Earth surface image analysis sensitivity improvement by means of combination of fuzzy and neuralnet segmentation methods
User Rating: 
/ 0
- Details
-
Category: IT technologies
-
Last Updated on Wednesday, 25 December 2013 11:33
-
Published on Wednesday, 19 December 2012 16:04
-
Hits: 4172
Authors:
I.M. Udovik, State Higher Educational Institution «National Mining University», Assistant Lecturer of the Department of Computer Systems Software, Dnipropetrovsk, Ukraine
Abstract:
Purpose. To increase sensitivity of visual analysis of low-contrast images of Earth surface and geophysical fields.
Methodology. We have increased the dimension of the space characteristics of the analyzed information in the image by using the method of fuzzy segmentation of initial data, and have formed new multi-dimensional array of images based on the obtained membership functions of classes with subsequent segmentation of the Kohonen neural network.
Findings. We have developed and experimentally proved the new method of segmentation of low-contrast images of the Earth surface and geophysical fields.
Originality. The new method of formation of multi-dimensional information space using images membership classes in the fuzzy segmentation algorithm followed by the formation of an adaptive single output image.
Practical value. The informativity of the method has been approved by experimental testing on segmentation of real examples of low-contrast images. The sensitivity of the procedure of segmentation, depending on the nature of the task may vary.
References:
1. Форсайт Д. Компьютерное зрение: современный подход / Форсайт Д., Понс Ж. – М.: Вильямс, 2004.
Forsyth, D. and Ponce, J. (2004), Computer Vision: A Modern Approach, Williams.
2. Jain, A.K., Dubes, R.C. (1988), Algorithms for Clustering Data, Engelwood Cliffs: Prentice-Hall.
3. Zheru Chi, Hong Yan, Tuan Pham. (1996), Fuzzy Algorithms: With Application to Image Processing and Pattern Recognition, World Scientific, London.
4. Осовский С. Нейронные сети для обработки информации / С. Осовский – М.: Финансы и Статистика, 2002.
Osovskiy S. (2002), Neyronnye seti dlya obrabotki informatsii [Neural networks for information processing], Finansy i Statistika, Moscow, Russia.
5. Ахметшина Л. Г. Сегментация мультиспектральных изображений на основе метода нечёткой кластеризации / Ахметшина Л. Г. // Сб. научных трудов НГАУ– Днепропетровск, 2000. – Т.1. – №9. – С. 90–93.
Akhmetshina, L.G. (2000), “Segmentation of multispectral images based on fuzzy clustering method”, Collected scientific papers NGAU, Vol.1, no.9, pp. 90–93.
6. Ахметшина Л. Г. Сегментация мультиспектральных изображений с использованием самоорганизующихся карт Кохонена / Ахметшина Л. Г., Егоров А.А. // Сб. научных трудов НГАУ – Днепропетровск, 2002. – Т.2. – №14. – С. 154–158.
Akhmetshina, L.G. and Egorov, A.A. (2002), “Segmentation of multispectral images using Kohonen self-organizing maps”, Collected scientific papers NGAU, Vol.2, no.14. pp. 154–158.
7. Ахметшина Л.Г. Информационные возможности модуляционного преобразования при сегментации мультиспектральных изображений. / Ахметшина Л.Г. // Системні технологіі. – Дніпропетровськ, 2004. – №6. – C. 122–127.
Akhmetshina, L.G. (2004), “Information capabilities of modulation conversion in segmentation of multispectral images”, Systemnі tehnologіі, no. 6, pp. 122–127.
2012_6_udovik
2013-12-24 
522.75 KB 
1557
Archive