Projection method for geometric modeling of high resolution satellite images applying different approximations

  • Authors:

  • Fahim Arif;Muhammad Akbar;An-Ming Wu

  • Affiliations:

  • Computer Science Department, College of Signals, National University of Sciences and Technology, Rawalpindi, Pakistan;Chief Instructor (Engineering Division), College of Signals, National University of Sciences and Technology, Rawalpindi, Pakistan;National Space Organization, Hsinchu, Taiwan

  • Venue:
  • PSIVT'06 Proceedings of the First Pacific Rim conference on Advances in Image and Video Technology
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Precise remote sensing and high resolution satellite images have made it necessary to revise the geometric correction techniques used for ortho-rectification. There have been improvements in algorithms from simple 2D polynomial models to rigorous mathematical models derived from digital photogrammetry. In such scenario, conventional methods of photogrametric modeling of remotely sensed images would be insufficient for mapping purposes and might need to be substituted with a more rigorous approach to get a true orthophoto. To correct geometric distortions in these, the process of geometric modeling becomes important. Pixel projection method has been devised and used for geometric correction. Algorithm has been developed in C++ and used for FORMOSAT-2 high resolution satellite images. It geo-references a satellite image while geolocating vertices of the image with its geo-locations extracted from ancillary data. Accuracy and validity of the algorithm has already been tested on different types of satellite images. It takes a level-1A image and the output image is level-2 image. To increase the geometric accuracy, a set of ground control points with maximum accuracy can also be selected to determine the better knowledge of position, attitude and pixel alignment. In this paper, we have adopted different techniques of approximations and applying three possible methods of interpolation for transformations of image pixels to earth coordinate system. Results show that cubic convolution based modeling gives best suitable output pixel values while applying transformation.