In this paper, the results of combining support vector machine (SVM) learning with discrete wavelet transform and contourlet transform in image compression are compared. The algorithm combines SVM learning with the discrete wavelet transform (DWT) and contourlet transform (CCT) of the image. An SVM selects the minimum number of training points, called support vectors which ensure modeling of the data within the specified level of accuracy. It is this property that is exploited as the basis for an image compression algorithm. Now, the SVMs learning algorithm performs the compression in a spectral domain of DWT and CCT coefficients.


 Peak signal to noise ratio (PSNR) is computed for the images compressed using wavelet and contourlet transforms. Results show that contourlet transform based image compression is more effective in capturing smooth contours (due to anisotropy property) than wavelet transform based compression.