Evolutionary Strategies for Tuning Continuous Processing Parameters
Article Sidebar
Main Article Content
Abstract
Modern methods of data and space image processing - filtering, segmentation, classification, and restoration - require precise setup of a multitude of continuous parameters that determine the behavior of algorithms and the quality of results. Traditional gradient optimization methods demonstrate limited effectiveness in multi-mode, non-smooth, or noisy target functions. This work examines the application of evolutionary strategies (ES) - in particular, CMA-ES and Differential Evolution - for automatic adjustment of satellite and hyperspectral image processing parameters. A CMA-ES/DE hybrid scheme is proposed, combining the adaptation of the covariance matrix with directed difference mutations, ensuring stable and precise convergence in anisotropic parameter spaces. Experimental results on Landsat-9 data showed a decrease in the root mean square error by approximately 58% compared to the initial values, as well as an advantage over traditional methods (Otsu, K-Means, Watershed) by 6-12% in terms of segmentation accuracy.
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
References
Rechenberg I. Evolutionsstrategie: Optimierung technischer Systeme nach Prinzipien der biol-ogischen Evolution. Stuttgart: Frommann-Holzboog, 1973. 170 p.
Schwefel H.-P. Numerical Optimization of Computer Models. Chichester: Wiley, 1981. 389 p.
Hansen N., Ostermeier A. Completely Derandomized Self-Adaptation in Evolution Strategies // Evolutionary Computation. 2001. Vol. 9, No. 2. P. 159–195.
Storn R., Price K. Differential Evolution - A Simple and Efficient Heuristic for Global Optimi-zation over Continuous Spaces // Journal of Global Optimization. 1997. Vol. 11. P. 341–359.
Wierstra D., Förster A., Peters J., Schmidhuber J. Natural Evolution Strategies // Proc. IEEE Congress on Evolutionary Computation (CEC). Hong Kong: IEEE, 2008. P. 3381–3387.
Beyer H.-G., Schwefel H.-P. Evolution Strategies - A Comprehensive Introduction // Natural Computing. 2002. Vol. 1, No. 1. P. 3–52.
Das S., Suganthan P.N. Differential Evolution: A Survey of the State-of-the-Art // IEEE Trans-actions on Evolutionary Computation. 2011. Vol. 15, No. 1. P. 4–31.
Liang J., Qu B., Suganthan P., Hernández-Díaz A. Problem Definitions and Evaluation Criteria for the CEC 2013 Special Session on Real-Parameter Optimization. Technical Report 201212. Nanyang Technological University, 2013.
Hansen N. The CMA Evolution Strategy: A Tutorial // arXiv:1604.00772. 2016. 39 p.
Chen J., Li X. Hybrid Evolutionary Algorithm for Satellite Image Segmentation // Remote Sens-ing Letters. 2019. Vol. 10, No. 7. P. 612–621.
Zhao Y., Zhang L. Adaptive Parameter Optimization for Hyperspectral Image Classification Using CMA-ES // IEEE Transactions on Geoscience and Remote Sensing. 2021. Vol. 59, No. 4. P. 3120–3134.
Zhang Q., Deb K. Multiobjective Evolutionary Algorithms: Recent Advances and Applications // Evolutionary Computation. 2018. Vol. 26, No. 3. P. 321–354.
Sun Y., Wang G. Hybrid CMA-ES and Gradient Search for Image Restoration // Pattern Recog-nition. 2020. Vol. 102. Article 107245.
Liu F. Differential Evolution for Robust Hyperspectral Image Analysis // IEEE Journal of Se-lected Topics in Applied Earth Observations and Remote Sensing. 2022. Vol. 15. P. 4422–4435.
Hansen N. Active Covariance Matrix Adaptation for Robust Global Search // Evolutionary Computation. 2023. Vol. 31, No. 1. P. 1–28.