{"title":"Maximizer of the Posterior Marginal Estimate for Noise Reduction of JPEG-compressed Image","authors":"Yohei Saika, Yuji Haraguchi","country":null,"institution":"","volume":63,"journal":"International Journal of Computer and Information Engineering","pagesStart":296,"pagesEnd":301,"ISSN":"1307-6892","URL":"https:\/\/publications.waset.org\/pdf\/11109","abstract":"We constructed a method of noise reduction for\nJPEG-compressed image based on Bayesian inference using the\nmaximizer of the posterior marginal (MPM) estimate. In this method,\nwe tried the MPM estimate using two kinds of likelihood, both of\nwhich enhance grayscale images converted into the JPEG-compressed\nimage through the lossy JPEG image compression. One is the\ndeterministic model of the likelihood and the other is the probabilistic\none expressed by the Gaussian distribution. Then, using the Monte\nCarlo simulation for grayscale images, such as the 256-grayscale\nstandard image \u201cLena\" with 256 \u00d7 256 pixels, we examined the\nperformance of the MPM estimate based on the performance measure\nusing the mean square error. We clarified that the MPM estimate via\nthe Gaussian probabilistic model of the likelihood is effective for\nreducing noises, such as the blocking artifacts and the mosquito noise,\nif we set parameters appropriately. On the other hand, we found that\nthe MPM estimate via the deterministic model of the likelihood is not\neffective for noise reduction due to the low acceptance ratio of the\nMetropolis algorithm.","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"International Science Index 63, 2012"}