|
Independent component analysis ( Independent Component Analysis, ICA ) is a very effective blind signal processing technology developed in recent years , has important theoretical and practical value , with a wide range in the field of wireless communications, speech processing, image processing and biomedical attractive prospect , are playing an increasingly important role . In this thesis, in positron emission computed tomography ( Positron Emission Computed Tomo - graphy , PET ) image denoising introduce a new feature extraction method - independent component analysis , mainly in the following aspects : First, , the outlined medical imaging development history , imaging equipment - the working principle of PET imaging characteristics , and related medical image processing status quo ; followed on the basic principle of ICA and constraints , as well as the development history of the ICA and research and gave details of the implementation process of the ICA algorithm ; again , PCA algorithm to do pre-processing of experimental data on the PET images FastICA algorithm based on negative entropy , and then use the image data in the preprocessed for feature extraction , and then on the extraction of sparse coding shrinkage PET images of the feature vector encoding denoising , and finally reconstructed PET image ; Finally, in the PET image denoising , the result of the processing methods described herein and the median filter and the processing result of the wavelet filter Comparative Analysis the experimental results show that compared with several commonly used in medical image processing denoising algorithm, used in this paper de-noising algorithm is more suitable for PET image processing , and the effect is significant .
|