|
As for the transformations of translation and rotation exist between sequence images in the image matching navigation system, it is hoped that the algorithm can simultaneously solve the position and heading parameters. Therefore, an algorithm for sequence image matching based on Harris corner and modified hausdorff distance was proposed. In this method, the coordinates of the corners are described in both polar coordinates and Descartes coordinates. Considering the rotational invariance of radius set in polar coordinates, Hausdorff distance is used to match radius sets between sequence images. Then, the rotation angle and the translation distance will be calculated. Experimental analysis has been carried out for the influence of different Gauss noise and the number of Harris corner. Results show that images can still be matched with rotation angle of ${0^ circ }$, ${60^ circ }$, and ${90^ circ }$ under Gauss white noise with 0.01 variance while the position error is less than one pixel, heading error is less than 5 degrees, and matching time is about 4 seconds. The algorithm has robustness and can meet the requirements of real-time and accuracy of sequence image matching for navigation system. |
|
Keywords:sequence image matching; Harris corner; Hausdorff distance; navigation |
|