Owing to the rapidly decaying autocorrelation function of synthetic aperture radar (SAR) imagery, predictive compression methods have not been widely used in image coding systems designed for SAR data. Because of the uncorrelated nature of SAR data, the prediction system design problem is approached in this paper from the point of view of statistics matching and decorrelation of reconstruction errors rather than minimization of the mean square error. It is demonstrated on 6 m resolution SAR magnitude data that a simple predictive coding system utilizing an unadaptive moving-average (MA) predictor and a Gaussian optimal quantizer can result in satisfactory reconstructed imagery at compression ratios of 2:1 to 4:1. Advantages of MA predictors are their lack of stability problems and their limited memory in the presence of channel errors.
It hardly seems possible that an entire year has passed since I wrote a similar editorial for the December 1986 issue of Optical Engineering. Perhaps the years seem to go by more quickly as we get older because each succeeding year occupies a smaller percentage of our total lifetime. Just a thought.
Synthetic aperture radar (SAR) imagery can be described as consisting of bright point objects embedded in homogeneous or textured low intensity backgrounds. When predictive coding is used to compress such data, high quality reconstructions and larger compression ratios are obtained if large predictive errors are emphasized to a greater degree than in a minimum mean square error quantizer design. A large prediction error threshold quantizer is used alone and in conjunction with a small prediction error delta modulator to yield high quality reconstructions at compression ratios of 4:1 to 5:1. Smoothing of SAR data, while increasing redundancy, does not lead to better compression ratios when the predictive coding methods described in this and the companion paper in this issue are applied.
An interferometric procedure is described that shows potential for obtaining surface figure error maps of grazing incidence optics at normal incidence. Such optics are found in some laser resonator configurations and in Wolter-type x-ray optics. The procedure makes use of cylindrical wavefronts and error subtraction techniques over subapertures. The surface error maps obtained will provide critical information to opticians for the fabrication process.
A hybrid optical architecture for digital matrix-matrix multiplication is discussed. The proposed system requires less hardware than outer-product optical matrix multipliers and offers a higher processing speed than systolic array matrix multipliers. The throughput and hardware requirements of the proposed architecture are discussed with reference to the availability of various electro-optic devices. The performance of the proposed system is compared to existing systolic array optical matrix multipliers and to outer-product-based optical matrix processors. The accuracy and error rate of the proposed architecture are evaluated.
A TV system has been developed for detecting and displaying a two-dimensional hologram image that permits concomitant (real-time) hologram interferometry. The optics employed are the same as for electronic speckle pattern interferometry (ESPI), wherein a uniform reference field interferes at zero offset angle with a speckled image of an object. Rather than detecting spatial modulation to indicate interference, as in ESPI, this system modulates the phase of the reference beam by 120° between picture frames to vary speckle irradiance. Each incoming frame is compared with the two previous frames, and only pixels that vary in brightness are passed as white pixels in the video output. The use of binary-valued pixels allows inexpensive data processing at standard video rates.
We have devised and demonstrated a technique that allows the simultaneous (in space and time) generation of two independent and well-separated Nd:YAG laser lines. We use this system to study three-wave mixing in nonlinear optical crystals. Application of this laser system coupled with suitable nonlinear optical crystals permitted us to simultaneously generate frequencies extending throughout the visible (including the three primary colors) and into the near-UV.
The design of a compact laser scanner for robot vision is presented. The design is based on synchronized scanners. It is shown that the Scheimpflug condition of focusing coupled with the low divergence of a laser beam provides an ideal arrangement for large depth of view. A geometrical analysis and a prototype description are presented.
A recent development in three-dimensional imaging is the application of VISIDEP technology to surveillance and reconnaissance. The results of initial efforts to develop a single-source system using a moving platform are reported. These tests clearly demonstrate the feasibility of such a system. Although film was used to achieve the present results, electronic imaging can be used to generate near-real-time three-dimensional video images with delays of less than 3 s. With the continued development of high density video and digital imaging, improved image resolution is easily achievable. The theoretical basis, testing results, and projections for future development are presented in full detail. Potential applications and expectations are discussed briefly as a part of the conclusion drawn from this basic effort.
Video camera intensity nonlinearities can be corrected without the use of light-measuring equipment or calibrated filters. The camera to be corrected is used to adjust two independent light sources such that they illuminate a test card equally. Digitized images made using these sources separately and together are analyzed to determine parameters for use in generating correction tables. An example is presented in which a ±2% nonlinearity observed in a General Electric model 2507 solid-state camera was reduced to less than ±0.2% using a least squares fit to a third-order polynomial. The technique can be used with camera output having any function (e.g., linear, logarithmic) and should be useful whenever strict adherence to a nominal output function is desired. Simple procedures for determining that a camera is nonlinear and for correction of vignetting also are described.
The longitudinal dimension of laser speckle was measured in the near field of a uniformly illuminated frosted-glass diffuser. Theory indicates that the speckle produced by the circular spot varies according to XL2/D2, where X is the wavelength of the laser, L is the distance from the observation position to the diffuser, and D is the diameter of the illuminated spot. A single wavelength was used, allowing variation with respect to L and D to be measured. The experimental results obtained closely match the theory. More than 400 speckle lengths were measured, normalized by multiplying by D2/L2, and statistically analyzed to deter-mine a mean of 11µm and a standard deviation of 2.5µm.
A unique experimental method has been developed for measuring weld pool surface temperatures. An equation for determining temperature was developed based on Planck's blackbody spectral distribution of emissive power, the definition of spectral directional emissivity, Lambert's cosine law for diffuse emitters, and Kirchhoff's law for spectral directional reflectivity. The necessary measurements are the spectral directional emissive power and the spectral directional emissivity of the weld pool surface. Detailed pool surface temperature measurements were made of gas tungsten arc (GTA) welding of 1.5mm thick stainless steel (SS) 304 plates using an argon cover gas. Because argon plasma arc spectroscopy measurements over molten SS 304 revealed insufficient emission gaps, the spectral emissive power was measured by high speed film radiometry as the arc was shut off so as to obtain information at the instant the obscuring arc emissions disappear. Similar measurements were made of the spectral directional reflectivity (and thus the spectral directional emissivity) by reflecting a focused laser beam off the weld pool. Detailed weld pool surface temperature maps were generated.
A simple field-rated transmissometer is described for rapidly determining the normal hemispherical transmittance T(0°, 2) of leaves measured in situ in the four Landsat wavelength bands. The transmissometer requires direct solar illumination of the leaf sample. It collects the transmitted light with an integrating sphere and measures the collected light using a commercially available radiometer. The transmittances determined by the transmissometer are comparable with those measured by a laboratory spectrophotometer with an integrating sphere attachment.
A large-aperture, flat-field, four-mirror optical system for large telescopes has been designed. Its main virtue is that it uses spherical surfaces for the primary and secondary mirrors, providing remarkably good imagery. Third-order theory and the optimization process are discussed for this optical system.