Several Digital Image Filter Algorithms

It is necessary for digital image to be preprocessed before being used for analysis due to limitation of imaging methodology, since the edge and minutia in noise may lead to difficulty in the following image process, such as edge detection, image segmentation and image registration. So preprocessing is one of the essential topics in image processing, which can greatly improve image quality. Image denoising is one of the most popular technologies in dealing with image processing. It is applied to increase SNR and to stress on expectation characters. It mainly uses the features that noise and signal distribution differently in the frequency domain. Signal is in low frequency, while noise in high frequency and minutia in image is also in this area. When performing denoising, traditional time or frequency domain denoising algorithms always destruct the edge and minutia of image. So it has been focus on how to keep line-like structures such as edges and minutia well while denoising for recent years.In this paper, the total research direction is divided into two:The first is spatial filtering. We propose an improved adaptive minutia preserving smoothing algorithm based on multiscale and multidirectional masks. This algorithm keeps the mask's good performance in preserving details. It divides image into sub-images and detects them with adaptive threshold value according to the gradient information of the whole and the local image. This method deals with the difficulty of choosing threshold and improves the automatization of image smoothing.Secondly, we studies on anisotropic diffusion filtering which based on partial differential equation (PDE) . The traditional Perona filtering often brings about blurred image. So we add linear masks to the minutia preserving smoothing algorithm which based on solving a nonlinear diffusion equation.This paper values theory and practice. To every algorithm, there are both theory analysis and relative programs, which are designed by Matlab to verify the properties of these algorithms. The results of experiments show that these two improved algorithms have more better performance than unimproved either in reduce noise efficiently in infra-region and at edges or in keep line-like structures such as edges and textures.