A preliminary study of compression efficiency and noise robustness of orthogonal moments on medical X-ray images

Abstract

This paper provides a preliminary study of compression efficiency and noise robustness of two orthogonal moments (Legendre and Tchebichef) on medical images. Similar to a typical JPEG compression, the medical images are first subdivided into 8x8 blocks and transformed into moment domain. Moments up to order of 2 to 8 are used to reconstruct the image. The mean square error (MSE) between the reconstructed and the original images are computed as the measure of compression efficiency. For the noise robustness test, Gaussian white noise is applied to the original image before the same processing steps mentioned above are done. Discrete Cosine Transform (DCT), a standard JPEG compression algorithm, is used as the benchmark for this experimental study. The result shows that the compression ability of Tchebichef moments is comparable to DCT while Legendre moment has the highest noise robustness among them. © 2011 Springer-Verlag.