Neural and genetic approximations of fractal error

Fractal error is an image processing metric that can be used to locate man-made features in aerial images. The main disadvantage of the fractal error algorithm is that it can take several seconds to compute on large images. Therefore, it is desirable to create an approximation of fractal error to provide real-time analysis. This paper presents two novel approximations of fractal error using a genetic algorithm and a neural network. The results obtained using the approximations are compared with those obtained from the fractal error algorithm. Results from the neural network and the genetic algorithm are compared with one another. The neural network provides and accurate representation of fractal error, while the genetic algorithm does in fact preserve all of the desired features of the original fractal error image. The genetic algorithm has been shown to be computationally faster than the neural network.

[1]  Michael C. Stein.  Fractal Image Models And Object Detection , 1987, Other Conferences.

[2]  Alex Pentland,et al.  Fractal-Based Description of Natural Scenes , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  Surya B. Yadav,et al.  The Development and Evaluation of an Improved Genetic Algorithm Based on Migration and Artificial Selection , 1994, IEEE Trans. Syst. Man Cybern. Syst..

[4]  Darrel L. Chenoweth,et al.  Fractal error for detecting man-made features in aerial images , 1994 .

[5]  Carey E. Priebe,et al.  An initial assessment of discriminant surface complexity for power law features , 1992, Simul..

[6]  Kenneth J. Hintz,et al.  Fractional Brownian motion models for synthetic aperture radar imagery scene segmentation , 1993, Proc. IEEE.

[7]  Carey E. Priebe,et al.  A PDP Approach to Localized Fractal Dimension Computation with Segmentation Boundaries , 1995, Simul..

[8]  Kurt Hornik,et al.  Multilayer feedforward networks are universal approximators , 1989, Neural Networks.

[9]  Bernard F. Buxton,et al.  Evolving edge detectors with genetic programming , 1996 .

[10]  Kenneth J. Hintz,et al.  Fractal image compression and texture analysis , 1991 .

[11]  Joseph Naor,et al.  Multiple Resolution Texture Analysis and Classification , 1984, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[12]  J. Knecht,et al.  Feature detection in synthetic aperture radar images using fractal error , 1997, 1997 IEEE Aerospace Conference.

[13]  F. Girosi,et al.  Networks for approximation and learning , 1990, Proc. IEEE.

[14]  Kenneth J. Hintz,et al.  Dimension and lacunarity measurement of IR images using Hilbert scanning , 1991, Defense, Security, and Sensing.

[15]  John R. Koza,et al.  Evolution of a 60 Decibel Op Amp Using Genetic Programming , 1996, ICES.

[16]  Ken-ichi Funahashi,et al.  On the approximate realization of continuous mappings by neural networks , 1989, Neural Networks.

[17]  James M. Keller,et al.  Texture description and segmentation through fractal geometry , 1989, Comput. Vis. Graph. Image Process..