Removal of High Density Impulse Noise Using Cloud Model Filter

The fact that makes image denoising a difficult task is uncertainties in the impulse noise. The most knowledge in dayflies is uncertainty and erratic, unfortunately it is similar to impulse noise. The mathematic implements for handling uncertainty mostly are probability theory and fuzzy mathematics. That means, among the uncertainties involved in impulse noise, the randomness and the fuzziness are the two most important features. In this paper we use a detail-preserving filter based on the Cloud Model (CM) to remove severe impulse noise. CM is an uncertain conversion model, between qualitative and quantitative description that integrates the concept of randomness and fuzziness. The normal random number generation method in normal cloud generator algorithm overcomes the insufficiency of common method to generate random numbers. It can produce random numbers which can be predictable and replicated, and this random numbers present to be a random sequence as a whole. The digital features of the normal cloud characterized by three values with the expectation Ex, entropy En and Hyper entropy He and are good enough to represent a normal cloud. First, an uncertainty-based detector, normal cloud generator, identifies the pixels corrupted by impulse noise. Then, the identified noise pixels are replaced by a fuzzy mean estimation of the processed noise free pixels within the detection window. Compared with the traditional switching filters, the CM filter makes a great improvement in image denoising. Especially, at high density noise level. Thus, the cloud model filter can remove severe impulse noise while preserving the image details. Key words—uncertainty, fuzzy median, normal cloud generator, cloud drops.