Band-to-band registration model for near-equatorial Earth observation satellite images with the use of automatic control point extraction

Abstract

The near-equatorial orbit (NEqO) image bands have large differences in distortion, scaling, and skewing. Therefore band-to-band registration of a near-equatorial orbit image is challenging. This study thus proposes a technique to overcome the band-to-band registration of NEqO image bands. The proposed method involves the conversion of image bands into greyscale, followed by image compression and automatic control point (CP) generation through scale-invariant feature transform. CPs are used to perform registration with both first- and second-order polynomials and spline transformation to correct the misregistration between NEqO image bands. The images used in this study are from the Malaysian RazakSAT satellite. Results show accurate registration performance. Moreover, the root-mean-square errors of the first- and second-order polynomial transformations are 4 and 3 m, respectively. These values are lower than those obtained with the use of spline transformation. Verification is performed by comparing the results of both automatic and manual transformations to the use of polynomial transformation. The results confirm the effectiveness of the proposed algorithm.