Research On Basic Problems Of Image Processing Based On Structural Feature Description

Nowadays,the human living environment is the stage of rapid development of information,where the image processing technology in digital media meets the needs of human expression and appreciation of things of texture information,and has played a crucial role in people's daily life.Image processing technology has been applied in various fields,and its basic researching problems include image denoising,edge detection,image enhancement,image segmentation,image recognition,etc.,where image denoising and edge detection are the bases of many other image processing tasks.Therefore,this paper presents some image denoising and edge detection methods based on structural feature description of images to improve the performance of existing image denoising and edge detection algorithms,which can provide the basis for the perfection of the theory of digital image processing technology.This paper includes the following contents:(1)It introduces the background and application fields of image processing briefly,elaborates the related works of image denoising and edge detection,and gives the content introduction of each chapter in this paper.(2)It introduces some basic knowledge of image processing,including the basic concepts of digital images,such as image and resolution,image neighborhood and image edges.It also describes the basic knowledge of image processing based on Matlab,such as the types of images in Matlab,the functions in Matlab about reading images,displaying images and saving images.(3)It presents a non-local mean deoising method based on the similarity measure(SSIM)of images to improve similar defects of weights is not accurate enough in the classical non-local mean filtering algorithm.The similarity in this method between pixels was carried out by using the structural similarity measure(SSIM),and the similarity between pixels was obtained by brightness,contrast and structure information.The experimental results show that the structural similarity measure based on the non-local mean filter for image denoising can effectively remove noises under the condition of the algorithm,the detail image information maintained is in a good condition.The PSNR quality evaluation standard for the denoising results are used to validate the effectiveness of our algorithm.(4)It proposes a LOG edge detection method based on Zernike moment to improve the problem in traditional LOG edge detection operator which weakens the image edges during the smoothing processing.It firstly calculates the weights of LOG operator based on Zernike moment in the Gauss filter.And then the edges are detected by Laplacian operator and zerocrossing method.Based on the image edge feature tests,our algorithm using Zernike moment weight calculation method based on Gauss filter can keep features better than the classical Lo G operater during the smoothing processing.(5)It summarizes the research contents of this paper,and discusses the future directions including how to further improve the proposed non-local mean filtering method and LOG edge detection method,and the application of the idea about the structural feature describtion to other image processing problems.