| Summary: | 碩士 === 國立東華大學 === 電機工程學系 === 93 === With the continued growth of multimedia and communications systems, the instrumentation and measurement fields have been a steady increase in the focus on image data. Developing tools and techniques to enhance the quality of image data plays a very relevant role.
In this thesis, we propose an efficient nonlinear algorithm to suppress impulse noise from highly corrupted image while preserving details and features. The algorithm is applicable to all impulse noise, including fixed valued impulses and random valued impulses, covering the whole image. The algorithm is based on fuzzy impulse detection and fuzzy noise cancellation techniques. The goal of the fuzzy impulse detector is to generate a noise identification binary matrix, the fuzzy noise cancellation module utilizes it to recommend a new value for every pixel. It is also used as the basis of the switch unit which controls each pixel of the output restored image in two different cases. From the result of simulation, the proposed method could remove noise effectively.
|
|---|
