The optical requirements of advanced systems have encouraged lens suppliers to devise new methods of evaluating optical parameters that will insure heretofore unattained system performance. Progress has been made in this direction by combining interferometric and data processing techniques to determine rapidly a satisfactory figure of merit for high performance lenses. There are two major advances that have made possible a new approach to the design and evaluation of optical systems. They are the high speed computer and the laser.
My subject this morning is lens testing. It is a subject as old as man's ability to make lenses and it seems a certainty that the first argument over lens quality occurred at the sale of the first lens. As the need for higher quality lenses grew, so grew the gap between the lens manufacturer and the lens user. The lens user primarily concerned that the lens perform adequately in his system and the lens manufacturer trying to provide this lens without really understanding the relationship between that which he could control and the users performance criteria.
The test I am about to describe is a simple test. In a nutshell, it consists of the following: We take a picture of a test object containing typical camera picture subject matter. This subject matter is arranged in a test pattern, and the pattern is repeated over the entire test object. We then make a black-and-white enlargement of the picture and visually compare the picture sharpness over the format with a graded standard. The standard contains a series of pictures of the same pattern that is repeated over the test object, and the range of sharpness of the series extends from excellent to poor. In developing the test, our goal was to satisfy the following basic objectives.
The concept of image contrast as determined by the modulation transfer function is now a well established principle in the field of image evaluation. To date little use has been made of this principle other than to predict resolution of a lens from design data. As with any new principle, it takes some time to determine from experience the problems that may be treated, and the method of applying the technique. This paper reports the application of contrast transfer analysis to solve a very puzzling problem in image evaluation.
More than 60 years ago, Hartmann' proposed a test procedure for the optical evaluation of large aperture astronomical objectives which remains to this day the principal tool by which astronomers accept or reject the claims of the manufacturer. Unfortunately, although relatively simple in application, astronomers have not always availed themselves of the Hartmann test, many indeed are ignorant of it, with the result that many have fallen heir to instruments of dubious or unknown quality. Noticeable exceptions to this situation have occurred in those cases where the astronomer has called for final shop testing based on Hartmann's method under controlled environmental conditions. For this and other reasons, American Optical has increasingly relied on the Hartmann test in the evaluation of large optical elements. Since 1961, American Optical has certified critical components on the basis of Hartmann tests . In fact, the standards of quality now expected of the optical industry require extensive Hartmann testing in the final stages of polishing. This allows the optician to bring the optical surface to a degree of perfection that exceeds his ability to evaluate by direct observation with the knife. Of course in situ tests of the objective remain the ultimate measure of the performance of an astronomical objective.
An instrument developed for the U.S Air Force, the Variable Contrast Illuminator, permits continuous variation of the contrast and color of a bar target source. The target consists of transparent areas ruled through a reflecting substrate. The colors of the transparent areas, the reflecting areas, or an area superposed over both clear and reflecting areas may be independently varied. The color variations are formed by selective attenuation and synthesis of red, green, and blue light paths. The white light in tensities of transparent and reflecting areas may be independently varied to produce contrast ratios from over 1000:1 to 1:30, i.e. from bright areas on a dark surround to dark areas on a bright surround. Recently a company investigation has shown the feasibility of providing a ruled target with three bar patterns(√2 progression) extending from 3.2 lines per mm to 228 lines per mm.
A central concept in the calibration of instruments is that of the mathematical error model. An adequately derived, physically related, error model provides for the correction in the data reduction process for measurable systematic errors. Error interactions and the nature of their propagation into the data as a function of instrument parameters are provided. Such a physically related mathematical error model is presented. While specifically intended for two axis Cinetheodolites, it is quite applicable to other types of mounts, such as radars. Techniques are indicated for its extension to instruments with more than two axes. Conceptually, the error model divides into three parts: a nodal point-image vector which may be perturbed to correct for certain errors; three or more matrices which represent the basic rotations and certain errors; and point by point corrections to the observed angular values used in the matrices. Vectors and matrices are used freely in the model. A brief tutorial presentation is provided on the mathematical methods employed. The model is useful in calibration of instruments, error analysis, acceptance testing, error correction, and provides a basis for a new philosophy of instrument design which potentially could result in cost savings.
Image evaluation and image mensuration are usually assumed to be problems to which a linear system's analysis is applicable. Hence, the aids used in viewing and studying the image, e.g. , viewers, tube magnifiers, microscopes, microdensitometers, etc. , are assumed to use incoherent illumination of the object resulting in a system which is linear in intensity. Departures from this ideal situation occur at moderate line frequencies in the object and the effects of illuminating the object with partially coherent light are of interest. Some examples of these effects are discussed with particular reference to evaluation and measurement of object information.
The need for a lens tester arose several years ago in connection with many projects associated with the Russian-English Translator, being built jointly by the United States Air Force and IBM Research. In order to determine if the lenses in the various components of the system were performing up to their specifications, and to resolve doubts as to their stability under high accelerations, a lens tester was designed and built according to the following specifications:
The application of laser technology has beeh extended to optical shop testing by incorporating a continuous wave, helium-neon gas laser in a package that houses a modified Twyman-Green interferometer. This modification provides for optical testing over large path differences with an auxiliary set of lenses used in the long path and a small reference flat used in the short path of the interferometer. With this technique, f/0.7 spherical mirrors have been tested at the center of curvature to an accuracy of 1/10 wavelength at the surface, and various other optical systems have been tested in both double pass and single pass. Two of the advantages of this testing method are (1) the capability of testing spherical concave surfaces without physically contacting the surface and (2) the ability to use small reference surfaces for large optical components or systems. The device, known as a laser unequal path interferometer, can be used with a set of null lenses to qualify aspheric surfaces. The unit is portable and capable of testing in any orientation under various environmental conditions. Several applications of this device are presented to illustrate its versatility.
An approach is explained for determining the three optical per-formance functions: The edge spread function, ES, the line spread function, LS, and the transfer function pair, MTF and PTF, from an edge trace, ET, obtained by photometric measurements at 2n + 1 points at equal intervals through the range, L, of the ET. The explanation is theoretical, but with as much consideration of the needs of the practical engineer as possible. Smoothing and normalization procedures are an integral part of this method, based on a rigid control by the method of least squares. The mean-square errors, MSE, of all individual results will be obtained. The MSE of a critical magnitude instead of Shannon's sampling theorem is used to determine the necessary and justifiable number of terms of the series applied in this case, the only assumption being the validity of the error propagation law. It will thus be possible to check the reliability or scope of applicability of the sampling theorem within this context, which differs from and goes beyond known methods.
An analysis of the performance of a cathode-ray-tube flying-spot-scanner film reading system requires a treatment of the gray level resolution of the system. This paper discusses three types of gray level encoding schemes for a digital film reading system; equal transmission, equal density and equal reliability levels. The maximum number of detectable levels, or the minimum gray level change detectable, to a given reliability is presented along with a trade-off analysis and a discussion of the utility of each method. Calibration and measurement of a hardware system by analysis of maximum signal available is discussed.
The construction, use and results obtainable from a relatively inexpensive ratio-type spectro-radiometer are described. The large range of lamp types, sizes and operating modes capable of being measured are discussed. The measurement of absolute spectral irradiance, Hλ is the primary intent of the instrument. However, values for spectral radiance, Nλ and spectral ra-diant emittance, Wx, and spectral radiant power, Pλ, can also be obtained providing source solid angle subtense is calculable. The methods of data taking, reduction and display are discussed. The overall accuracy of the device is presently limited to approximately ±5% (Hλ), primarily because of the standards currently available. While the optics in the device permit use over the 250nm to 2.5 micron spectral region, the needs for present limitation to the 300nm to 1000nm region and possible additions and uses for the instrument are discussed.
Why all this sophistication in specifications? First, to communicate one's wants in sufficient detail to permit another to unambiguously execute hardware to fulfill these needs. Secondarily, specifications serve as a quasi-legal means to determine when a contract between parties has been fulfilled.