LIBSVM: Low accuracy rate

Hi guys

I am currently using LIBSVM to train and test my data. My data are actually features of an image. I have performed scaling on the data (Using sample code I found on LIBSVMFAQ website.) but my classification accuracy is not as good as I expected.(shown below).

Accuracy = 48.5714% (102/210) (classification)

Accuracy = 54.2857% (114/210) (classification)

Accuracy = 56.1905% (118/210) (classification)

I have used 1040 images for training (50% are professional photos and 50% are amateur photos) and 210 images for testing (not labelled). I can provide more information when required.

Since I am still new to LIBSVM, could anyone please provide some help to me? I your help will be greatly appreciated.

Thanks

Marcus