{
"title": "A Modified Cross Correlation in the Frequency Domain for Fast Pattern Detection Using Neural Networks",
"authors": "Hazem M. El-Bakry, Qiangfu Zhao",
"country": null,
"institution": null,
"volume": "11",
"journal": "International Journal of Computer, Electrical, Automation, Control and Information Engineering",
"pagesStart": 3754,
"pagesEnd": 3761,
"ISSN": "1307-6892",
"URL": "http:\/\/waset.org\/publications\/2525",
"abstract": "Recently, neural networks have shown good\nresults for detection of a certain pattern in a given image. In\nour previous papers [1-5], a fast algorithm for pattern\ndetection using neural networks was presented. Such\nalgorithm was designed based on cross correlation in the\nfrequency domain between the input image and the weights\nof neural networks. Image conversion into symmetric shape\nwas established so that fast neural networks can give the\nsame results as conventional neural networks. Another\nconfiguration of symmetry was suggested in [3,4] to improve\nthe speed up ratio. In this paper, our previous algorithm for\nfast neural networks is developed. The frequency domain\ncross correlation is modified in order to compensate for the\nsymmetric condition which is required by the input image.\nTwo new ideas are introduced to modify the cross correlation\nalgorithm. Both methods accelerate the speed of the fast\nneural networks as there is no need for converting the input\nimage into symmetric one as previous. Theoretical and\npractical results show that both approaches provide faster\nspeed up ratio than the previous algorithm.",
"references": null,
"publisher": "World Academy of Science, Engineering and Technology",
"index": "International Science Index 11, 2007"
}