Research on CA Differencing for Remote Sensing Change Detection

Abstract

To improve the precision of land-cover change detection and remedy the shortage of single change detection technique, an integrated correspondence analysis (CA) differencing method was developed, which combined CA method with image differencing. After the spectral transformation of the individual date images into component space using CA, the first component (PC1) of the date 1 image was subtracted from the PC1 of the date 2 image to produce difference image highlighting change areas. The experimental results that the integrated CA differencing can restrain the noise interference and better automatically detect land-cover change, whose precision was respectively increased by 4.38%, 1.24% and 1.88% relative to the conventional principal component analysis (PCA) differencing, multi-band PCA and spectrum feature variance. By this token, the CA differencing can be applied to change detection and deserved further extensive applications.