Study On Independent Component Analysis Method And Its Applications To Image Denoising

Independent Component Analysis (ICA) is a recently developed a very effective blind sourceseparation, has excellent blind identification, capacity and feature extraction, widespread concern inmany practical applications. ICA not only has important theoretical but also high value, especiallyin the field of biomedical signal processing, telecommunications, speech signal processing, imageprocessing has a broad application prospects. Many literatures on ICA were published and lots ofalgorithms were proposed during the past ten years in a large number of journals and conferenceproceedings. Currently, ICA has become a hot research topic.In this thesis, the principle and algorithms of ICA are researched in detail. Theoretical analysisof the lack of the classic ICA algorithm, proposed an improved ICA algorithm. And do someresearches in image denoising which is based on ICA. Contribution and innovation of this paperinclude the following aspects:(1) Arranged the basic principles of the ICA. Meanwhile derived from the perspective ofinformation theory the metric measure ICA independence standards, and summed up the generalprocess of ICA solution, also compared the performance of the algorithm.(2) ICA classical algorithm local optimum problem that is theoretically proved, as to this issue, thispaper presents an algorithm based on chaotic particle swarm optimization ICA. In the end,simulation results show that the algorithm is better than the others.(3) ICA transform image denoising based denoising using soft threshold shrinkage function, andtend to make the edges of the image is blurred and characteristic loss. Improvements on the systolicfunction can effectively avoid the fuzzy image edge and feature loss phenomenon.