AUTOMATIC POSITION OF FIDUCAL MARKS BY THE IMAGE-MATCHING TECHIQUE

Abstract

The paper was a study of automatically positioning of 8 fiducial marks on the digital aerial photograph using image matching technique. Four methods of finding center of fiducial marks were summarized and evaluated, namely, manual locating, center of gravity, correlation coefficient, and least square methods. All methods were then implemented and tested on 8-bit gray-scale images with the pixel size of 15, 30 and 60 micron. The positions given from these methods were integer and decimal. The research was also to develop a computer program using C language to calculate the coordinate by center of gravity, correlation coefficient, and least square methods. Then the performances of each were compared considering of precision, time consuming, and reliability. The results revealed that when using implemented program to calculate the fiducial marks on the digital photo. By using center of gravity method as reference, the center of fiducial marks from least square methods were within 0.2 pixel. The method of correlation coefficient with sub-pixel accuracy type were within 0.4 pixel. For time consumption, the center of gravity method took 1.5 seconds per 8 fiducial marks which was the fastest, while the correlation coefficient method and least square methods needed 1.7 and 4.2 seconds, respectively. In term of reliability, the coordinate calculated from center of gravity method was mostly closed to those from manual locating method and was assumed to be the most reliable.