Medical image compression using block-based PCA algorithm

PCA algorithm can be employed to aid in image compression. In this paper, two extended-PCA algorithms that manipulate the block information of the image are tested and compared. The first algorithm is termed as block-by-block PCA in which general PCA algorithm are applied on each block of the image. In the second algorithm- the block-to-row PCA, all block information are first concatenated into row before general PCA algorithm is then applied on the transformed matrix. Digital fundus image is used as the input image in this work. Using the newly-derived compression ratios, the result shows that block-to-row PCA outperforms block-by-block PCA in terms of image quality and compression rate. At equal block size and compression ratio, block-to-row PCA can achieve higher PSNR than block-to-block PCA. Blocking effects are discernible on the reconstructed image using block-to-block PCA with block size = 16 and compression ratio as low as 0.25 while no apparent distortion are seen on the reconstructed image using block-to-row PCA with block size = 32 and compression ratio as high as 0.90.