imaging and machine vision europe june/july 2010 www.imveurope.com


Image courtesy of MVTec Software


16


3D vision Strike a pose


Greg Blackman finds that being able to view an object in 3D is providing much needed information for object inspection and positioning


The term 3D vision is a very broad one and encompasses many different technologies. Dr Lutz Kreutzer, manager of PR and marketing at MVTec Software, notes that there are two main objectives for 3D imaging systems, both of which can be achieved by different means: 3D alignment, or finding the 3D pose (position and orientation) of an object, and 3D reconstruction, which is determining the 3D shape of arbitrary objects. ‘To solve a 3D vision problem, it is often necessary to employ and combine different technologies,’ he says. MVTec’s Halcon software platform provides functionality for various ways of acquiring 3D information, which can be laser-based, stereo-based using two or more cameras, monocular 3D, time-of-flight, and Moiré interferometry, among others. One of the more established 3D techniques,


and one commonly used in industry, is laser triangulation. Here, a laser line is projected onto the object and a camera images the deviations in the laser as the part moves across the beam to build up the 3D shape. The technique creates a 3D elevation map of the object and is commonly used for defect detection, measuring volumes, or measuring elevation dimensions. Tyre inspection is a good example of where a scanning laser can be particularly advantageous, as it’s difficult to identify any defects or read a black serial number on a black tyre with standard 2D techniques. Food processing is another area that has adopted laser-based 3D vision. The technique is good for estimating volumes of foodstuffs, such as chicken breasts, as they move along a conveyor belt to portion the food accurately. A 3D setup based on laser triangulation will


give a height output in pixel values. Pierantonio Boriero, product line manager at Matrox


3D vision technologies allow a robot arm to locate and pick up objects in any pose


Imaging, comments that it’s difficult to get the mechanical setup correct with laser-based ranging. ‘Firstly, there are certain constraints on how the laser is projected and where the camera is positioned in order to achieve the desired measurement accuracy and range, which tend to work against each other,’ he says. The Matrox Imaging Library (MIL) contains various 3D vision software tools. According to Boriero, a lot of effort has gone


into making the calibration process for laser triangulation as straightforward as possible without compromising accuracy. There are many ways of calibrating the system, but often a positioning stage will be used to move a known object accurately through the field of view. From this, the system can build a look- up table for converting xyz pixel values to real- world coordinates. Anders Murhed, product unit manager, advanced vision at Sick IVP, notes some disadvantages with this type of calibration method: ‘The object has to be moved very accurately and the stages can be expensive. It


is also often not practical to have to use this equipment to recalibrate cameras on production lines.’ Sick has developed a saw-tooth calibration


tool that the user can move back and forth across the laser plane by hand to build up the calibration matrix. By programming the vision system with the dimensions of the tool’s teeth, the matrix can be produced without the absolute position of the target. For making very accurate 3D measurements,


Stemmer Imaging’s Common Vision Blox (CVB) software now incorporates Metric 3D, a metric calibration system for laser triangulation. The tool uses a complex calibration process that provides accuracies beyond that of standard laser triangulation calibration methods, which use a simple sum of pixels defining the area of the slice. The Metric 3D system profiles a known object


that changes in size, rather like a cone with sharp edges, moving from small to large and then small again, so that every single point in the 3D space


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32