The Modified Algorithms Of Median Filtering Applicatied In Image Processing

Image signal is often interfered by many kinds of noise when produced, transmitted and recorded. The quality of image must be improved by filtering technology before edge detection, image segmentation, feature extraction, pattern recognition and so on. Image filtering is expected to both reduce the noise and keep the image details. But it is difficult to design such a good filter for the overlapping of noise and image details. Image filtering is a very important and hot research field in image processing.In the pretreatment, image filtering is the most important. Linear and nonlinear filters are in common use. Linear filter can filter gauss noise, but pulse noise and high frequency can not be filter. On the contrary, Nonlinear filter can filter pulse noise, but can not filter Even distribution noise and gauss noise. Also hybrid filtering is one of the popular image filtering, and has some advantages.This thesis firstly studies the image filtering, analyses Mean filtering in linear filtering and median filtering in nonlinear filtering, On the base, it analyses hybrid filtering. For resolving the contradiction of removing noise and holding image details in median filtering, this thesis gives two modified algorithms in different ways. One is extremum median filtering based on threshold, the other is multilevel median filtering based on directions. And then the results of experiment prove that these modified algorithms are valid. They both can remove noise and meanwhile retain image details. At the end of thesis there are some more ideas for perfecting modified algorithms of the median filtering.