Research On Image Filtering Algorithm For Mixed Noise And Its Application In CT Image

In real life, we can get varieties of digital images, but influenced by kinds offactors, these images are corrupted by mixed noise more or less. The noise will not onlyaffect our vision, which will lead to the deviation to the understanding of the image, butalso has great impact on the later image processing, which will change the results of theimage processing. All of these will have a great influence on our life and work.Firstly, the paper reviews the classification of image noise, research status andresearch methods of the image denoising.Secondly, the paper discusses the processing of the impulse noise in the image.According to previous researches, the median filter can help to remove the impulsivenoise effectively. So the author does a lot of study on the traditional median noise aswell as switching median and extreme valued median, and finally the impulse noise isremoved by using an adaptive median filtering based on the pulse detection. Thisalgorithm can distinguish the pixels and noise point through two levels of judging. Thedetection of the removal after the noise points removed and the experimental simulationshows that the algorithm is really superior to the traditional median algorithm.Thirdly, the paper discusses how to remove the Gauss noise. After a lot of reading,the author found that wavelet denoising is an effective way to remove the Gauss noise.This paper does a certain amount of research on the wavelet image denoising,introduces the related knowledge about wavelet, and then studies the principle andprocess of the wavelet denoising, especially on the wavelet threshold denoisingalgorithm. This paper mainly focuses on the adaptive threshold selection. Simulationresults show that adaptive threshold method is better than the wavelet threshold.Finally, the author proposes that we can use the two algorithm together to removethe mixed noise (here refers to the noise mixed with impulse noise and Gauss noise).Specifically, we can remove the impulse noise with the median filter based on pulsedetection in the airspace, and then use the wavelet adaptive threshold filtering toremove the rest of the noise. What’s more, that this special method can be used toremove the mixed noise in the CT images, and experiments show that the effect issatisfactory.