Zero-aliasing correlation filters

Traditional correlation filters are designed and implemented via the frequency domain, where the correlation of two signals may be computed efficiently. However, when the discrete Fourier transform (DFT) of length N is used, multiplication in the frequency domain results in an N-point circular correlation, rather than a linear correlation. The resulting correlation filter output is therefore corrupted by the aliasing effects of circular correlation. One solution is to design and implement the correlation filter directly in the space domain. However, this is more computationally intense. Recent literature has discussed ways in which to minimize circular correlation effects, but the effects are not completely removed. We propose a new frequency domain method for completely eliminating circular correlation effects when designing correlation filters. We demonstrate this idea with the well-known minimum average correlation energy (MACE) filter and show how the reformulated MACE filter in the frequency domain outperforms the original formulation of the MACE filter.

[1]  B. V. K. Vijaya Kumar,et al.  Space-time correlation filters for human action detection , 2013, Electronic Imaging.

[2]  Abhijit Mahalanobis,et al.  Unified framework for the synthesis of synthetic discriminant functions with reduced noise variance and sharp correlation structure , 1990 .

[3]  B. V. K. Vijaya Kumar,et al.  Dealing with circular correlation effects , 2013, Defense, Security, and Sensing.

[4]  Bruce A. Draper,et al.  Simple real-time human detection using a single correlation filter , 2009, 2009 Twelfth IEEE International Workshop on Performance Evaluation of Tracking and Surveillance.

[5]  G.B. Moody,et al.  The impact of the MIT-BIH Arrhythmia Database , 2001, IEEE Engineering in Medicine and Biology Magazine.

[6]  Andy Harter,et al.  Parameterisation of a stochastic model for human face identification , 1994, Proceedings of 1994 IEEE Workshop on Applications of Computer Vision.

[7]  B. V. K. Vijaya Kumar,et al.  Maximum Margin Correlation Filter: A New Approach for Localization and Classification , 2013, IEEE Transactions on Image Processing.

[8]  Mubarak Shah,et al.  Action MACH a spatio-temporal Maximum Average Correlation Height filter for action recognition , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[9]  D. Casasent,et al.  Minimum average correlation energy filters. , 1987, Applied optics.

[10]  Bruce A. Draper,et al.  Average of Synthetic Exact Filters , 2009, CVPR.

[11]  B. V. K. Vijaya Kumar,et al.  Correlation Pattern Recognition , 2002 .

[12]  Marios Savvides,et al.  Correlation Pattern Recognition for Face Recognition , 2006, Proceedings of the IEEE.