A novel approach for scale and rotation adaptive estimation based on time series alignment

摘要

This paper proposes a novel approach for target scale and rotation adaptive estimation based on template matching, which is robust to undergo brightness, contrast invariance, and noise corruption. Improved features based on ring projection transform are extracted, which can not only improve the matching ability of some special scenes by taking into account changes of pixel intensity and structure information, but also automatically recommend the sampling rings involved in the angle estimation. Moreover, treating image features from the perspective of signal time series, we have designed a hierarchical adaptive estimation strategy to solve the problem of scale invariance while reconstructing the transformation of brightness and contrast. Eliminating the limitations of the pre-prepared fixed-scale vertex template, the proposed approach implements an adaptive estimation of the scale. Additionally, rotation angle calculation based on the normalization cross-correlation can be used as the secondary verification of the candidate solution to further improve the matching accuracy. Numerical evaluation shows that the method enjoys attractive results.