Research On Surface Feature Extraction Of Functional Structures Based On Shearlet Transform

With the development and progress of precision manufacturing industry,the requirements for the performance and lifetime of precision components are also continuously increasing.The defects and damages of individual components can affect the performance and lifetime of the entire system.However,in actual production,many defects of precision components are identified by the human eye,which is not only inefficient but also prone to errors.In order to provide the entire system with better performance and longer service life,it is particularly important to quickly identify and automatically detect the features of functional structural surfaces.This paper proposes a method for segmentation and feature extraction of functional structure surface regions based on shear waves,which is verified by experiments and compared with traditional methods.Experiments show that shear waves have considerable advantages in feature extraction.Firstly,this paper introduces the research background and significance of functional structure surface feature extraction,and introduces the current situation of functional structure surface feature extraction methods at home and abroad.Next,the article gives a detailed introduction to the development process from wavelet to shearlet,and analyzes the limitations and shortcomings of wavelet,contourlet,and curvelet.Then,it briefly introduces shearlet,a novel multiscale geometric analysis tool,including its properties and advantages: good directional sensitivity,reconstruction accuracy,and sparsity.It also introduces the implementation and principle of nonsubsampled shearlet transform without nonsubsampled.In order to quickly extract the features of functional structural surfaces,based on nonsubsampled shearlet transform,this paper utilizes the directional sensitivity advantage of nonsubsampled shearlet transform,and combines mathematical morphology to successfully segment functional structural surfaces.Experiments have shown that feature edges can be extracted for surfaces with obvious boundaries,and feature edges can also be extracted for smooth edges that are difficult to separate,to achieve functional surface region segmentation,Compared to the traditional Canny operator,the method in this paper has better results for smoothing edges.Using the multiscale decomposition characteristics of nonsubsampled shearlet transform,combined with L0 gradient minimization and relative total variation,the texture features of the optical element surface were successfully removed and the defects on the element surface were isolated.Experiments have shown that the third dimension is the key dimension for separating surface defects when performing multiscale decomposition surface feature analysis.The separation method proposed in this paper has the characteristics of good directional recognition,accurate separation and positioning,and can completely extract defect features.This has important reference application value in precision optical element detection and surface metrology.Compared to traditional methods,texture feature extraction based on shearlet transform has better representation capabilities.