Authors : S.T. Lim, D.F.W. Yap and N.A. Manap
Abstract: Principal Component Analysis (PCA) has received growing attention in its latent potential in image compression. However, the image reconstructed from PCA compressed data can be improved in terms of image quality and compression ratio. In this study, a modified PCA algorithm was considered. In this algorithm, the eigenvectors derived from the original image was used to reconstruct the compressed data. Performance evaluation show that PSNR and SSIM obtained for image compressed by the proposed modified PCA are significantly higher than the conventional PCA algorithm (p<0.05). The objective evaluation results were further confirmed by the visual inspection of the output images where less streaks and noise were found on image compressed by the proposed modified PCA at compression ratio as high as 90%.
S.T. Lim, D.F.W. Yap and N.A. Manap, 2018. Image Compression Using a Modified Principal Component Analysis Method. Journal of Engineering and Applied Sciences, 13: 3104-3109.