Landsat image compression using lifting scheme

Abstract

In this paper, Karhunen Loeve transform is applied to the Landsat image for removing the spectral correlation, and then wavelet transform is applied for removing spatial correlation followed by which coding is done. Biorthogonal wavelets are applied in one method and lifting scheme using Haar wavelet is applied in another method. SPIHT and EZW are applied for coding in each of the two methods and the results are compared. The correlation coefficient is calculated to verify spectral decorrelation and the compression is measured in terms of bit rate, compression ratio and peak signal to noise ratio is calculated. The experiment was tested with the Landsat images and found that the correlation coefficient has been reduced and hence KLT is effective for spectral decorrelation. In terms of compression ratio, the compression approach using KLT, Lifting scheme, EZW provides better results than other compression approaches. In terms of bit rate, the compression approach using KLT, Biorthogonal wavelet, SPIHT provides better results than other compression approaches.