(August 18, 1999)
Machine vision applications are expanding rapidly into areas as diverse as collision avoidance and the inspection of produce. The speed and accuracy with which automated inspection systems operate provides the opportunity to not only make real time accept/ reject decisions but also perform process control by measuring critical dimensions. CCD cameras and structured lighting provide the basis for the vast majority of inspection and process control applications but there are imaging niches where the use of laser scanning can provide superior vision. Laser scanning will not replace CCD systems in most applications but it is important to understand this technology as a complement to existing machine vision solutions. The system integrator that has a good working knowledge of both technologies can select the best solution for a given application.
Applications where laser scanning based machine vision should be considered include applications where ambient lighting is not ideal, the item being inspected is light sensitive, highly polarized light is required to increase contrast, a large number of pixels is desired, fluorescence imaging is used or narrow wavelength bands are needed to increase contrast. Laser scanning systems are also not as adversely affected by inclement weather. The properly selected laser can see through rain or fog where a conventional camera system would be blinded.
A drawback of laser scanning systems is the use of moving components. This risk has been significantly reduced with the introduction of long life, self-generating air bearings. These bearings exhibit no wear at operating speed since there are no physical contacting surfaces. Figure 1 shows the new compact configuration of these assemblies.
Laser scanning based machine vision applications can be segregated into 2D and 3D. In the 2D applications a rotating polygonal scanner or galvanometer is used to produce a scanning spot. The second axis of the raster is produced by either object motion or by a secondary scanning means such as a galvanometer. The scene is interrogated pixel by pixel, line by line serially. This is quite different than the CCD camera systems that view the scene in a parallel fashion and then extract scene information serially. CCD cameras now employ many output taps to increase readout speed.
Inspection of continuous moving webs of material is well suited to machine vision. CCD systems and laser scanning systems can inspect on the fly by freezing image motion. Linear array CCDs perform well in web inspection but if you need more than 2,000 pixels across the web the prices climb rapidly. Laser scanning can provide up to 25,000 pixels across a web at high speeds. The scanning systems can be set up to scan various web widths with user defined pixel sizes. Laser wavelength can be selected to optimize imaging and also to avoid interacting with the material being inspected (such as film). Light collection can be designed to acquire images using specular reflection, diffuse reflection, fluorescence, transmission or a combination of these techniques. Figure 2 provides a schematic of a typical web laser scanning system.
Application example – web inspection
Laser scanning is well suited to interrogate developed film on a pixel by pixel basis. A custom vision system was created to satisfy a customer’s need to digitize film images from 70mm film. The film was in a continuous web form so a web scanner was designed that used a polygon to generate one axis of scan and a film transport to provide the second. The microdensitometer was designed with a HeNe as the interrogation source and transmitted light was collected. A collection integrating cylinder was used to gather the diffusely transmitted light and detectors located at each end were summed to produce the signal. The system was operating from a density 0 to 3.0 so small amounts of stray light would negatively impact data integrity. Spatial filtering was used to reduce the optical scatter emitted from the laser. The requirements for low scatter off the polygon prohibited the use of diamond turned optics and required the use of conventionally polished nickel plated aluminum polygons. Diamond turned polygons can be produced to about 50 Angstroms rms surface finish while conventionally polished nickel plated polygons can achieve less than 15 Angstrom rms finishes.
A lens turret enabled the use of four different spot sizes with polygon and film transport speeds changing automatically to maintain the proper pixel aspect ratio and data rates. Spot sizes ranged from 12.5 microns to 100 microns. The system interrogated the film to 8 bit digitize the images at a 2 Mhz data rate.
Conveyor based 2D inspection
This common application for machine vision enables manufacturers to inspect the quality of the goods they are producing at high speed. The systems can be designed to look at size, shape, alignment, presence or absence of features, or special feature characteristics. A standard inspection head is not usually practical since each application requires a special configuration of similar generic techniques.
Printed Circuit Board Inspection Machine
This is an application where machine vision helped improve inspection by eliminating the fatigue factor associated with inspectors looking at large numbers of bare printed circuit boards. This application required bare 18 inch by 24 inch printed circuit panels to be inspected at one mil resolution for a variety of defects including opens, shorts, missing traces, and narrow traces. Design rules and data base comparison techniques were used to evaluate the acquired image. The scanner used fluorescence to gather information with the 488nm line from an argon ion laser scanned across the board in a 4 inch long swath. PCB substrates fluoresce when exposed to this wavelength whereas copper traces do not, providing good contrast that could not be achieved with conventional lighting and CCD arrays. Fluoresced light is collected with an imaging rod, imaging the scan line onto the face of a linear fiber optic array. The array converts the collected light to a circular profile and delivers the light to photomultiplier detectors. Three scanning heads were used in parallel to increase system throughput. Figure 3 is a picture of a scan head on an alignment bench.
While 2D systems make up the majority of machine vision applications the promise of 3D-machine vision captures the publics’ imagination. Robots moving about with their own vision systems, automobiles that see and react without the need for steering input, 3D input scanning of home interiors for renovation projects all provide exciting possibilities that would impact our lives. The technology to achieve these goals is approaching rapidly, from the vision systems needed to acquire the images to the image processors required to enhance the images and automatically make decisions based on the scene information content.
CCD technology provides solutions in this 3D world with stereoscopic imaging and structured lighting but this is an area where laser scanning technology can have an impact with a variety of means available to gather scene information. Laser scanning can be used at the microscopic level, mid-range levels and long range levels with appropriate technology switching. 3D scanning incorporates many of the features of 2D scanning but adds the third dimension by a number of means including confocal imaging, triangulation or time of flight measurement.
At the microscopic level laser scanning is used in scanning confocal microscopy. Images are acquired by imaging back through the optical system and using an aperture to define the image plane. The sample is stepped though focus and the reimaged object is observed. Two-dimensional slices through the object are taken until the three dimensional representation is built up.
Mid Range Sy
Laser Scanning, the Other Machine Vision Technology
(August 18, 1999)