Automatic segmentation of digitized data for reverse engineering applications

Reverse engineering is the process of developing a Computer Aided Design (CAD)model and a manufacturing database for an existing part. This process is used in CAD modeling of part prototypes, in designing molds, and in automated inspection of parts with complex surfaces. The work reported in this paper is on the automatic segmentation of 3-Dimensional (3-D) digitized data captured by a laser scanner or a Coordinate Measuring Machine (CMM) for reverse engineering applications. Automatic surface segmentation of digitized data is achieved using a combination of region and edge based approaches. It is assumed that the part surface contains planar as well as curved surfaces that are embedded in a base surface. The part surface should be visible to a single scanning probe (21/2D object). Neural network algorithms are developed for surface segmentation and edge detection. A back propagation network is used to segment part surfaces into surface primitives which are homogenous in their intrinsic differential geometric properties. The method is based on the computation of Gaussian and mean curvatures of the surface. They are obtained by locally approximating the object surface using quadratic polynomials. The Gaussian and mean curvatures are used as input to the neural network which outputs an initial region-based segmentation in the form of a curvature sign map. An edge based segmentation is also performed using the partial derivatives of depth values. Here, the output of the Laplacian operator and the unit surface normal are computed and used as input to a Self-Organized Mapping (SOM) network. This network is used to find the edge points on the digitized data. The combination of the region based and the edge based approaches, segment the data into primitive surface regions. The uniqueness of our approach is in automatic calculation of the threshold level for segmentation, and on the adaptability of the method to various noise levels in the digitized data. The developed algorithms and sample results are described in the paper     

A noise model for digitized data

The paper analyzes the overall conversion error of a noisy digitizer affected by Gaussian noise at its input or generated within the digitizer itself. It is shown that, under mild conditions concerning the ratio between the input noise standard deviation and the quantization step, the overall conversion error can be modeled by a Gaussian random variable uncorrelated with the input sequence. The power of the global conversion error is evaluated in closed formulae together with its degree of variability. Numerical simulations support the proposed analysis     

Registration and integration of multiple laser scanned data for reverse engineering of complex 3D models

In reverse engineering or rapid prototyping of a complex 3D object, there is often a need to scan a complete 3D model using laser digitizers. However, most digitizers scan objects in a 2.5D way such that the multiple data sets need to be aligned and integrated. In this paper a surface registration algorithm is proposed to solve the problem using a nonlinear minimization approach. Multiple data sets with different orientations can thus be aligned and integrated to construct a complete 3D data set. To control the accuracy of the registration process, uncertainty analysis of the registration parameters is investigated. A registration uncertainty model is developed to predict the uncertainty of the registration process. Using this model, we can predict the minimum number of the scanned data points to satisfy the required registration accuracy.     

Improved neural network model for reverse engineering

While conventional engineering transforms engineering concepts into real parts, in reverse engineering real parts are transformed into engineering models. The construction of a surface from three-dimensional (3D) measuring data points is an important problem in reverse engineering. This paper presents a reconstruction method for the sculptured surfaces from the 3D measuring data points. The surface reconstruction scheme is presented based on a neural network. The reconstruction of the existing surfaces is realized by training the network. A series of measuring points from existing sculptured surfaces is used as a training set. Once the neural network has been trained, it serves as a geometric model to generate all the points that are needed. However, the learning rate for the neural network is relatively slow, and the learning accuracy is often unacceptably low. In this paper, to improve the performance of the neural network, a pre-processor is proposed before the input layer. The pre-processor maps the input into the larger space by generating a set of linearly independent values. The effect of the pre-processor is to increase modelling accuracy, and reduce learning time. Based on this method, experimental results are given to show that the reconstructed surfaces are faithful to the original data points. The proposed scheme is useful for regular or irregular digitized data.     

Design model generation for reverse engineering using multi-sensors

Reverse engineering is the process of creating a design model and a manufacturing database for an existing part or prototype. The applications of reverse engineering are in redesigning of existing partstools or prototype parts where the CAD model of the part is not available. Reverse engineering, for the most part, is performed as an interactive process where the designer identifies the surface features from digitized data and then models the surfaces accordingly. This paper presents the algorithms and implementation results for a reverse engineering system which is intended to automatically create CAD representations of part prototypes. An integrated sensory system combining contact and non-contact sensors has been developed to digitize parts surfaces. The sensory system fuses data from machine vision and a coordinate measuring machine (CMM) in order to automatically digitize the part surface. Machine vision is used to capture the orthographic views of the part. The images of these orthographic views are processed and vectorized to create five views of the part in the form of an engineering drawing. The system utilizes the generated orthographic projections to automatically drive the CMM to capture a grid of point coordinates from the part surface. The CMM digitization process is guided by the segmentation provided from the orthographic views. The segmented data from the part surface is input to the surface modeling module of the system where parametric surfaces are fitted through the digitized points. The surfaces are then extended and intersected using the Hermite approximation method to develop the 3-D CAD model of the part. Accuracy and automation is achieved by combining global shape information obtained from part images with the accurate point data acquired by a CMM. Algorithms for surface segmentation, part digitization, surface extension, and surface intersection modeling are described in this paper.     

Product development for Chinese calligraphy using reverse engineering and rapid prototyping

Chinese calligraphy is considered to be one of the most beautiful and meaningful patterns in the world. Calligraphy is an art dating back to the earliest day of Chinese history, and widely practised throughout ancient China even to recent days. Owing to the passage of time, many old carvings of stamp characters have gradually faded. How to preserve these old Chinese cultural artifacts will be some crucial issues for today. Reverse engineering (RE) and rapid prototyping (RP) technology are considered effective tools for product development. This research combines RE, grey prediction theory in pattern processing, geometric modeling for constructing the characters of Chinese calligraphy and RP for model making. First, the written Chinese calligraphy is scanned by a scanner. By using image processing, the contours of the Chinese characters can be detected. The contours for the characters then turn out to be point data which can be easily processed in any computer aided design (CAD) software by using B-Spline curves to fit the points. An example is demonstrated by writing a Chinese compound word ‘the fortune and treasure are coming’ to show the process. Finally, a RP model is fabricated to show various applications for the products in Chinese calligraphy pattern. By this research, the gap between the Chinese calligraphy in handicraft and digital virtual design can be realized. More usefully, with this research the aesthetic aspects in the characters can be preserved.     

Product development for Chinese calligraphy using reverse engineering and rapid prototyping

Chinese calligraphy is considered to be one of the most beautiful and meaningful patterns in the world. Calligraphy is an art dating back to the earliest day of Chinese history, and widely practised throughout ancient China even to recent days. Owing to the passage of time, many old carvings of stamp characters have gradually faded. How to preserve these old Chinese cultural artifacts will be some crucial issues for today. Reverse engineering (RE) and rapid prototyping (RP) technology are considered effective tools for product development. This research combines RE, grey prediction theory in pattern processing, geometric modeling for constructing the characters of Chinese calligraphy and RP for model making. First, the written Chinese calligraphy is scanned by a scanner. By using image processing, the contours of the Chinese characters can be detected. The contours for the characters then turn out to be point data which can be easily processed in any computer aided design (CAD) software by using B-Spline curves to fit the points. An example is demonstrated by writing a Chinese compound word 'the fortune and treasure are coming' to show the process. Finally, a RP model is fabricated to show various applications for the products in Chinese calligraphy pattern. By this research, the gap between the Chinese calligraphy in handicraft and digital virtual design can be realized. More usefully, with this research the aesthetic aspects in the characters can be preserved.     

Managing engineering data for complex products

In this article we propose a design framework that integrates many application programs and manages the enormous quantity and variety of data generated while designing a complex product. Our discipline-independent framework provides data management services, such as version control and configuration management, needed by many engineering disciplines. This discipline-independent framework may be extended to satisfy the particular needs of specific engineering disciplines. The framework supports a message passing system and a flexible formula language, which facilitate the integration of design data from multiple disciplines.Our framework also includes powerful features for designing products with customer selectable options. First, we provide constructs, called option variables and option restrictions, for indicating what options are to be offered. Then, our option language allows designers to specify the product parametrically as a function of the option variables. Finally, we provide features for verifying that the design satisfies specified constraints for all combinations of options to be offered.     

On the reliability of reverse engineering results

Determination of actual parameters of manufactured optical coatings (reverse engineering of optical coatings) provides feedback to the design-production chain and thus plays an important role in raising the quality of optical coatings production. In this paper, the reliability of reverse engineering results obtained using different types of experimental data is investigated. Considered experimental data include offline normal incidence transmittance data, offline ellipsometric data, and online transmittance monitoring data recorded during depositions of all coating layers. Experimental data are obtained for special test quarter-wave mirrors with intentional errors in some layers. These mirrors were produced by a well-calibrated magnetron-sputtering process. The intentional errors are several times higher than estimated errors of layer thickness monitoring, and the reliability of their detection is used as a measure of reliability of reverse engineering results. It is demonstrated that the most reliable results are provided by online transmittance data.     

Automatic skin segmentation and tracking in sign language recognition

Skin segmentation and tracking play an important role in sign language recognition. A framework for segmenting and tracking skin objects from signing videos is described. It mainly consists of two parts: a skin colour model and a skin object tracking system. The skin colour model is first built based on the combination of support vector machine active learning and region segmentation. Then, the obtained skin colour model is integrated with the motion and position information to perform segmentation and tracking. The tracking system is able to predict occlusions among any of the skin objects using a Kalman filter (KF). Moreover, the skin colour model can be updated with the help of tracking to handle illumination variation. Experimental evaluations using real-world gesture videos and comparison with other existing algorithms demonstrate the effectiveness of the proposed work.     

集成逆向工程系统的开发与研究

本文介绍了逆向工程理论，建立了一个基于激光三维测量的集成逆向工程(Integrated Reverse Engineering，IRE)系统模型，并对其关键技术进行了深入研究。该系统首先利用先进激光测量方法对实物进行三维数字化处理，借助专用的曲面重构软件及CAD系统反求出实物的CAD模型，然后再按照传统的制造流程生产出产品或原型。     

Optical imaging: three-dimensional approximation and perturbation approaches for time-domain data

The reconstruction method presented here is based on diffusion approximation for light propagation in turbid media and on a minimization strategy for the output-least-squares problem. A perturbation approach is introduced for the optical properties. Here we can strongly reduce the number of free variables of the inverse problem by exploiting a priori information such as the search for single inhomogeneities within a relatively homogeneous object, a typical situation for breast cancer detection. We achieve higher accuracy and a considerable reduction in computational effort by solving a parabolic differential equation for a perturbation density, i.e., the difference between the photon density in an inhomogeneous object and the density in the homogeneous case being given by an analytic expression. The calculations are performed by a two-dimensional finite-element-method algorithm. However, as a time-dependent correction factor is applied, the three-dimensional situation is well approximated. The method was successfully tested by use of the University of Pennsylvania standard data set. Data noise was generated and taken into account in a modified data set. The influence of different noise on the reconstruction results is discussed.     

Computer-aided reverse engineering of the human musculoskeletal system

The objective of this study is to extend the applications of reverse engineering technology from manufacturing industries to the biomedical industry. By obtaining nearly exact geometric data of human tissues, such as bones, tendons, and ligaments, in a high-speed and inexpensive manner, potentially groundbreaking research becomes possible for applications in injury rehabilitation, injury prevention and strengthening. Previous applications of reverse engineering technologies in the biomedical community have dealt largely with prosthetic design and plastic surgery. This study expands this research to include muscular and skeletal applications. The primary advantage provided by scanning technologies is an improved quality of data. Secondly, the time investment for the medical researcher is greatly reduced. And finally, the use of a scanning technology will occasionally provide a less expensive alternative to the medical imaging counterparts. The goal of the research is to develop guidelines and methodologies for reverse engineering human structures.     

Optical characterization and reverse engineering based on multiangle spectroscopy

We perform characterization of thin films and reverse engineering of multilayer coatings on the basis of multiangle spectral photometric data provided by a new advanced spectrophotometer accessory. Experimental samples of single thin films and multilayer coatings are produced by magnetron sputtering and electron-beam evaporation. Reflectance and transmittance data at two polarization states are measured at incidence angles from 7 to 40 deg. We demonstrate that multiangle reflectance and transmittance data provide reliable characterization and reverse-engineering results.     

Computer-aided reverse engineering of the human musculoskeletal system

The objective of this study is to extend the applications of reverse engineering technology from manufacturing industries to the biomedical industry. By obtaining nearly exact geometric data of human tissues, such as bones, tendons, and ligaments, in a high-speed and inexpensive manner, potentially groundbreaking research becomes possible for applications in injury rehabilitation, injury prevention and strengthening. Previous applications of reverse engineering technologies in the biomedical community have dealt largely with prosthetic design and plastic surgery. This study expands this research to include muscular and skeletal applications. The primary advantage provided by scanning technologies is an improved quality of data. Secondly, the time investment for the medical researcher is greatly reduced. And finally, the use of a scanning technology will occasionally provide a less expensive alternative to the medical imaging counterparts. The goal of the research is to develop guidelines and methodologies for reverse engineering human structures.     

Automatic identification of functional kinematic bone features from MRT segmentation for gait analysis

We present a method for the segmentation of human leg bones and extraction of functional parameters of the femur using MRT images. The novelty consists in the use of dynamic models which will be adapted to the images of individual patients both globally to a whole leg bone and locally to individual parts of a bone. Thresholding and region growing procedures are applied for pre‐processing the images. For some parts of bones, for example the femur ball, we use a three dimensional VRML‐based (Virtual Reality Modelling Language) femur model as a reference in order to make the segmentation method more robust. Based on the segmentation and the 3D VRML model, we can extract functional (biomechanical) femur parameters which are needed for gait analysis.     

Gigacycle fatigue data sheets for advanced engineering materials

Gigacycle fatigue data sheets have been published since 1997 by the National Institute for Materials Science. They cover several areas such as high-cycle-number fatigue for high-strength steels and titanium alloys, the fatigue of welded joints, and high-temperature fatigue for advanced ferritic heat-resistant steels. Some unique testing machines are used to run the tests up to an extremely high number of cycles such as 10¹⁰ cycles. A characteristic of gigacycle fatigue failure is that it is initiated inside smooth specimens; the fatigue strength decreases with increasing cycle number and the fatigue limit disappears, although ordinary fatigue failure initiates from the surface of a smooth specimen and a fatigue limit appears. For welded joints, fatigue failure initiates from the notch root of the weld, because a large amount of stress is concentrated at the weld toe. The fatigue strength of welded joints has been obtained for up to 10⁸ cycles, which is an extremely high number of cycles for large welded joints. The project of producing gigacycle fatigue data sheets is still continuing and will take a few more years to complete.