Improving the Effectiveness of Remote Sensing Data Correction Using Phase Cross-Correlation Algorithm with Upsampling Factors Variable

Geometric accuracy is one of the problems in using satellite imagery and geometry correction is a necessary step before using the imagery for various applications and analysis. Image-to-image registration is one of the geometric correction methods that can be easily used to find the shifted pixel in satellite imagery. This paper specifically analyzes the phase cross-correlation method implemented in two images with different acquisition times. The goal of this research is to find the impacts of the upsampling factor by adjusting its variable in order to improve its effectiveness in terms of fixing distorted images. Using a study area in Pacitan Regency, two highresolution photos of Pleiades are divided into four polygons, each with four bands. In each band of polygons, the upsampling factor variables are applied sequentially on a scale of 10 - 10.000 to discover the relation between its effectiveness and processing time. The result shows that the higher the upsampling factor, the greater the sub-pixel precision and vice versa. At some point, the greater processing time does not outweigh the higher accuracy. This paper found that the best upsampling factor is 1000 with accuracy up to three digits after decimal points and a processing time of less than 23 seconds. Finding the precise upsampling value can simplify the geometric correction process with image registration. Determining the precise upsampling value will be very helpful for remote sensing data users to carry out the correction process, the highest sub-pixel precision can be achieved with equitable processing time.