| In medical image processing fields, medical image will be always affected by different kinds of noise pollution in the forming and transmission record processing due to the imaging system, transmission medium and records of the equipment is not perfect. In general, there are not related between the noise signal and the studying object, the observable information of medical images will be disturbed by the noise signal with useless information. It is the primary work to remove the noise in image processing,otherwise it is meaningless to directly use the noise image without image denoising, and that can even lead the doctor to make mistakes of judgment in the diagnosis of the patient’s condition. In order to reduce the noise on the influence of the image quality, Main research completed is as follows:Studying on the theory of knowledge about the relevant image denoising aspects. This thesis studys the current popular three kinds of testing noise mechanism:switching threshold value method, extreme value method and the poles threshold method, and how to evaluate the effect of noise testing, with the misjudgment rate, the leakage rate, the accuracy rate and the error rate to measure, and use the current evaluation denoising algorithm of two kinds of quality criteria:subjective evaluation standards and objective evaluation standards, putting forward to judge the mixed noise filtering algorithm.Studying on the mainstream of the technical knowledge about the current related image filtering technology research and the improvement methods. This thesis studys several regular image denoising methods:average filtering, median filtering, gaussian smoothing filtering and their improvement algorithms, then briefly analyse the performance of the quality and practical application about these algorithms, test and verify the standard mean filtering algorithm, the standard median filtering algorithm and the gaussian smoothing filtering algorithm through simulation experiment.To address the problems of denoising mixed noise in medical image, through comprehensive consideration the respective advantages between the median filtering algorithm and the mean filtering algorithm, this thesis puts forward a mixed denoising algorithm, first, to remove salt&pepper noise that exist in mixed noise with the assistance of the improved median filtering algorithm——adaptive threshold median filtering algorithm, then to remove gaussian white noise that exist in mixed noise with the assistance of improved average filtering algorithm——local extreme average filtering algorithm, in order to reach the purpose of removing the mixed noise that exist in the medical images.Finally, test mixed noise filtering algorithm on the matlab software platform, verify that this algorithm is more effective than the standard median value filtering algorithm or the standard mean filtering algorithm. Images show more clearly and the details of the image information will be kept more complete. |
