The potential for automatic optical inspection in industry is clearly great since virtually every product is inspected at least once. Although the ready availability of the basic components needed, such as light sources, scanners, image analysers and display systems, has led to the design and application of numerous systems, many quality assurance engineers still need convincing that inspectors can in fact be replaced by machines. The advantages and benefits to be derived from using automated inspection will be discussed, together with techniques which can be used to overcome some of the problems met in high speed automated systems which may be operating in hostile environments. Thresholds of acceptance need to be established in relation to the particular problem and this aspect of inspection traditionally gives rise to the greatest area of disagreement between the supplier of equipment and the customer. The need to agree on the precise method of measurement of the relevant parameter and means for system calibration will be stressed when describing three recently developed inspection systems using a scanned laser, a white light TV system and infrared radiation for the inspection of strip products, microscopic flaws on optical components, and stress patterns on structures subjected to dynamic loading.

This paper reviews our efforts to develop the theory and instrumentation needed to measure surface roughness of manufactured surfaces by optical scattering methods. We are addressing three key problems: developing a valid and sufficient optical scattering theory for this roughness range, applying appropriate mathematical inversion techniques so that practical roughness parameters can be calculated from scattering distributions, and finally evaluating a compact commercial instrument for a wide variety of problems. Recent results from our group suggest that the simple phase screen approximation model of optical scattering validly describes light scattering from machined metal surfaces with a predominant surface lay in the 0.01 pm R to 3.0 pm R range. A model for scattering in the entire farr-field hemisphere and obsera vations on our r approach to the inverse problem is given.

An optical flaw detector with laser as the external light source, which is called LST ( laser scanning tester ), has been developed. This equipment automatically inspects the entire surface of hot slabs. The results are used to examine the suitability of those slabs for hot charge rolling. The characteristics of LST are its high optical resolving power and the signal processing method with which two-dimensional information on the type of the flaw is processed. For the opening width of O.4mm and over, the detection ratio is nearly 100%. This equipment started commercial operation in January 1983 in Nippon Steel's Yawata Works and its application has increased the hot charge rolling ratio.

The determining factors of the glossiness, the relation between surface roughness parameters and glossiness, and the method of discrimination of the grade of glossiness by detecting light reflection intensities were investigated. A new glossmeter was developed as results of the investigation. The grade of the surface glossiness of stainless steel sheets determined by this method coincided with the results by visual inspection with the accuracy of 90%.

Two aspects of the surface evaluation of bright finished aluminum sheet are discussed. The first section of this paper addresses the relationship of optical gloss to surface roughness provided by first-order light scattering theory and the use of this relationship in characterising the surface roughness of bright aluminum sheet from optical gloss measurements. The second part of this paper examines the attributes of Nomarski DIC microscopy for identifying defects on bright-rolled aluminum surfaces. Comparisons are made between DIC micrographs of a typical surface area containing defects and 3-dimensional microtopographical data generated from identical areas by contact profilometry.

Two types of new on-line color measurement systems were developed for the continuous measurement of the color difference or the whiteness of travelling steel sheets. Applying these systems to a production line of colored galvanized steel sheets, it was found that the color difference measured by them coincided well with those obtained by a tristimulus colorimeter.

This paper is concerned with inspecting surfaces using the textural properties of the surface. The approach taken here is that of modelling the surface texture by a "stochastic" model which is parametric, synthesis, compact and parsimonious. Two such models are discussed: the Markov Random Fields and the Fractal models. The first model is very useful for modelling textured surfaces such as textile, lumber, etc; whereas the second one is useful for modelling perceptual surface roughness. Surface inspection is cast as a statistical classification and hypothesis testing problem based on the maximum likelihood estimate (MLE) of the model parameters (or on the sufficient statistics). The image is divided into disjoint square windows and a MLE a* (or a sufficient statistic) is computed for each window . A Mahalanobis metric 11 a* - a 11,p weighted by the Fisher information matrix 'P is computed and compared to a predetermined threshold. This metric is shown to be the quadratic of the likelihood of the data for a large number of samples, and the test is the corresponding chi-square test.

Optical sensors can be used for the non-contact control of mechanical parameters (e.g. surface quality) in fully automated production and assembly systems. The applicability of such optical sensor systems to a particular production line mainly depends on the detection limits (i.e. which size of scratches can be detected on a particular surface). A simple dimensionless number, D, was found describing the detection limit (D>1). The important parameters within D are: the reflectivity of both, the unperturbed surface and the defect, the area of the surface flaw and the noise generated by the statistical surface structure. The detection limits for different common flaws on various characteristic surfaces often present in the metal processing industry will be discussed together with examples of fiber optic sensor designs.

For many years, C.R.M. has been involved in the application of optoelectronic noncontact methods for inspecting steel products especially during hot rolling. Among the first prototypes, the OPTIMAC gauge measures the slab thickness on roughing mills according to an optical triangulation method. A measuring system, called ROMETER-5 has been developed for the objective assessment of the flatness of hot steel strips at the runout table of the finishing mill, i.e. the wavy length of 5 strip 'fibers'. Relative differences in fiber lengths as small as 10^-6, undetectable by visual inspection, can be determined. On a wide flange beam mill, a new system measures the flanges width of hot H-beams whose width x height dimensions are ranging from 100 x 100 up to 500 x 1100 (mm²). This automatic system, called DIMPRO, provides measurements with a ± 1 mm-accuracy in the whole range without requiring any additional sliding guides of the beams in the reversing mill.

A non-destructive technique for processing semiconductor material based on Faraday Rotation (FR) measure-ment has been developed and successfully utilized for evaluation of single crystal Cadmium Sulfide (CdS) wafers. We have extended the technique to the infrared region of the spectrum by utilizing a CO₂ laser, thus significantly enhancing wafer mapping capability of the technique. Since homogeneity becomes a serious problem for large size wafers, considerable cost savings are realized if the wafer is screened prior to array fabrication. CdS Schottky barrier detectors are presently used in the two-color STINGER-POST missile system. In the band edge region, FR was found to vary as λ^-2, where X is the wavelength of the incident radiation, as predicted by the theory for interband transitions. In the infrared, e.g., at 10.6 microns, FR is primarily due to free carriers and varies as λ ². Free carrier concentration calculated from FR measurements agrees well with results obtained by Hall measurements. Advantages of the FR method are: 1) no contacts need be alloyed, hence no material contamination, 2) surface preparation such as etching/polishing is not required, 3) it is a rapid technique and does not require a skilled operator, and 4) by the use of the CO₂ laser, auto-mated wafer mapping is possible. This technique, used here for CdS, is applicable for any semiconductor material where rapid measurement of carrier concentration and a determination of material homogeneity is desired. Development of a self-contained, automated and computer controlled wafer mapping system based on carrier concentration measurement by FR will be described. Results obtained for CdS wafers will be presented

It is today's emergence of thermoelectrically cooled, highly accurate infrared linescanners and imaging systems that has definitely made on-line Infraread Thermography (IRT) possible. Specifically designed for continuous use, these scanners are equipped with dedicated software capable of monitoring and controlling highly complex thermodynamic situations. This paper will outline some possible implications of using IRT on-line by describing some uses of this technology in the steel-making (hot rolling) and automotive industries (machine-vision). A warning is also expressed that IRT technology not originally designed for automated applications e.g. high resolution, imaging systems, should not be directly applied to an on-line measurement situation without having its measurement resolution, accuracy and especially its repeatability, reliably proven. Some suitable testing procedures are briefly outlined at the end of the paper.

An automatic detection system of hot slab subsurface inclusions has been developed for the purpose of eliminating demerits in the conventional flame observation method. The system takes images of the melting area while scarfing the surface of hot slab, and auto-matically detects sparks emitted from defects. The special designed nozzles can perform fin-free scarfing. The system can also classify hot slabs into grades in a way similar to an expert inspector.

Two new methods have been developed to measure the temperature of a moving wire subjected to superimposed vibrations. The first technique is concerned with the measurement of temperatures up to several hundred degrees Kelvin whilst the second is concerned primarily with temperature measurements from about 900K and higher.

On-line infrared absorption gauging has moved from being a curiosity in the early 1970's to a vital inspection and measuring tool in the 1980's. This paper-reviews the response to changing demands and looks at the challenge of the future.

Computer simulations are performed to compare effectiveness of band averaging, angle averaging, Brewster angle, multi-pass, and Fourier filtering methods of interference effect cancellation in film thickness measurement. Results indicate that the Brewster angle and multi-pass methods are superior for a 2-wavelength measuring system. Fourier filtering in the full spectrum systems offer new methods of film thickness measurement.

An automatic, transparent, thin-film-thickness inspection system for complex shapes is discussed. The basic elements of the system are a light section microscope (LSM) used to produce an optical signal proportional to the film thickness and a multiple-axes robot used to automatically position the LSM at the desired inspection points. The system measures film thickness from 0.013 to 0.250 mm (0.0005 to 0.010 in.) and has a repeatability of +0.003 mm (+0.0001 in.). The major development efforts consisted of (1) designing a vision system for theLSM, (2) providing for automatic read-out of the film-thickness data, and (3) final positioning of the LSM using the viewed optical patterns. Also, a simple standard was designed and fabricated that provides for traceability to the National Bureau of Standards (NBS) using a coordinate measuring machine. The completed inspection system has been installed in a production shop.

This paper describes a vision system designed to control the insertion of integrated circuits in PC boards. This task needs both high processing speed and great reliability. To achieve the goal, we use very accurate optical devices and a real time image processor as well as powerful computing resources. A.I. programming techniques based on a prediction/verification principle are implemented.

Product inspection continues to play a growing role in the improvement of quality and reduction of scrap. Recent emphasis on precision measurements and in-process inspection have been a driving force for the development of noncontacting sensors. Noncontacting sensors can provide long term, unattended use due to the lack of sensor wear. Further, in applications where, sensor contact can damage or geometrically change the part to be measured or inspected, noncontacting sensors are the only technical approach available. MTI is involved in the development and sale of noncontacting sensors and custom inspection systems. This paper will review the recent advances in noncontacting sensor development. Machine vision and fiber optics sensor systems are finding a wide variety of industrial inspection applications. This paper will provide detailed examples of several state-of-the-art applications for these noncontacting sensors.

Active optical vision systems rely on the analysis of images on an object illuminated by known structured optical patterns. Approaches differ by the choice of projected pattern (dots, stripes, grids, etc.) and the type of analysis used to extract shape information. A novel approach is reviewed, whereby a sinusoidal grating structure is projected onto an object, and the resulting deformed grating images are detected by a solid state array camera and analyzed by a microcomputer using digital interferometric phase-measuring algorithms. Large arrays of data points can be acquired and processed in parallel with very high accuracy and speed. Two systems are described, one that yields range measurement from a single point of view over a two-dimensional array of points, and a second that permits full 360 degree measurement. The case of parallel laser light illumination, as well as that of general white light diverging illumination are described. Experimental results and performance analysis are presented.

A three dimensional machine vision system, based on-a phase measuring version of moire topography, is described. By rotating an object in front of the system, surface co-ordinates can be acquired from many thousands of points around the periphery of that object. A computational method for comparing the dimensions of an object with its design specification is described. The application of this method is restricted to the subset of objects which can be described in terms of simple primitive shapes.

This report presents a constructed 3-D image acquisition system based on time-of-flight measurements. The basic structure of the system is described and the main factors affecting the performance of the system are introduced. Some recent experimental results of resolution tests are presented. The current image resolution is ca. 1 cm at a working range of 2.5 m to 4 m.

Our multisensor panoramic range finder versatile 3D vision system is described. The problem of the necessary calibration is considered and some solutions proposed, especially with mathematical methods. These solutions are compared with the experimental views. Some parameter changes must be made and the multisensory concept permits these adjustments. The field of applications in view for this system indludes both : - locating and apprehending by a third generation robot. - 3D advanced peripherals for computer vision for picture synthesis and analysis. This latter part of our work is being conducted jointly with the group // synthese d'images tridimensionnelles (3D picture synthesis) of the C.C.E.T.T. (Common Study Center of Television and Telecommunications of Rennes (FRANCE).

We describe how to expand the dynamical range of active triangulation systems. The method is based on the expansion of depth of focus, in order to achieve a larger measuring range - and on the reduction of coherent noise, in order to achieve better resolution of depth. With a 1024 element linear photodiode array we could achieve about 15000 resolvable depth positions, in preliminary experiments.

This paper outlines the development of a 3-D vision system based upon the synchronized scanning principle. The concept uses triangulation and through the use of a novel scanning mechanism, a large field of view is achieved while maintaining resolution. The design of the prototype camera is described. The optical design is based on the requirements set by the first application, that of measuring marine propellers with diameters in the range of 5 metres. The benefits of using this technology over current propeller measuring practices will be discussed. Furthermore the application of the 3-D camera in the measurement of museum artifacts is also demonstrated and results shown.

An autosynchronized laser scanning mechanism is integrated in a compact, mobile unit mounted on a six axis, articulated robot wrist. This camera combines a new geometrical arrangement for improving the performance of optical triangulation with the latest solid-state laser and CCD sensor technology. The camera enables in-process, 3D mea-surements of the welded workpiece and the optimization of the robotic arc welding process. Joint and weld geometry analysis can be performed in real time with high precision, even during high current arc welding. This provides information about weld joint geometry in front of the arc (e.g. seam tracking) and behind the weld pool (e.g. inspection of finished welds), both of which are required for closed loop adaptive welding. High dimensional resolution combined with high signal to noise ratio provides an ideal tool for the further development of expert welding systems. Furthermore, the vision system can be used for the generation of 3D object data, which can be used in conjunction with a computer graphics system for offline robot programming.

An installation for automatic acquisition of tridimensional objects, in particular human feet is described. The helical method, the mechanical setup, the computations on data acquired by the optical camera and the data smoothing process by cubic spline functions are discussed. Typical results of tridimensional acquisitions are presented.

This paper presents a method for building the model of a 3D object from a range image provided by a fast range finder. The model is an adjacency graph describing the connectivity properties of neighbouring regions (facets) of the object. Each region is obtained from the segmentation scheme based on the analysis of the hemispheric histogram. The histogram organizes the surface orientation information. The model is built in an unsupervised manner and provides a compact description of the object. The model builder could be integrated in an automatic 3D object recognition system.

An approach to object recognition is discussed in which surfaces of objects in 3-D space are approximated by, pieces of spheres,cylinders and planes,and these pieces are recognized. This processing is realized by partitioning an image into regions and processing these regions in parallel. The purpose of this paper is to describe a technique which uses a recorded image of a camera with laser scanner to measure coordinates of 3-D surface points. The three of them in measured points in a region are then used to determine a plane and calculate further distances from feature points in the region to the plane. The multiple distance features can be used for recognizing whether the surface in a region is a piece of a sphere, a piece of a cylinder,or a piece of a plane. The surface orientations can be discribed by surface normal which has two oblique angles. This technique is useful for many purposes such as finding the three-dimensional positions and orientations of parts and calculating their surface shape properties which provide informations for industrial inspection.

We have used Fourier transform infrared (FT1R) spectroscopy to evaluate the contamination of stainless steel tubing with inner diameters of less than one mm, and lengths of 25 to 100 mm. We will discuss the development of a non-destructive analytical tech-nique which is sensitive to both organic and inorganic contaminants which may arise during fabrication processes. Experimental results will be compared with theoretical calculations for thin films of hydrocarbon lubricants and metal oxides on metallic surfaces. The main advantages of infrared measurements are their non-destructive nature, ease of application, short measurement times, and the ability to interface the necessary auxilliary optics with existing FTIR instruments. The sensitivity of the technique and its application to the evaluation of current cleaning processes will be presented.

Among the many advantages of Raman spectroscopy for polymer characterization, low frequency data have been the least exploited. Results are presented on the use of low frequency Raman data on polyester fibers for determining the relative specific heats, and the degree of amorphous orientation. Background theory is presented.

The feasibility of using diffuse reflectance Fourier transform infrared spectroscopy as a rapid non-destructive technique for detecting mould release agent on the surface of epoxy-based composites has been investigated. It has been shown that it is a viable method if the amount of release agent is greater than about 200 μg/cm². This limit of detection is determined mainly by the effects of front-surface reflection and scattering. These cause some degree of variation in band shapes and relative intensities which makes it difficult to perform spectral subtraction in order to better detect weak features resulting from surface contamination.

The techniques of Fourier transform infrared microspectroscopy are reviewed. The technique which can be used to obtain the infrared transmission and reflection infrared spectra of very small samples is extremely useful for the characterization of a variety of chemical systems such as polymers, biological materials, and semiconductor materials. In this paper, the applications of the microspectroscopic techniques to the characterization of polymer systems will be illustrated and discussed. The method can be used to identify contaminants in polymer materials, identify different layers in multi-layer polymer packaging materials and characterize paint samples. Unlike other microsampling methodologies, the infrared microsampling techniques will yield direct information about organic materials and functional groups and has the potential to solve a number of real life, industrial problems.

Laser excited fluorescence of black liquor was investigated as a possible monitoring technique for pulping processes. A nitrogen pumped dye laser was used to examine the fluorescence spectrum of black liquor solutions. Various excitation wavelengths were used between 290 and 403 nm. Black liquor fluorescence spectra were found to vary with both excitation wavelength and black liquor concentration. Laser excited fluorescence was found to be a sensitive technique for measurement of black liquor with good detection limits and linear response over a large dynamic range.

A study of the light absorption behaviour of kraft pulp suspensions has been carried out. The purpose of the study was to determine whether the measurement of the ultra-violet light absorption coefficient could be exploited as a basis for the development of an on-line lignin content sensor. In order to obtain this measurement, it was first found necessary to compensate for the effects of the pulp light scattering and consistency variation. A method was developed to provide this compensation; however, it appears that the measurement is also significantly affected by species variation.

A study of the infrared (IR) absorption characteristics of dried kraft pulp sheets was made. This was done in order to assess the potential of using this approach as the basis for determining residual lignin, or Kappa number, in pulp after cooking. Strong positive linear correlations were obtained between Kappa number and IR absorbance at 1509 cm-for pulps made from different wood species, produced in different mills and having a Kappa number range of 13-37. For pulps from some mills, made from the same wood furnish and having a small Kappa number range, the degree of correlation was seriously reduced. The method requires the use of moisture-free pulp specimens in the measurement of absorbance. It is suggested that it would be more suitable as the basis for a laboratory instrument than for an on-line, process Kappa number sensor.

The following describes a recently developed rapid scan spectrometer (5 scans per second) used for the measurement of process constituents. The analyzer is a continuous high resolution, scanning instrument using a wide spectral range, and is optimized for laboratory and on-line use. It uses a unique SpectraMetrix software package which enables quantitative determination of constituents and physical parameters within complex samples. The instrument and the technology is now used extensively in industrial applications including, petrochemicals, automotives, pharmaceuticals, textiles, adhesives and more.

Color is a very important property of many products and an essential feature of some. The commercial value of color is evident in the fact that customers reject product that is satisfactory in every other way, but is not the right color. Color isrumerically specified, measured, and controlled just as length or weight are. It has three dimensions: Hue, Value, and Chroma, and may be represented in a three-dimensional space. Colors of objects depend on the illumination and pairs of colors may match in one light but not in another. Controlled illumination is required for color matching. Illuminants were standardized by the International Commission on Illumination (CIE). As a basis for color measurement, the CIE adopted three spectral sensitivity functions representing a standard observer. Color may be measured by instruments using standard illumination and simulating the standard observer. It is better to measure spectral reflectance or transmittance and compute colorimetric quantities. Color may be inspected on a production line and the data obtained can be used to control the process. When production cannot be controlled as precisely as required, product may be sorted by color.

A remote sensing technique, based on the principles of emission spectroscopy, is being developed for temperature measurements in black liquor recovery boilers. Several tests have been carried out, both in the laboratory and at a number of recovery boilers, to characterize the emission spectra in the wavelength range of 300 nm to 800 nm. These tests have pointed out the potential for temperature measurements using the line intensity ratio technique based on a pair of emission lines at 404.4 nm and 766.5 nm observed in the recovery boiler combustion zone; these emission lines are due to potassium, a common constituent found in all the black liquors. Accordingly, a fiber optics based four-color system has been developed. This in-situ, nonintrusive temperature measurement technique, together with some of the more recent results, is described in this paper.

This paper discusses the concept, design and testing of a fiber optic viewing system that records flame images, wavelength spectra and flicker frequencies from inside an operating gas turbine combustor. The image, wavelength spectrum and flicker frequencies give a new type of signature of the combustor that can be related to certain engine operating conditions and malfunctions. Flame data along with other engine and sensor data taken over long periods of time and under different engine conditions can give signatures to help diagnose developing, impending or suddenly occurring problems in the engine hot gas path. The diagnosis of problems should help prevent further or additional damage to the engine or predict overhaul schedules. We would also hope to identify unusual operating conditions or modes that could be avoided in normal operation. A program has been underway with the Electric Power Research Institute to develop such a viewing system. The program has been through planned stages of conceptual design, fabrication, testing in a high pressure combustor rig and is currently being field tested on a utility gas turbine at Houston Light and Power. The flame image is taken from the combustor at the distal end of a small diameter probe with a fixed wide angle field-of-view lens. Fiber optics transfers the image outside the combustor to a nearby video camera. The video image is digitized and calculations of magnitude, position, size and orientation are made on the flame image pattern. Contour plots and differences between image patterns can also be generated for further analysis. Changes in wavelength and flicker frequencies can be recorded and the flicker signal can be cross correlated with acoustic signals from the same combustor. During the rig testing phase of the program several modifications were made to the combustor to simulate problems. These modifications included: generation of air flow nonuniformities, fuel nozzle changes, simulated liner damage, fuel, water injection and air flow changes. Some results from these tests and the field engine tests will be presented.

The objective of this research is to apply new sensing techniques, artificial intelligence, and robotics to improve the welding process through control of penetration depth and width parameters. By producing a constant depth and width of penetration through on line control, the quality and strength of welds may be improved to a new level of reliability previously unobtainable. Experimental data presented in this paper indicate that depth and width of penetration can be monitored by infrared thermography and hence controlled dynamically during the weld process. Infrared thermography is used to monitor surface temperature distributions in the vicinity of the molten metal pool. Characteristics of temperature distributions perpendicular to the direction of arc travel are analyzed and correlated with weld bead width and depth as measured by destructive examination after the weld production. The results show that there is a linear relationship between peak weld plate surface temperature and depth of penetration. Additionally, surface puddle width is found to be directly correlated to measured distance separating solidus temperatures of the molten metal as measured by infrared thermography.

Active vision for the real-time sensing of the joint geometry in front of the weld-ing torch is considered by many as being one of the best sensors for arc welding process automation as well as to improve the quality of the resulting welds. It is currently used for seam tracking and work is being executed to use it efficiently for adaptative process control [1-4]. One of the main problems in active vision systems for adaptive welding is the pre-sence of strong environmental noise from arc light and molten-particle emission. An analysis of such noise source is presented in this paper. The arc light intensity is first characterized for the three (3) main welding processes used for industrial robotic welding: GMAW in short-circuit, spray and pulsed spray transfer mode. Light noise generated by spatter is then considered. The relevance of such an analysis in terms of the choice of effective noise-reduction techniques is pointed out. The frequency bandwidth of the arc-emission signal determines the optimum modulation range for a synchronous-detection technique within the limits imposed by the characteristics of the viewing device. Similar considerations are valid concerning the travelling speed of the projected particles. Examples are given of several noise-reduction techniques, and their implementation with either flying-spot or bidimentional sensors is discussed.

A very rugged, compact (4x4x10 inches), gas purged "PINHOLE CAMERA" has been developed for viewing electron beam materials processing (e.g. melting or vaporizing metal). The video image is computer processed, providing dimensional and temperature measurements of objects within the field of view, using an IBM PC. The "pinhole camera" concept is similar to a TRW optics system for viewing into a coal combustor through a 2 mm hole. Gas is purged through the hole to repel particulates from optical surfaces. In our system light from the molten metal passes through the 2 mm hole "PINHOLE", reflects off an aluminum coated glass substrate and passes through a window into a vacuum tight container holding the camera and optics at atmospheric pressure. The mirror filters out X rays which pass through the AL layer and are absorbed in the glass mirror substrate. Since metallic coatings are usually reflective, the image quality is not severely degraded by small amounts of vapor that overcome the gas purge to reach the mirror. Coating thicknesses of up to 2 microns can be tolerated. The mirror is the only element needing occasional servicing. We used a telescope eyepiece as a convenient optical design, but with the traditional optical path reversed. The eyepiece images a scene through a small entrance aperture onto an image plane where a CCD camera is placed. Since the iris of the eyepiece is fixed and the scene intensity varies it was necessary to employ a variable neutral density filter for brightness control. Devices used for this purpose include PLZT light valve from Motorola, mechanically rotated linear polarizer sheets, and nematic liquid crystal light valves. These were placed after the mirror and entrance aperture but before the lens to operate as a voltage variable neutral density filter. The molten metal surface temp being viewed varies from 4000 to 1200 degrees Kelvin. The resultant intensity change (at 488 nm with 10 nm bandwidth) is seven orders of magnitude. This surface intensity variation is contrast reduced if the observation wavelength is a narrow band as far red as high intensity blooming will allow an observable picture. A three eyepiece camera allows an image plane where photo gray glass functions as a neutral density filter only over the high intensity portion of the image, thus reducing blooming. This system is enclosed in a water-cooled housing which can dissipate 15 watts/cm², keeping the camera below 40 degrees Celsius. Single frames of video output are acquired for feature enhancement and location by a Data Translation DT2803 image processing board housed in an IBM PC.

Among the production problemswhich the snowmobile industry is faced with, the precise alignment of the motor pulley with respect to the driving pulley is particularly stringent. The actual commercially availa-ble snowmobiles are designed for high performance. As an example, one of the model to be aligned is pointing at more than 170 km./hr. One of the consequences of these higher performances is the much more stringent tolerance necessary on the precision alignment of these two pulleys. The engine has to be moved in order that the desired position between the pulleys get within a tolerance of .4 or .75 mm according to the parameter. The belt that links the two pulleys can have a life expectancy as short as five hours if the engine alignment is not properly done. To obtain such a high precision alignment on the production line in the short period of time allowed to this operation, a complete optical alignment system has been developped to perform this operation. That system takes into account the four usual parameters the snowmobile manufacturer was dealing with. The resolution of the position detectors with the related electronics is within .08 mm and the four measured parameters are displayed in real time permitting the complete alignment of the snowmobile engine by looking at the panel meters instead of the calliper square and the trial and error technique they used to employ.

The LaserMike optical micrometer system uses a scanning helium-neon laser to perform non-contact gauging of profile dimensions. High-speed, accurate gauging of stationary or moving products is possible. By combination with microprocessor-based controllers, the LaserMike can be effectively utilized in a variety of integrated automatic systems.

For many applications with plastics the base systems are insufficient and use of reinforcing fillers is required. Glass and mica flakes are currently available for reinforcement. It has also been shown that properties of reinforced microcellular polyurethane are strongly affected by the void volume fraction. In this paper some experimental results on the applicability of far-infrared techniques to measure the state of dispersion and the concentration of mica and glass fillers in polyethylene and polyurethane are presented and discussed. It is shown that fillers can be analyzed with little perturbation from the presence of porosity, while the average orientation of the filler particles can be evaluated by varying the polarization or the angle of incidence of the far-infrared laser beam.

In the pulp and paper industry, the problem of regular and fast evaluation of wood fiber characteristics such as length and specific area is an important one. With this in view, we have been studying an acousto-optical technique based on the acoustic agglomeration of fibers in a water suspension, where a stationary ultrasonic field is created at about 150 kHz. Under the influence of radiation forces, fibers re-orient themselves parallel to the nodal planes of acoustic pressure, and regroup or agglomerate in these planes in different characteristic times. These are mesured by means of the light scattered at small angles. We have found that these times depend on the size distribution of fibers, particularly length. We present results obtained with an assortment of fiber types, under various experimental conditions which indicate eventual applications in the automatic control of pulp production.

For the pulp and paper maker, product quality and production costs are the major factors that determine profitability. Quality has to be high enough to satisfy the customer and costs low enough to maintain competitiveness. Accurate and readily available fiber length information is fast becoming one of the most important control factors to achieve these targets. Measurement of fiber length has been difficult and time consuming in the past --- appli-cation to production almost impossible due to the historical nature of the data. The Kajaani fiber length analyzer has revolutionized fiber length analysis. Even more accurate than the microscopic method and infinitely faster than mechanical classifiers, such as Bauer McNett or Clark classifiers, the Kajaani analyzer opens new horizons for the paper maker. The Kajaani method is an optical method and is based on the ability of fibers to change the direction of light polarization. With no critical sample preparation, the results are ready in a few minutes. During this time, the analyzer counts and measures over 3000 individual fibers. Results are printed out either in graphic or numerical form. Some of the typical applications of the Kajaani fiber length analyzer are to determine hardwood/softwood ratios in pulp and paper mills both in brownstock and stock preparation areas, to predict strength properties of mechanical pulp based on the fiber length information, to measure the coarseness of the fibers, to evaluate screening and refining processes and to check the quality of purchased pulp.