Research On A Variable Order Fractional Integral Image Denoising Algorithm

With the development of modern science and technology,image processing technology has been widely used in aviation,information and communication,biotechnology,medicine and agriculture,machine vision and many other fields.As one of the important branches of image processing,image denoising has become a hotspot of many scholars.Its main purpose is to improve the visual effect of the image,highlight the edge and texture details of the image,and prepare for the subsequent image processing.In the image denoising process,the details of the preservation and denoising is a pair of irreconcilable contradictions.Many classical image denoising algorithms can remove the noise,but at the same time destroy the image edge and texture information.In this paper,a variable order fractional integral image denoising algorithm is proposed,which can remove the noise adaptively and balance the contradiction between the two.In this paper,the image denoising model is studied by the combination of theoretical analysis and simulation experiment.Many classic denoisingalgorithmsare easy to remove the edge information and other information of the image as noise,which results in the image blur.This paper,under the background of salt & pepper noise,mainly aims at improving this problem.So this paper firstlydetects the noise points by the method that detectssalt & pepper noise based on the number.Then use the discriminate function of “noise-edge”to do the secondly detect to the noise points and edge points which is detected in the first step.And the noise point is finally determined.Then,aiming at the problem that the order of the traditional fractional integral is given by a human or given by a large number of real numbers,it is proposes a method to determine the order of fractional integral by image local gradient modulus and image entropy adaptively.Finally,an adaptive fractional-order integration algorithm is used to filter out the detected noise points.The image denoising algorithm is simulated by MATLAB.And using the subjective and objective evaluation of image qualityanalyzes the experimental results.Compared with many classical denoising algorithms,this algorithm not only improves the signal-to-noise ratio of the denoised image and reduces the normalized mean square error,but also does not destroy the other texture information when removing the noise,proving the feasibility of the algorithm and effectiveness.