A HIGH-PRECISION ELLIPTICAL TARGET IDENTIFICATION METHOD FOR IMAGE SEQUENCES

Image sequences which obtained by close-range videogrammetry, have been widely used in the field of target identification and object tracking. The target identification is one of the key technologies and the corresponding requirement of identification precision are always at a high level in close-range videogrammetry. Therefore, this paper propose a high-precision target identification method for image sequences based on elliptical mark. The proposed approach adopts a coarse-to-fine strategy to identify elliptical mark based on pixel-level detection of Canny method and sub-pixel level identification of Zernike moments, which combines a series of constraint condition methods, thinning method and the least square fitting (LSF). In the process of coarse strategy, the constraint conditions of recursive segmentation method, morphological methods and shape feature are used to remove non-edge points caused by Canny identification. For the fine strategy, the reliable and accurate sub-pixel level edge points are identified by the thinning method and Zernike moments, and sub-pixel center position of elliptical mark are obtained by LSF method. Both simulation experiment and real experiment demonstrated that the proposed method can obtain high precision and reliability results of edge information and center position at a sub-pixel level, and its results are closer to the value by comparing the other two methods in simulation experiment.