuper- vised change detection is used. The resultant optimal binary change mask is obtained by minimisation a mean square error based cost function us- ing Binary Particle Swarm Optimisation (BPSO). The proposed method is compared with three other satellite change detection methods and the results demonstrate that our method provides a signi cant improvement of obtaining the changed and unchanged regions on the diference image. The total errors show that our method at least 39.62% better than the best compared method and are utmost 176.18% better than the worst one.
DSpace@Karatay by Karatay University Institutional Repository is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 Unported License..