PTGC: A parallel triangular geometric correction algorithm for remote sensing images

Geometric correction with the compute-intensive characteristic is the critical step in the processing of remote sensing image. As is known, the correction based on polynomial transform will lead to significant computing errors when applied to the region with fluctuant terrain. In this paper we present a novel and more precise parallel triangular geometric correction algorithm to deal with the image of the fluctuant terrain with large amount of data. In the algorithm, dynamic cut-off points are adopted to achieve good load balance, and the triangle-scan-line structure is utilized to preserve the index of the triangle, which is hashed by the pixel. Hence, the redundant calculation is avoided. The algorithm is equipped with good scalability, and is able to handle arbitrary geometric distortion. The accuracy of our approach is approximate 80% higher than the polynomial geometric correction algorithm with similar computational overhead.