Mixel decomposition of remote sensing images is one of valid assistance means for improving quality of feature extraction from the images. Some problems exist in normal linear mixels decomposition, for example, registration error is too large, and neighbourhood information can be took full advantage of and so on. These result in that distortion of new images generated after linear mixels decomposition is more serious. Aiming at an above-mentioned circumstance, particle swarm intelligence searching method is put forward in this paper. New algorithm implements mixel decomposition of remote sensing images, combined with linear mixels decomposition model. The algorithm takes full advantage of neighbourhood information, makes the decomposition result more human, and presents better robustness to environment.
[1]
Shi Pei-jun.
Sub-pixel Model for Vegetation Fraction Estimation based on Land Cover Classification
,
2001
.
[2]
Agostinho C. Rosa,et al.
Self-Regulated Artificial Ant Colonies on Digital Image Habitats
,
2005,
ArXiv.
[3]
Vitorino Ramos,et al.
Artificial Ant Colonies in Digital Image Habitats - A Mass Behaviour Effect Study on Pattern Recognition
,
2004,
ArXiv.
[4]
Leonardo Bocchi,et al.
A New Evolutionary Algorithm for Image Segmentation
,
2005,
EvoWorkshops.
[5]
David Zhang,et al.
A universal texture segmentation and representation scheme based on ant colony optimization for iris image processing
,
2009,
Comput. Math. Appl..