Image Denoising Based On Mathematical Morphology

The digital image is an important way for our human to access and deliver effectiveinformation. Digital image processing originated in the twenties. In recent decades, with thegrowing popularity of the digital image, digital image processing is becoming one of theresearch focuses in the computer field. And there are many branches in digital imageprocessing , of which image denoising is an important one.As for image denoising, there have been many classics algorithms such as: mean filtering,median filtering, Wiener filtering. These algorithms can remove noise, but at the same time theywill bring a lot of loss of image detail,which can not reach people’s expectations.Mathematical morphology theory was first put forward by G. Matheron and his student J.Serra in 1964 when they engaged in the iron ore nuclear quantitative rock analysis at ParisSchool of Mines, France. After decades of development, the theory has matured. The theorybases on set algebra, so in essence it is a mathematical method. This approach uses the idea ofset theory, and finally find out the structure and geometry of the image through the interactionof structural elements and images. Mathematical morphology has perfect theory, high efficiency,simple usability and is suitable to deal with many shape-related problems in specializedhardware. Using mathematical morphology to remove noise can base on the priori geometricinformation and use morphological operators to filter out the noise effectively as well as retainthe details. This article mainly focuses on mathematical morphology theory to designalgorithms which can remove gaussian noise.The work done in this paper includes the following: the research background, basicconcepts ,significance and development trends of image denoising, subiective and objectiveevaluation methods of image quality, MSE, SNR, PSNR, the classification of image noise,several noise models like impulse noise and gaussian noise, the application and weak points ofclassical denoising algorithms such as mean filter, median filter and wiener filter, thedevelopment, basic concepts and principles of mathematical morphology, four basic operationof binary and grayscale morphology: dilation, erosion, open and close, and their applications in image denoising. The paper proposes a new mathematical morphology algorithm which iscalled switching morphology filter to remove the gaussian noise. It first set a threshould T, thenuse T and morphology opening and closing to detect gaussian noise, and use multi-structureopening and closing combination operators to remove the noise. In this paper, we use thisalgorithm to deal with gaussian noise of different variance in Lena image and document image.MATLAB simulation results show that compared to the traditional median filter and wienerfilter, this new algorithm can get better PSNR and has better image denoising and detailpreservation performance. Finally, we summarize this paper and prospect the future researchdirection of the switching morphology filter that the paper have proposed.