A re-configurable cross-sectional imaging system for reverse engineering

1、英文翻译A re-configurable cross-sectional imaging system for reverse engineering based on a CNC milling machineThis paper presents a novel reverse engineering digitizing system for full part geometry, which is based on a cross-sectional imaging device built in a NC milling machine. The system successive

2、ly captures a picture for each planar cross-section contour of the part by end-milling and CCD imaging, and builds the geometry for both external and internal surfaces of the complex 3D part based on a set of the vectored cross-sectional contours. The system mainly consists of three components: a NC

3、 milling machine, a cross-sectional imaging device and a computer control unit. Some issues involving the principle and process flow of the system, encasing materials, cross- sectional imaging and NC code generation, etc. are described in detail. Built on an existing NC milling machine, a portable d

4、evice for capturing the cross-sectional images is designed, which includes an isolated light source, a digital camera, a protective case, a rigid arm and a robust tripod. The device, connected to a computer control unit, serves as a highly flexible accessory for the NC milling machine, constructing

5、the cross-sectional imaging systemor reverse engineering. Furthermore, the error analysis and accuracy assessment of the system are also addressed. A typical case is discussed in detail to illustrate the applications of the system. Such are-configurable digitizing system for reverse engineering offe

6、rs a number of advantages, such as the functional extension of an existing NC milling machine, low costs, and rapid construction. As a result, this system provides a feasible and useful scheme for many enterprises to construct their own reverse measuring system based on existing equipment to aid in

7、rapid product development and extend the function of existing equipment.1 Introduction Reverse engineering (RE) is a methodology for constructing the vectored 3D digital model for an existing physical part by various digitizing processes. The digital model can usually be imported into a CAD system f

8、or subsequent redesign and manufacturing process planning. Unlike the traditional manufacturing philosophy about conceptual designs being transmitted into physical products, RE digitizes, analyses, modifies and fabricates products based on the existing objects 14. RE is necessary when working from a

9、 physical prototype rather than starting from a CAD concept model. This is also particularly useful for the product development where CAD was not used in the original design and a part must be replicated. RE can significantly reduce the production lead time and the costs of the part duplication proc

10、esses. The three primary steps in the RE process are part digitization, features extraction and CAD model reconstruction 5. Therefore, the process of RE always starts from part digitization which is a process of acquiring point coordinates on part surfaces by a variety of digitizers. There are vario

11、us commercial systems available for part digitization. Correspondingly, RE methods differ depending on the digitizing devices used, which range from conventional coordinate measuring machines (CMM), laser scanners, ultrasonic digitizers to various optical measuring devices. They can also be classifi

12、ed into two broad categories: contact and non-contact. The CMM is the most commonly used contact digitizing device equipped with a touch-triggered probe, and can usually produce 3D point coordinates of external part surfaces with micron-level accuracy. However, its digitizing speed is relatively slo

13、w compared with most non-contact optical measuring systems. In addition, a radius compensation of the probe has to be considered. Due to the contacting force, the process is not suitable for soft or fragile objects. Another approach is the non-contact digitization of surfaces using optical technique

14、s, which are usually much more efficient in measuring speed and human labour. A number of systems based on optical methods have been developed, such as laser scanners and camera-based vision systems. Laser scanners have a very high measuring speed and adequate resolution 6, but they are so sensitive

15、 to the ambient lighting that the digitizing process usually has to be performed in a specially dedicated or well conditioned lighting environment, and the digitizing accuracy can also strongly depend on the brightness, texture and curvature of the part surfaces . Vision systems have a lower resolut

16、ion (roughly on the order of 10100 m); however, they can capture millions of data points simultaneously over a large spatial range without moving the optical head 7. Obviously, none of these methods are fully suitable for digitizing complicated objects with undercuts, hidden and internal features. A

17、lthough industrial computer tomography (ICT) and magnetic resonance imaging (MRI) are able to image the internal structure of a part, they are quite expensive and poor in accuracy, and usually require a well trained person to operate them. Furthermore, MRI is not suitable for various metallic parts.

18、A number of studies from both the academia and industrial bodies have been performed to explore new RE digitizing methods or devices which can obtain both the internal and external geometry of complex objects simultaneously. Chang and Chiang 8 presented a method of three dimensional image reconstruc

19、tions for complex objects by an abrasive computed tomography apparatus. The apparatus uses an abrasive process to remove the inlaid object, layer by layer, and to capture the cross-sectional image of each layer with a CCD camera. A numerical scheme is applied to obtain the Bezier curve of the contou

20、rs for each layer. Yang and Chen 1 described a new reverse engineering methodology based on haptic volume removing. Liu et al. 9 proposed an integrated system of cross- sectional imaging based on reverse engineering and rapid prototyping for reproducing complex objects. Chow et al. 10 developed a la

21、ser-based reverse engineering and machining system that would significantly reduce time for CAD model creation and NC code generation. Feng 11 addressed a methodology of Internet-based reverse engineering, and provided a case study to illustrate its applications in integrating CAD and CAM. Aoyama an

22、d Yun 12 described a system to autonomously measure the shape of an unknown physical object for constructing the computer model of a physical object. The system is composed of two subsystems, namely, one for the rough recognition of a physical object and the other for the precise measurement. Liet a

23、l. 13 presented a reverse engineering system for rapid modelling and manufacturing of products with complex surfaces. The system consists of three main components: a 3D optical digitizing system, a surface reconstruction software and a rapid prototyping machine. The unique features of the 3D optical

24、 digitizing system include the use of a white-light source, and a cost-effective and quick image acquisition. CGI has developed a cross-sectional imaging and digitizing system based on a milling machine for simultaneously capturing both the external and internal geometry of any complex parts 14, whe

25、re the milling process is performed successively to capture the planar image of each cross section. However, CGI cross- sectional scanning system is actually a dedicated machine tool, and rather expensive, and the vibration of mechanical components may affect the measuring accuracy of its imaging co

26、mponents. A number of related studies, such as the cross-sectional imaging process, interpolation, data reduction, 3D model reconstruction, and error analysis have also been investigated profoundly 1523.From the preceding literature review, it can be seen that it is usually difficult to capture exte

27、rior and interior shapes of a part concurrently with both high accuracy and low cost by using the current reverse engineering methods. This paper is organized as follows. Section 2 presents the principle and architecture of the system. Implementation of key techniques is described in detail in Sect.

28、 3. The error analysis and accuracy assessment of the system are discussed in Sect. 4. An actual case study is demonstrated in Sect.5. Finally, Sect. 6 concludes the paper.2 Principle and architecture of the system2.1 Principle and process flowThe process flow of the RE system based on cross-section

29、al imaging can be divided into three main steps: object preprocessing, cross-sectional image acquisition, and CAD model reconstruction. The overall flow chart of the process is shown in Fig.1. First, an object is encased within a special encasing material that covers and fills all the internal and e

30、xternal features of the object. The preprocessing stage mainly completes the three sub-tasks: determining the optimum encasing orientation, preparing the encasing material and selecting the proper encasing process. Subsequently, the encased part is clamped by a fixture lying on the table of a CNC mi

31、lling machine. The encased part is then chipped away layer by layer and the cross-sectional image of each layer is captured by a digital camera mounted on a combined device. While the cross-sectional image of each layer is sent to the data processing software. The software automatically files data o

32、f each layer to the hard disc in a control unit to ensure no information is lost. The process, including clamping the encased part, chipping the encased part layer by layer, and capturing the cross-sectional image of each layer, is called cross-sectional image acquisition. A schematic diagram of the

33、 cross-sectional imaging system is shown in Fig2. Finally, all the acquired cross-sectional images are processed through special data post-processing software to reconstruct a CAD model of the object. The data post processing for the cross-sectional images involves the image pre-processing, image se

34、gmentation and edge detection, as well as contour extraction and representation. Figure 3 illustrates the data processing flow for the RE system proposed. The 3D CAD model reconstruction involves fitting a surface to a sequence of evenly spaced planar contours that have been extracted from cross-sec

35、tional images of the part. A number of commercial software systems can be used to aid the reconstruction of a3D CAD model (e.g. Mimics, Digisurf, Surfacer, etc.).2.2 System architectureIn order to implement the reverse engineering process based on slicing digitization, a novel cross-sectional imagin

36、g system for reverse engineering based on an existing CNC milling machine, which can simultaneously acquire the full cross-sectional contours of a part with complex external and internal surfaces, has been developed. The system is comprised of two subsystems: hardware and software. The hardware subs

37、ystem mainly consists of the following components: a combined device to capture the cross-sectional images of an object which includes an isolated light source, a digital camera, a case, a rigid bar and a robust tripod; a CNC milling machine (or a CNC grinding machine); and a computer control unit.

38、The computer control unit, including a computer (PC), an I/O board and a position switch, carries out the automatic control of cross-sectional imaging. The computer with a USB interface and PCI expansion slot is employed to implement the control functions. A digital I/O card which can be plugged int

39、o the PCI expansion slot is employed to transfer the signals between the computer and the CNC milling machine. A normally closed position switch produced by Omron is used to automatically generate the control signal for the image acquisition system. As a critical component, the cross-sectional imagi

40、ng system is described in detail in Sect.3.2. A robust tripod which is an independent adjustable fixture device is used to mount the cross-sectional imaging system. The mechanism can implement the quick positioning between the CNC milling machine and the image capturing unit. In addition, it is inde

41、pendent of the CNC milling machine. Therefore, the combined device and the computer control unit can be considered as a high compliant accessory which might be integrated with a CNC milling machine to form quickly a cross-sectional imaging system for reverse engineering. Such a re-configurable RE di

42、gitizing system has a number of advantages, such as extending the function of an existing CNC milling machine, and reducing the cost and lead time. A number of CNC milling machines can be utilized for the RE digitizing system. Based on the accuracy of the reverse digitizing system and the existing N

43、C milling machine, a universal tool milling machine is adopted for the prototype system developed by the authors here. Figure 4 illustrates a diagrammatic sketch of the cross-sectional imaging system.3 Implementation of key techniques3.1 Encasing material and processThe process starts by encasing a

44、part using a special material which covers and fills all the internal and external features of the part. He encasing material plays a significant role in the cross-sectional imaging system. As the most important function, the encasing material can provide a background image with high grey gradient c

45、ompared with the part digitized, so that the cross-sectional contour of the part can be easy detected and extracted in the subsequent image processing phase. In addition, the encasing material can also reinforce the intensity of the measured part, and improve machining properties of parts with thin-

46、 walls and multi-cavity features.Cross-sectional images acquired are gray-scale images. Therefore, both region and edge based segmentation methods can be employed. The binary segmentation of image processing involves thresholding and the grey gradient of the image under the control of a value. Thres

47、holding is the process of converting a multi-level image into a binary image. In a binary, each pixel value is presented by a single binary digit. In its simplest form, thresholding is a point-based process that assigns the values of “0”or“1”to each pixel of an image based on comparison with a globa

48、l threshold value T.A cross-sectional image only has one object that is distinct from its background. The grey-level histogram of the image will always be a bimodal distribution located apparently far from each other. Therefore, one threshold value is sufficient to segment image into object and back

49、ground. Given segments, edge detection is based on the calculation of gradient operator G.Based on the analysis above, in order to acquire the high-quality cross-sectional images, the encasing material must satisfy the following requirements: high grey gradient between the digitized part and the enc

50、asing material, good mechanical properties and proper toughness. The authors have developed a proper encasing material which used a E-44 bisphenol epoxy resin as the bulk material, a 650 silon resin as the firming agent and elasticizere, a black graphite powder, as the weighting agent. Furthermore,

51、a number of experiments to evaluate the machining capability have also been conducted for determining the optimum charge ratio. The formulation of the encasing material obtained is shown in Table 1. A number of actual cases have demonstrated that the encasing material developed, which has a high gre

52、y gradient, low shrinkage ratio, good bonding and machining performance, can well satisfy the demands of practical engineering.After the material is fully mixed according to the above formulation, the part is gradually filled with the encasing material. The completely encased part is then placed in

53、a vacuum chamber where all air bubbles are removed and all voids are 100% filled. In addition, the part needs to be purified by using an acetone (gasoline oil or alcohol) to remove the greasy dirt, dust, and oxidants in the internal and external surface of the object before the encasing process. The

54、 solidifying time of the encased part should typically be no less than 12 h. The solidifying speed needs to be adjusted according to the ambient temperature.3.2 Cross-sectional image acquisition systemAs the most critical component, the image acquisition system can apture and file automatically the

55、cross- sectional images of the digitized part. The three primary modules including the image capturing device, the light source, and the automatic generation of NC codes will be discussed in detail in the following sections.3.2.1 Selecting the capturing image deviceA combination of a CCD and a data

56、acquisition card is used frequently as an image capturing device. However, it has several limitations: low resolution, high fluctuation of the image grey gradient, and uncertainty in positioning an image. With the rapid development of digital imaging techniques, various digital cameras as new image

57、capturing devices which are characterized by the low cost and high resolution present a good cost/performance ratio to obtain the high quality images. Furthermore, a digital camera can provide a USB interface which can implement high speed communication between a computer control unit and a can be r

58、eal-time processed. The cost-effective ratio for a variety of digital cameras was compared and a digital camera produced by Kodak (DC 4800 model) was chosen as the capturing image device in the prototype system developed. The resolution of the camera is 3.1 million pixels. In other words, it has a r

59、esolution of 2,160 pixels horizontally and 1,440 pixels vertically.3.2.2 Designing the light sourceThe light source has a significant effect on the image quality captured. It is especially important to select a proper light source and determine optimum layout between the camera and light source for capturing high-quality cross-sectional images. The light source for the system needs to meet the following conditions: constant, uniform and matching with ambient intensity. The ideal candidate should