Generalized bending is a one -parameter family of changes to two curvatures and related thicknesses of a previously defined optical system consisting of spherical and plane refracting surfaces. This family of changes leaves first-order properties invariant at all other surfaces in the system. Thus, third-order aberrations at the other surfaces are also unchanged. The third-order aberrations may then be expressed as functions of independent generalized bends at different locations; therefore, simultaneous correction of several aberrations is possible. Comparison of ray fan plots for real rays through an optical system shows marked differences for various degrees and locations of generalized bending. Surfaces at which a generalized bend would make significant changes to aberrations of the original lens are easily identified. This use of generalized bending would be helpful in advanced stages of a design routine.

Multiband photography has been used for several years as a tool for aerial reconnaissance. Recent experiments indicate a potential value in earth resources studies, particularly those utilizing an electro-optical image sensor. This report summarizes the test parameters and testing philosophy of several electro-optical devices. The results of camera tube testing indicate that vidicons similar to the RCA 4589 are suitable for a multiband electro-optical camera system. Results of image tube testing indicate that careful specifying of imaging performance is necessary, and results of coupled camera tube /image tube testing indicate that improvement is necessary for coupling the fiber optics. It is recommended that the vidicon be used at rather high dark current (~100 nA) and that dark current subtraction be employed for the flight program.

The atmospheric transmission and airlight in three spectral bands as a function of an angle off nadir were calculated from radiometric measurements with cameras and film for a particular solar irradiance and atmosphere; the sun zenith angle was 49 °, the airpaths were from 15,000 ft m.s.l. to the surface at 1,900 ft on a clear day in Tucson, Arizona, and the azimuth was into the sun. The three spectral bands had peak transmissions at 430 nm (blue), 530 nm (green), and 800 nm (infrared). The statistics derived from numerous measurements show that a standard deviation of 2% can be attained in the relative radiances read out of multiband photography obtained with calibrated cameras and processed with carefully controlled sensitometry. In the blue and green bands, the atmospheric effect on contrast as a function of an angle off nadir became statistically noticeable at about 35° and 50 °, respectively. The standard deviations of the relative radiances measured in the blue and green bands were 3% and 2 %, respectively. The effect in the infrared band probably became significant at even larger angles; greater inaccuracy in the infrared band data precludes a more definitive statement. Also presented is a solution for an optimum ratio of playback lamp luminances for false color recombinations in two bands.

The papers presented in this document were given at a technical session in conjunc- tion with the dedication of the Optical Sciences Center on January 22, 1970. Session chairman was the Honorable Harry Davis, Deputy Under Secretary of the Air Force. Mr. Davis was an appropriate choice for the chairman, as he has been active in stimulating much of the current thought and effort in this area. The concept of synthesizing a large optical aperture from a collection of smaller, related apertures has appealed to optical physicists for a long time. The difficulties of manufacturing a large telescope or other optical system using a single massive blank for the primary objective are in themselves extreme. In the case of a large astronomical telescope, even though the primary image- forming element may be monolithic, it cannot be rigid to the tolerances required. As a result, most telescopes are active in the sense that the primary mirror is supported on a mount that adjusts to differing mechanical or thermal loads. The development of large space telescopes will lead ultimately to actively controlled mirrors in the adverse environment of any useful orbit. Under such conditions, it is possible to conceive of a synthetic aperture telescope consisting of a set of reflectors, mounted on an actively controlled mechanical system, that can respond to environmental changes. In this way, proper phasing of the optical elements can be achieved for significant exposure times. In fact, some of the phasing can be done optically in the region of the image plane, which is typically much smaller than the apertures themselves. In addition to the above passive systems, active systems using laser illumination with either spatial or temporal integration can be devised and are close to realization. In effect, these would be optical radar systems, with all of the advantages of coherent detection, along with the fineness of resolution available at optical wavelengths. The state of the art in the field has not reached the engineering stage. Indeed, at present we cannot speak even of a developmental stage. However, the time is fast approaching when synthetic aperture optical systems are going to be realized in practice. Atmospheric limitations imply that the first applications will occur in space, but certainly not before the techniques are well explored on the ground, perhaps in the far infrared region where the resolution limitation of the atmosphere is not so restrictive as in the visible range. Before this happens, technological developments in several areas are required. Needed are active control and phasing of the optics, post-processing of the sampled pictures, and an over-all concept for system design. Any interested person can likely add more areas of technology that would be stretched to attain the desired goal. This meeting was organized to permit discussion of the state of the art in this area and to encourage further work in the field. Appended to this set of papers is a bibliography on the subject of optical synthetic apertures, which was prepared by Mr. R. Brian Hooker, currently a graduate student at the Optical Sciences Center. The list should be useful as a starting point for anyone who wishes to delve deeper into the subject. -R. R. Shannon

The 13.7 -m Czerny -Turner spectrograph at the McMath Solar Telescope is evaluated with regard to polarization and image -forming properties. The ratio of the transmissivity of the spectrograph for incident light linear polarized parallel to the slit, to the transmissivity for light polarized perpendicular to the slit is determined as a function of wavelength. Values of this ratio range from 0.3 to 20. The diffraction grating is responsible. It is shown that spectrograph transmission as a function of polarization may cause large photometric and radiometric errors. This fact is not unique to this particular instrument. Two mechanisms appear to contribute to the polarizance of the grating. These are (1) the Rayleigh or Wood's anomalies, and (2) a vector wave interaction introduced because the groove dimension is nearly the same as the wavelength. Peak polarizance Xp occurs at the Wood's anomalies, and for X > groove depth it is found that the wavelength of peak polarizance Xp is given by λp = (0.7) d cos(B) where d is the ruling separation and O the grating angle. Photoelectric scans of the solar spectrum were made across several of the Wood's anomalies to obtain their profiles.

It is possible to use a target of two coherent point sources, with a separation much greater than the resolution limit of a diffraction -limited lens, to determine the best focus of a photographic system. The interference pattern formed is a function of the focus error. Measurement of its characteristics (fringe period, fringe number, etc.) for different focus errors should indicate the best focus. Since the measurement is made at a long distance from focus, there is no ambiguity as to the direction of the best focus position relative to the film. An experiment, in which known focus errors were introduced on both sides of a visual best focus for a two -coherent -point- sources target, was designed and carried out for a 35 -mm camera system to determine the usefulness of such a method of determining best focus. A comparison test was made against a standard resolution vs focus error method. The experiment showed an unambiguous best focus within an error of 10 pm. The focus position was unambiguous because the final relationship studied was linear, and only one value was obtained. In general the new method appears to be faster, simpler, cheaper, and more accurate than the standard resolution method, and it requires no special instrumentation on the camera being tested.

Multiband photography has been employed for several years as a tool for aerial reconnaissance. Recent experiments indicate a potential value in earth resources studies, particularly those utilizing an electro-optical image sensor. This report lists the imaging properties of an electro-optical multiband camera that uses one combined image tube /camera tube sensor to achieve high- quality imagery with maximum registration between separation images. A study was made to determine which camera tube would be the best choice for the earth resources program. The results of the study indicate the best choice is a standard vidicon with fiber optic faceplate coupled to an image intensifier with an extended red photocathode. This decision was based on lifetime, image quality, ruggedization, and cost criteria.

The results of the first complete tests of the Optical Sciences Center digital image processing facility are reported. The instrumentation for digitizing film transparencies is operating well enough to provide suitable in- puts to the computer program IMPR(C, which processes (restores) the degraded image. Some manual correction of the digital data is required (mainly for missing digits) before IMPROC can be used. Although IMPROC will be expanded and modified during future research efforts, the debugging of its present form is essentially completed. First attempts at restoration of a defocused image show contrast enhancement and edge sharpening. Current techniques for displaying the processed images, however, are slow and have a limited gray scale. The best technique to date, which uses the computer's plotter to produce nine distinguishable gray levels, requires 1 hour to produce a 100 x100 matrix of points. A CRT film display device, driven by a digital computer, is being developed to solve this problem. Current and future research efforts are described.

New materials of low thermal expansion are finding wide application. The expansion coefficient (a) is a function of temperature, and this function must be known for each material before its applicability can be assessed. A novel method for determining a, which is at once precise and easily implemented, has been devised. It is based on the dependence of mode frequencies in a Fabry-Perot interferometer on the mirror separation. The expansion sample is formed into an interferometer spacer with ends polished flat and parallel. Spherical mirrors are optically contacted to the ends, forming a confocal interferometer. The assembly is maintained at controlled temperatures in an environmental chamber. The two lowest -order transverse modes are probed by variable -frequency sidebands derived from a 633 -nm He- Ne laser by amplitude modulation. A change in sample temperature AT causes a change in interferometer length AL, which shifts the resonance frequencies by Av. Then a = (1 /AT) (AL /L) _ - (1 /AT)(iv /v). Thus, a can be measured with precision limited ultimately by the stability of the source laser, in practice 1:109 with presently available commercial lasers. For a sample of Owens -Illinois Cer -Vit, a has been measured at 10 temperatures in the range 3.0 to 32.4 °C, with a mean error of 2 x10-9 and a maximum error of 3 x10 -9. For a sample of Corning ULE silica, a has been measured at six temperatures in the same range, with a mean error of <1 x10 -9 and a maximum error of <1.3 x10 -9.

Simple inverse filtering of each spectral component of a color picture can lead to serious misregistration of the restorations. This is a consequence of the nonlinearity of the photographic process, which distorts edges of objects in an unknown manner. To force three restorations that do register, we tack onto the usual term describing mean -square resemblance to the object scene the constraint of mutual resemblance among the three restorations themselves. This term is highly sensitive to phase errors (misregistrations) among its three members. The optimum three filters complying with this minimization problem are found. A simple case is examined, in order to judge the manner by which the restoring filters force a mean-square resemblance to the original object scene and, simultaneously, a mutual registration among the three restorations.