Abstract
With the widely applications of digital images, the lossy compression will gradually become the mainstream of compression methods. Because this method might cause the lost of important information, only compression with low ratio are considered on medical images. The goal that we are pursuing is to maximize compression ratio while keeping image quality unchanged. Therefore image quality assessment plays an important role for it. ROC (Receiver Operating Characteristic) is a well-known method for evaluating wavelet compression and JPEG (Joint Photographic Experts Group) compression. Human eyes are difficult to differentiate the nuances of compressed images at low compression ratio. Our research focuses on evaluating image quality by using statistic methods. We develop some image quality indexes using Moran’s I test, kurtosis, etc to evaluate the image quality of wavelet and JPEG compression at low compression ratio (about < 25 times). The results indicate that JPEG compression is better than wavelet compression when the ratio is below 10 times and the reverse is true at compression ratio from 10 to 25 times.