The Research And Application Of Digital Image De-noising Algorithm

In the era of informationized, digital and intelligent, the proportion of information in the image is increasing day by day. It has become a very important tool to obtain and transmit information. However, the digital image usually suffers from the interference of different noises during the acquisition and transmission process. As a result, the quality of image is reduced. In order to suppress the interference of the noise during image processing, image de-noising is a critical task.There are many kinds of image noise. Salt and pepper noise as well as Gaussian noise are common to see. Median filter is a kind of nonlinear filtering algorithm, which is widely used to de-noise the salt and pepper noise. Compared with the traditional linear filtering algorithm, it not only owns more efficient de-noising ability but also makes the image clearer and more vivid. Low pass filter has been the classic algorithm to reduce the Gaussian noise. It can keep the high frequency detail of image signal in the process of de-noise. However, disadvantages of these algorithms are more or less revealed as the application field is widening gradually. Therefore, a lot of improved median filtering and low pass filtering algorithm are appeared in the field of image de-noising in recent years. Meanwhile, some progress has been made.The main content of this paper is digital image de-noising problems. According to the characteristics of the salt and pepper noise and Gaussian noise in the image and the deficiency existing in image de-noising, the author compares and studies image de-noising algorithm thoroughly. The main work includes the following four aspects:First, knowledge about digital image de-noising processing and common filtering algorithm are introduced. Also, the characteristics, the classification and the evaluation standards of noise de-noising method are summarized. What’s more, the advantages and disadvantages of various filter algorithms are analyzed.Second, the improved median filtering algorithm is studied. According to the advantages and disadvantages of the two algorithms’ form in literature fourteen and sixteen, this paper proposes an adaptive median filter algorithm based on extremum. In order to prove that the algorithm is better than other median filtering algorithms, theoretical analysis and Matlab simulation experiment are carried out.Third, based on the study of traditional exponential filter principle, the author extract the root of transfer function which affects the filtering effect. Then, she proposes a Gaussian noise de-noising algorithm based on the index of filtering. And the effectiveness of this algorithm is verified through simulation experiment, the PSNR evaluation and histogram analysis.Forth, the research carries out the common image de-noising algorithm and hybrid noise filtering algorithm in experimental environment of Matlab by using some classical image such as portrait, tires and medical. It shows that image de-noising algorithms are all effective in processing these images. For the mixed noise image, the multi-step filtering algorithm is better than the single filtering algorithm.