A three-dimensional scanner is a device that analyzes the objects or their immediate environment to gather specific information on the form and possibly on the appearance (color, texture) of these. The data collected can then be used to build CGI three-dimensional (digital objects) for various purposes.
These devices are much used by the industries of entertainment for movies or video games. Digital images of scanned objects in 3D are also used for industrial design, the design of prostheses, reverse engineering, for quality (digital repository) control or for the documentation of cultural objects.
Sans-contacts scanners may be subdivided into two main categories, active and passive scanners. They are themselves into many subcategories based on their technological principle.
This Lidar scanner can be used to scan buildings, geological formations, etc. in order to produce a model in three dimensions. Its RADIUS is adjustable over a very wide horizon : thanks to the horizontal rotation of the head, a mirror directs it vertically. The laser beam is used to measure the distance with the first object cutting beam
The 3D Lidar scanner is an active device that uses a laser beam to probe the subject. At the heart of this type of scanner is a laser rangefinder to determine the distance from the surface of the object studied by counting the time required for return of the pulse of the reflected laser beam.
Since the speed of light c is known, the round-trip time to determine the distance traveled by light, which is twice the distance between the scanner and the surface. Of course, the accuracy of the scanner by time of flight depends on the accuracy of the measurement of the return time t, knowing that 3.3 picoseconds is approximately the time taken by light to travel one millimeter.
The laser rangefinder detects only one point at once in the direction it is pointing. For this, the device scans all of its field of view point by point and must change its direction of view to each measure. It can be changed by rotation of the camera itself or by using a system of rotating mirrors. This last method is the most commonly used because mirrors are lighter and can change direction more quickly with more precision.
Time of flight 3D scanners can measure the distance from 10 000 to 100 000 points per second.
Another technology used by laser scanners to measure distances is the measure of phase shift. The scanner emits a laser beam, which, in contact with the object, is reflected to the laser scanner. The wavelength of the laser emission varies according to the provider. The mirror of the scanner returns the laser beam vertically towards the same object. The vertical angle is encoded at the same time as the distance measurement.
The laser scanner rotates 360 ° on itself in the horizontal. The horizontal angle is calculated simultaneously with the distance measurement. The distance and the angle vertical and horizontal give a polar coordinate (δ, α, β) that is converted to Cartesian coordinate (x, y, z). Some laser scanners use the phase shift measurement technology to measure the distance to a surface. The device projects an infrared laser beam which returns to the scanner by reflection. It calculates the distance to the nearest millimeter by analyzing the phase shift between the emitted beam and received RADIUS.
The laser of a known sine wave is broadcast by a laser source.
It's the 'light '. Some of the laser beam is reflected from the target to the source. Is called \back light\. The phase of this \back light\ is compared to that of the light emitted known to determine the 'light history'. The difference between the two peaks is called \phase shift\. The phase shift obtained corresponds to 2π x flight time x the frequency of modulation. Phase shift scanners are usually faster and more accurate than 3D in time of flight laser scanners, but they have a smaller scope.
Principle of a detector using laser triangulation. Two positions of the object are displayed.
The laser triangulation scanner is an active scanner that also uses laser light to probe its environment. He points to the subject with a beam as for one by flight time and uses a camera to locate the point. Depending on the distance to a surface, the point appears at a different place in the field of vision of the camera. This technique is called triangulation because the point laser, the camera and the laser emitter form a triangle. The length of a side of the triangle, the distance between the camera and the laser transmitter is known.
The angle on the side of the laser transmitter is also known.
The angle on the side of the camera may be determined by looking at the location of the laser dot in the field of vision of the camera. These three data determine the shape and the dimensions of the triangle and give the position of the laser point. In most cases, a laser instead of a period band, scans the object to accelerate the acquisition process. The national Council of research Canada was among the first institutes to develop a technology of scan based on triangulation in 19782.
In a conoscopic system a laser beam is projected onto a surface, then thinking through the same beam passes through a birefringent Crystal and is sent on a CDD sensor.
The frequency of diffraction patterns can be analyzed and used to determine the distance to the surface. The main advantage of conoscopic holography is collinearity, that is, a single beam (round trip) is needed to perform the measurement, to measure for example the depth of a hole drilled finely which is impossible by triangulation.
Manual laser scanners create images 3D from the triangulation principle described above : a point or a laser line is projected onto an object using a manual device and a sensor (typically a CDD sensor or position sensitive device) measures the distance to the surface.
Positions are saved to an internal coordinate system and the scanner itself being moving its position must be measured. The position can be determined by the scanner using characteristic landmarks on the surface being scanned (typically of adhesive reflective strips) or using an external tracking method. The unit responsible for this identification comes in the form of a Machine to measure three-dimensional equipped with a camera incorporated (to set the orientation of the scanner) or as a device for Photogrammetry using three or more cameras allowing the six degrees of freedom of the scanner.
Both techniques tend to use infrared leds incorporated to the scanner which are perceived by the (camera (s) through the filters to see them despite ambient lighting.
The information is collected by a computer and saved as the coordinates of points in three-dimensional space, using computer processing, these can be converted by triangulation in a canvas and then in a computer model, most often in the form of NURBS surfaces. Hand-held laser scanners can combine this data with passive receivers of visible light - that record the textures and colours - to restore (see reverse engineering) complete a modeling in 3D model.
Structured light 3D scanners project a bright pattern on the subject and to observe the deformation. The pattern may be in one or two dimensions.
Example of a line as a one-dimensional ground. It is projected on the subject using an LCD projector or laser. A slightly offset the projector camera, records his possible deformation. A technique similar to triangulation is used to calculate the distance, and therefore the position of the points representing. Ground sweeps the field of vision in order to save a bunch at a time, information about distances.
Now take the example of a grid or strip-shaped pattern. A camera is used to record the deformations and a complex computer program is used to calculate the distances of the points making up that ground. The complexity is due to the ambiguity. Take a group of vertical stripes sweeping horizontally a topic. In the simplest case, the analysis is based on the presumption that the sequence of bands visible from left right matches the image projected laser in such a way that the image of the band the leftmost is the first laser projection, the following is the second and so on.
In the case of non-triviales targets with holes, some occlusions, rapid depth changes, however, the order is necessarily verified that bands are often hidden and may even appear in a different order, giving rise to an ambiguity in the bands lasers.
This specific problem has recently been resolved by an advanced technology called Multistripe laser Triangulation (MLT). The structured light 3D scanning is still an active area of research, giving rise to a number of publications each year.
The highlight of the structured light 3D scanners is speed. Instead of scanning a point at a time, they scan the entire field of vision at the same time. This limits or eliminates distortion problems related to the movement. Existing systems are able to scan objects in motion in real time. Recently, Song Zhang and Peisen Huang from Stony Brook University have developed a scan on the fly using a digital fringe projection and a modulated phase technique (another structured light method).
This system is able to capture, rebuild and restore the details of objects deforming in time (as a facial expression) at a frequency of 40 frames per second.
The modulated light 3D scanners illuminate the subject using a changing light. Usually, the light source has a cycle whose amplitude describes a sinusoidal pattern. A camera detects reflected light, measures the importance of its variation and determines the distance the light has traveled.
The modulated light also allows the scanner to ignore the source of light other than a laser, so that there is no interference.
Passive scanners without contact, being issuing any type of radiation, are based on the detection of reflected ambient radiation. Most scanners of this type detect visible light because it is immediately available. Other types of radiation, like infrared can also be used. Passive methods can be cheap, because in the majority of cases they don't require device specific show.
Stereoscopic systems usually two cameras videos, slightly apart, pointing to the same scene. By analyzing the slight differences between the images of the two devices, it is possible to determine the distance of each point in the image. This method is based on the vision stereoscopic humaine5.
These types of 3D scanners use outlines created from a sequence of photos taken around an object in three dimensions against a contrasting background. These silhouettes are detached from their background and assembled to each other at the location of the axis of rotation of the camera to form a \Visual hull\ an approximation of the object. With this type of techniques all sorts of concavity of the object - like the inside of a bowl - are not detected.
Scanners seeking the assistance of the user
There are other methods, based on detection and identification assisted the user characteristics and forms a series of different images of an object, which allow to construct an approximation of it. This type of technology is useful to quickly achieve an approximation of an object composed of simple shapes like buildings. Various commercial software are capable as iModeller, D-Sculptor ou RealViz-ImageModeler.
These types of 3D scanners are based on the principles of Photogrammetry. Somehow they use a methodology similar to panoramic photography, with this instead to take images from a fixed point to take a panorama, a series of images from different points is taken from a fixed object to replicate it.
Modeling of data collected by the scanner
The clouds of points produced by 3D scanners are often not usable as what. Most applications do not directly use, but use instead of a 3D model. This means for example in the context of a 3D polygonal modeling to determine and connect adjacent points in order to create a continuous surface. A large number of algorithms are available for this work (for example, photomodeler, imagemodel).