You can use adaboost along with SVM to improve the performance.

First state please the software you use for classification. According my my experience there are some tuning parameters which are of importance and in different products have different recommended values.

I suppose that the following papers can be helpful:

TARABALKA, Y., FAUVEL, M., CHANUSSOT, J. & BENEDIKTSSON, J. A. 2010. SVM-and MRF-based method for accurate classification of hyperspectral images. Geoscience and Remote Sensing Letters, IEEE, 7, 736-740.

AGHIGHI, H., TRINDER, J., TARABALKA, Y. & LIM, S. 2014. Dynamic Block-Based Parameter Estimation for MRF Classification of High-Resolution Images. Geoscience and Remote Sensing Letters, IEEE, 11, 1687-1691.

Have you tried ecognition?

There is a nice chapter for comparing the classification techniques accuracy in fallowing link:

http://www.intechopen.com/books/landscape-planning/land-use-cover-classification-techniques-using-optical-remotely-sensed-data-in-landscape-plannin

@Onur - thank you for the source.

I suggest you to use feature selection procedure in combination with SVM, it signficantly improves the accuracy by eliminating ininformative bands.

For example, genetic algorithm based SVM (GA-SVM) robust over regular SVM

@Rajasheker Reddy Pullanagari, Thank you for your answer and I'm interested in the process of feature selection based on GA.

@Shiv Prakash, @Onur Satir , @Hossein Aghighi, @Hristo Nikolov, @Shayak Sadhu And thankful to everyone for your answer and source. It's useful for my research.

@Qichun Yang I'm not yet try to use ecognition.