ANALYSIS OF EFFICIENCY OF USE OF HARRIS AND KANADE–LUCAS–TOMASI DETECTORS FOR VISUAL NAVIGATION TASKS

The article examines methods of images analysis based on computer vision. We made a comparison between the detectors of feature points determined by Harris and Kanade–Lucas–Tomasi (KLT) methods. Found points are represented by Speed-Up Robust Feature (SURF) descriptor and then used to determine homography matrix. Analyses of accuracy of visual navigation is done by estimation of a camera rotation angle via factorization of homography matrix obtained from two detector methods. Errors of visual navigation follow the normal distribution for the given sample.