Aerospace Defense

Use of Correlation Coefficients to Provide an Optimal Algorithm of Template Matching, Invariant to Scale and Angle with Implementation on FPGA

Abstract
Template matching techniques are widely used today in image-based tracking systems to identify and track military targets. Due to their passivity, these systems are resistant to common electronic warfare techniques. One of the problems with using template matching algorithms is that they are slow, especially when rotation and scaling occur in targets relative to a predetermined pattern. In this paper, we perform an optimized algorithm for grayscale template-matching based on correlation coefficients which is invariant to scale and angle. The ‘brute force’ algorithm performs template-matching between the image to analyze and the template query shape rotated by specific angle, translated to specific scale. This takes too long and thus is no practical. The optimized algorithm includes three cascaded filters for scale, angle and template matching which results in probability of scaling, rotated angle for each pixel and point of template-matching, respectively. This algorithm accelerates searching and is 400 times faster than ‘Brute Force’ algorithm. By implementing hardware on the FPGA, it is possible to identify and track military targets at a speed of 10 frames per second.

Keywords

Template Matching
Invariant to Scale and Angle
Tracing
Correlation Coefficients
[1] N. Mayasari and A. P. Siahaan, “Vehicle Plate Recognition using Template Matching” International Journal For Innovative Research In Multidisciplinary Field , Vol 4, pp.259-263, 2018.
[2] S. Caraiman et al., “Computer Vision for the Visually Impaired: the Sound of Vision System,” in 2017 IEEE International Conference on Computer Vision Workshops (ICCVW), pp. 1480–1489,2017.
[3] S. Nikan and M. Ahmadi, “Partial Face Recognition Based on Template Matching,” in 2015 11th International Conference on Signal-Image Technology & Internet-Based Systems (SITIS), pp. 160–163. 2015.
[4] F. Essannouni and D. Aboutajdine, “Fast Frequency Template Matching Using Higher Order Statistics,” IEEE Trans. Image Process., Vol. 19, No. 3, pp. 826–830, 2010.
[5] M. Suzuki, Y. Tanida, T. Maruyama, “Detecting patterns in various size and angle using FPGA”, Int. Conf. on Field Programmable Logic and Applications, pp.151-154, 2010.
[6] W. Chia Lee, C. Hsing Chen, “A Fast Template Matching Method Rotation Invariance Using Two Stage Process” , Int. Conf. on Intelligent Information Hiding and Multimedias Signal Processing, pp.109-112, 2009.
[7] L. Ming, j.Z. Guang, “A novel algorithm for a rotation invariant template matching”, Optoelectron. Lett, Vol. 4, No 5, pp. 379-383, 2008.
[8] J.Flusser, T.Suk, “Rotation moment invariant for recognition of symmetric objects” IEEE T.Image Processing, Vol. 15, No 12, pp.3787-3790, 2006.
[9] C.R.P. Dionisio, H.Y. Kim, “A supervised shape classification technique invariant under rotation and scaling”, Int. Telecommunications Symposium, pp.533-537,2002.
[10] H.Y. kim, S.A. Araujo, “grayscale Template Matching Invariant to Rotation, Scale, Translation, Brightness and Contrast” IEEE pacific-Rim Symposium on Image and Video Technology. Leacture Notes in Computer Science, Vol. 4872, pp.100-113, 2007.
[11] Y. Tanida, T. Maruyama, “An Approach for Downscaling Images for Real-time Pattern Detection”, pp.253- 256 2008.
[12] M. Arif and K. Sohaib, “Correlation-Coefficient-Based Fast Template Matching Through Partial Elimination” IEEE Signal Processing Society. Volume: 21, Issue: 4, April 2012
[31] A. Lindose, L. Entrena, “High performance FPGA-based image correlation”, J.Real-time Image Proc, pp.223-233, 2007.
[14] R. C. Gonzalez and R. E. Woods, Digital Image Processing, Englewood Cliffs, NJ: Prentice Hall, 2014.

  • Receive Date 11 February 2022
  • Revise Date 31 May 2022
  • Accept Date 21 August 2022

Home

Submit Manuscript

Reviewers

Contact Us