Efficient Wavelet-Based Scale Invariant Features Matching

Feature points’ matching is a popular method in dealing with object recognition and image matching problems. However, variations of images, such as shift, rotation, and scaling, influence the matching correctness. Therefore, a feature point matching system with a distinctive and invariant feature point detector as well as robust description mechanism becomes the main challenge of this issue. We use discrete wavelet transform (DWT) and accumulated map to detect feature points which are local maximum points on the accumulated map. DWT calculation is efficient compared to that of Harris corner detection or Difference of Gaussian (DoG) proposed by Lowe. Besides, feature points detected by DWT are located more evenly on texture area unlike those detected by Harris’ which are clustered on corners. To be scale invariant, the dominate scale (DS) is determined for each feature point. According to the DS of a feature point, an appropriate size of region centered at this feature point is transformed to log-polar coordinate system to improve the rotation and scale invariance. To enhance time efficiency and illumination robustness, we modify the contrast-based descriptors (CCH) proposed by Huang et al. Finally, in matching stage, a geometry constraint is used to improve the matching accuracy. Compared with existing methods, the proposed algorithm has better performance especially in scale invariance and blurring robustness. Key-Words: Matching, Discrete Wavelet Transform (DWT), Dominate Scale (DS), Scale Invariance, Log-Polar Transform, Feature Point Descriptor