Multiple-Scale Wavelet Decomposition, 3D Surface Feature Exaction and Applications

Abstract

This paper presents a method of applying wavelets to decompose three-dimensional surface into multiple-scale subsurfaces, and of using the subsurface features to predict surface functions and detect machining errors. The one-dimensional discrete wavelet decomposition is first introduced, and then, it is extended to decompose and analyze three-dimensional surfaces. In this study, applications of wavelets decomposition are demonstrated in several automotive case studies, including abrupt tool breakage detection, chatter detection, cylinder head sealing/mating surface leak path detection, and transmission clutch piston suiface none lean up region detection. These case studies successfully demonstrate that the proposed multiple-scale two-channel wavelet decomposition method can be served as a useful tool for surface functions prediction and machining errors detection.

A shearlet-based separation method of 3D engineering surface using high definition metrology

“…This high definition metrology (HDM) provides a good platform to develop new surface analysis methods that involving surface texture separation and surface feature characterization. In the last few years, some efforts have been made to use it to analyze engineering surfaces [16,26]. A large number of high-resolution data of engineering surfaces can be obtained by this device, which is served as the input of filtering.…”

“…The discrete wavelet transform (DWT) is a widely used wavelet method in digital signal and image analysis, and it can be easily extended from one dimension to the condition of two dimensions. The application of 2D DWT in analyzing 3D engineering surfaces can be found in [16,17]. Zeng et al [18] adopted two-dimensional dual-tree complex wavelet transform (2D DT-CWT) to separate engineering surfaces, which is more superior to 2D DWT in the aspects of shift-invariance and directional selectivity.…”

A Selective Multiclass Support Vector Machine Ensemble Classifier for Engineering Surface Classification Using High Definition Metrology

“…But with the development of 3D measuring techniques especially faster optimal methods, a lot of data can be obtained from engineering surfaces in quite a short time [5,6], it is inappro priate to simply use one-dimensional (ID) parameters to character ize surface textures. Consequently, more surface characterization approaches have been proposed to obtain abundant information about 3D surfaces, such as a 3D parameter set [4,[7][8][9], gray level co-occurrence matrix [10], two-dimensional (2D) autocorrelation function and spectral analysis [11], and a 3D Monte Carlo model [12], In recent years, some approaches that first used in signal processing areas have been adopted to extract features of engi neering surfaces, such as Gabor filter banks [13], Gaussian filter banks [14], and wavelet packets [15][16][17], and after filtering, some numerical surface parameters are calculated for each sub-band to represent a given surface. The goal of feature extraction is to improve the effectiveness and efficiency of classification.…”

“…In this study, attention is mainly focused on wavelet filters, due to the fact that it can provide multiscale/orientation analysis, which makes it a powerful tool in fea ture extraction and is superior to traditional filters [16,18]. There are many kinds of wavelets, and discrete wavelet transform (DWT) is the most widely used one in the analysis of workpiece surfaces in previous studies [16,19,20]. However, DWT mainly has two disadvantages: lack of shift-invariance and poor direc tional selectivity for diagonal features, which impair its applica tion in engineering surface analysis.…”

Characterization of Cutting Force Induced Surface Shape Variation in Face Milling Using High-Definition Metrology1

“…Among them, Zhang et al [25] proposed a histogram estimator of surface parameters to classify surface pattems using the LHI teclinology, and Liao et al [26] used extracted surface features for machining error detection. We notice that the cutting force variation induced surface pattems such as those reflected in Fig.…”