Template matching using the parametric template vector with translation, rotation and scale invariance
Yi-Hsien Lin,Chin-Hsing Chen
DOI: https://doi.org/10.1016/j.patcog.2008.01.017
IF: 8
2008-07-01
Pattern Recognition
Abstract:A new method for the template matching, invariant to image translation, rotation and scaling, is proposed. In the first step of our approach, the ring-projection transform (RPT) process is used to convert the 2D template in a circular region into a 1D gray-level signal as a function of radius. The advantages of the RPT process are that it owns the characteristic of rotation invariance and reduces the computational complexity of normalized correlation (NC). Then, the template matching is performed by constructing a parametric template vector (PTV) of the 1D gray-level signal with differently scaled templates of the object. The merits of our approach are that it not only obtains rotation invariance, but also scale invariance Additionally, our approach is conceptually simple, easy to implement and only one training image is needed for the training phase. Experimental results show that the computational time of the proposed approach is faster and the performance is better than the parametric template (PT) method and the affine moment invariants (AMIs) method in the image rotation or scaling. Moreover, our approach not only enjoys high accurate rate under the changes of translation, rotation and scale, but also estimates the scaling value of the target object in the input scene. Experiments with Gaussian noise demonstrate that the proposed algorithm is robust to detect the target object with the changes of translation, orientation and scale. This indicates that our approach is suitable for on-line template matching with scene translation, rotation and scaling.
computer science, artificial intelligence,engineering, electrical & electronic