Research On Image Denoising And Edge Detection Based On Wavelet Transform

Digital image processing technology has a wide range of applications in production and life,and digital images play a decisive role in the ever-changing technological development.Noise affects the expression of image information.The edge is the reflection of the main structure of the image and the main outline.It is also the direct interpretation of image understanding and the basis for further segmentation and recognition.Therefore,suppressing noise and improving edge detection accuracy are an important content of image processing.Compared with the traditional wavelet denoising threshold function,the continuity is poor,and the shortcomings of insufficient adaptability.In this paper,through the analysis of the existing improved functions,relevant parameters and high-order variables are introduced to improve the existing threshold functions,and the denoising ability and fitness of the threshold function are improved.Traditional edge detection methods have problems such as poor continuity and insufficient noise suppression ability in the edge detection of noisy images.Based on wavelet theory,this paper proposes a parallel algorithm combining wavelet Mallat decomposition and reconstruction algorithm and Canny algorithm from multi-scale and multi-direction.It improves the continuity of edge detection and the edge retention rate of noisy images.Finally,it is demonstrated through experiments.The main work of the thesis includes:(1)Analyze and understand the traditional filtering denoising algorithm theory,use Matlab R2016 simulation software to compare the effects of denoising the image.The quantitative and qualitative analysis of the denoised image was introduced by using the evaluation index,and the advantages and disadvantages of the traditional filter denoising were obtained.(2)Analyze the denoising performance of the classical gradient edge detection operator using experimental demonstration,and compare the edge detection effect of pure and noisy images.The introduction of the edge detection evaluation index verifies the advantages and disadvantages of the traditional operator,and concludes that the classical operator has disadvantages and advantages in edge detection and noise suppression.(3)Based on the characteristics of the wavelet transform method,the commonly used methods based on wavelet transform denoising,wavelet modulus maxima denoising,spatial correlation denoising,and wavelet threshold denoising are introduced.In this paper,based on the mathematical analysis and experimental results of wavelet threshold denoising,aiming at the shortcomings of the continuity of the traditional threshold function and the insufficient range of adaptation,a corresponding method to improve the denoising threshold function is proposed.1 The semi-soft threshold and the compromise threshold function are used.The combination of the structure to add a parameter threshold function,2 through the construction of higher-order functions to improve the existing improved threshold function,the use of mathematical analysis and experimental demonstration of the effectiveness of the improved threshold function.(4)Based on the theoretical basis of wavelet transform,this paper analyzes the method of multi-scale modular maxima noisy image edge detection.Combining the Mallat decomposition reconstruction algorithm,and the detection of image edge continuity and integrity evaluation index,using Mallat and Canny algorithm combined to propose a better method of noise suppression and edge detection integrity,followed by a description of the direction of the adjustable wavelet transform The application of image edge detection,and demonstrated the feasibility of the method through experiments.