Multi-target optimization SAR image change detection method based on integrated learning

TL;DR: In this paper, a multi-target optimization SAR image change detection method based on integrated learning is proposed to solve problems of large spot noise, incapability in keeping the local information of SAR images and low classification accuracy in the prior art.

Abstract: The invention discloses a multi-target optimization SAR image change detection method based on integrated learning and mainly solves problems of large spot noise, incapability in keeping the local information of SAR images and low classification accuracy in the prior art. The method comprises steps that a fluctuation parameter division method is employed to generate an original difference graph D1 for two inputted SAR images Y1 and Y2; de-noising of the original difference graph D1 is carried out to acquire a de-noising difference graph D2; two target functions are constructed based on the original difference graphs D1 and the de-noising difference graph D2, a solution set of the two target functions having the smallest function values is acquired through calculation, and multiple binary images are acquired; an integrated learning method is further utilized to acquire a final change detection image R based on the multiple binary images. The method is advantaged in that spot noise is reduced, the image local information is kept, classification accuracy is improved, and the method can be applied to remote sensing, medical diagnosis and video monitoring.