Application Of Information Fusion Technology In Image Filtering

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 filter 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.According to the classification of noise, the dissertation firstly discusses Guassian noise filters, impulse noise filters and hybrid noise filters, which include modified trimmed mean filter (MTM), center-weighted modified trimmed mean filter (CWMTM), classified average and weighted mean filter (CAWM), fuzzy weighted average filter (FWA), and adaptive median and weighted mean filter (AMAWM). The simulation results show that these hybrid filters can effectively reduce hybrid noise. The adaptive median and weighted mean filter (AMAWM) has the best performance among the five filters.Based on research and academic analysis of information fusion technology, the dissertation brings forward the idea of compositive entropy and its algorithm in order to avoid the defect caused by the judgement of the edge points with a single criterion. The new compositive entropy algorithm is implemented in a linear and non-linear hybrid filer, and two simulation results are shown to verify its performance. One is about the comparison between the hybrid filter based on compositive entropy and single-criterion hybrid filters, including...