A review of adaptive image transform coding techniques is given. Adaptive techniques are superior to nonadaptive procedures; however, their implementation requires a significant degree of added complexity. A major aspect of adaptive transform coding is the understanding and utilization of the local statistical structure of the image. Rate adaptive communication and buffer feedback are also important aspects of adaptive coding procedures. The discussion emphasizes the image quality aspect of the coding process.
Whether an image is produced on a CRT display, a printer, or a photographic plate, that image can be described as a two-dimensional modulation of luminance and color across space. A psychophysical approach to image quality is one which considers the detectability and discriminability of these images in terms of the spatial, temporal, and color-sensitive mechanisms of perception. The aim is to develop a technology-independent method for characterizing and predicting image quality.
The concepts of statistical decision theory were applied to the photographic tone reproduction system to develop an objective measurement of tone reproduction accuracy called the "expected loss." Reproductions were made of four scenes using systems with a wide range of tone reproduction characteristics. The expected loss for each system/scene combination was calculated, and the perceived tone reproduction accuracy was determined using the psychophysical scaling method of magnitude estimation. A statistically significant relationship was found between the subjectively determined accuracy and the system expected loss, but the degree of correlation was low enough to indicate that additional factors may need to be included.
Based on the assumption of the vision's frequency channel model in perception of edges, a numerical procedure is suggested to obtain an effective measure of image raggedness. This measure can be obtained from a bit map of bilevel image with consequent Fourier analysis of the effective edge profile function. The advantage of the procedure is that no "standard" or control image is used for comparison and only single variable FFT is performed.
To extend the life of an original document or for convenience of handling the image, it is often desirable to make reduction copies of the document. However, current microimaging systems have not been optimized for high-quality color reproduction. This paper describes the image quality and optics considerations for high-resolution color imaging. As a result of this study, two new 105 mm fiche-size color film systems are proposed. System I is a positive-positive system, using KODAK EKTACHROME Professional Film 6118, that can be used for image reductions up to six times printed on KODAK EKTACHROME Duplicating Film 6121. The second procedure, System II, is negative-positive and offers reductions up to 10 times with excellent image quality. System II uses EASTMAN Color Intermediate II Film 5243 as a camera negative and is printed on EASTMAN Color Print Film 5384 to produce a positive transparency. Reductions higher than 10 times are possible but with a corresponding loss in image quality, which is primarily due to optics limitations. Another major concern is the maintenance of image quality from the center to the corners of the original.
Criteria are studied and derived that relate lens MTF data to image quality of photographic prints. These criteria are based on the results of earlier experiments correlating MTF-data of photographic prints with observed judgements. Three criteria are considered, they are based on an integral over the weighted lens MTF, the lens MTF at a critical frequency, and critical magnification. The preliminary results show the limitation set by the photographic reproduction process, resulting in only small differences in image quality with standard lenses on axis. Improvements, however, are possible from lens performance off axis, by reducing the depth-of-focus range, by reducing effects like camera vibration, and especially from the photographic process.
The measurement of Minimum Resolvable Temperature Difference (MRTD) has been shown in general terms to be a good criterion for assessing the overall performance of a thermal imaging camera since it takes into account both the effects of MTF and that of noise.
A computer model has been developed to predict the reproduced reflectance distribution of an electrophotographic image, given a specified document reflectance distribution and a set of parameters which characterize the system. The model is general enough to handle not only the classic large solid area reproduction characteristics, but also spatially varying distributions such as halftoned images and text. The model is run with various input targets to determine system characteristics such as tone-reproduction curves, reproduced line widths, and system MTF.
One important application of electronic digitizers is the scanning of line art (engineering drawings, parts drawings, etc.). Typically, line art drawings are digitized with the ultimate goal of obtaining binarized data. Resolution requirements for scanning of line art (engineering drawings, part drawings, etc.) are sometimes difficult to calculate, and are usually determined by subjective evaluation or trial and error. Too much resolution is inefficient and consumes unnecessary resources, while too little resolution may be incapable of resolving detail adequately. One method often used is to set the scanning pixel size to twice the minimum line width found in the drawing. It can be shown that this method will yield good results for a fixed threshold binarization, assuming even illumination, but equivalent results can be obtained by scanning at lower resolution (larger pixel size) and using an adaptive threshold binarization which operates on local areas.
The visual display presents a major source of critical information to the pilot in a flight training simulator. At the same time, due to underspecification or underdesign of the visual display system, the information presented to the pilot is grossly incomplete. If the pilot saw in the real aircraft as well as he sees in the simulator, he would not be allowed to fly (or even drive a car): he would be legally blind! These problems in the visual display are the result of misapplication or misunderstanding of the pilots visual perceptions and the limitations of resolution and scene detail of the visual display. This paper presents a summary of the principles of limiting resolution and the definition of scene detail, as well as a discussion of other visual parameters; and the importance of these factors in obtaining satisfactory standards in initial and continuation training in a flight simulator.
Image contrast enhancement sometimes complicates image understanding. A scene that consists of slightly dissimilar target and background emittances may not be readily identifiable without image enhancement. Even if the emittance differential can be sharply contrasted, those image surface patterns that convey subsurface thermal information may not be visible because of the wide dynamic range that must be accommodated by the thermal imaging system. This paper describes how emittance complicates the interpretation of thermal images. High and low emittance values affect the logic required for understanding thermal scenes. Thermal scenes containing emittance differentials are easier to interpret if there is a large contrast between the object and the background.
In recent years the quality of infra-red lenses has improved considerably. Deficiencies due to manufacturing errors and lack of readily available test equipment have been reduced, and lenses are now being made which are consistently close to diffraction limit. Now that the optics are not the limiting factor in a thermal imaging (TI) system, it is possible to consider what relaxations may be permitted to reduce optical costs without compromising TI system performance. As a first step in this consideration, this paper describes initially experiments concerning the subjective effects of using TI systems with degraded optics. We then describe experiments both in the laboratory and also in the field to quantify these subjective effects, and methods of MTF measurement on complete thermal imagers are then given. Finally some results and conclusions from these experiments are presented.
A new electronic speckle pattern interferometry technique (ESPI) has been developed and is used for non-contact examination of the head-tape interface in a revolving head configuration of a video tape recorder (VTR). Dynamic tape motion is measured by illuminating the surface of the tape with a pulsed Nd-YAG laser (wavelength of 532 nm). A television camera (NTSC standard), combined with the pulsed laser, makes it possible to measure the transient motion of the tape. The pulse repetition rate of the laser is 30 pps, such that the speckle pattern of the tape's surface can be observed and recorded at intervals of 1/30 of a second and stored in frame memories. Isothetic fringes for in-plane displacement are obtained by computing the difference between the initial and succesive images. These isothetic fringes are processed automatically to obtain the values of the displacement by using digital techniques. By this method, the transient behavior of the tape can be analyzed. An example of the results is presented to show the dynamic motion of a tape traced by a revolving head at a speed of 5.77 m/s in a VTR.
Large numbers of borescopes (endoscopes) are used by the Royal Air Force for inspection inside engines, wing panels, etc. There is a need to establish specifications for borescopes, and test procedures to ensure that they meet the specifications both initially and after use or maintenance. A representative sample of 59 units was studied, by Sira Limited, to confirm the suitability of certain test methods and to facilitate writing a test specification. The trends of the results are presented, together with a description of the proposed test equipment under the general headings of resolution, glare, photometry and sundries. It is expected that equipment will be installed within the year, and that borescope specifications of the new type will gradually be introduced.
A versatile test device for slide projectors is described. At first the concept is discussed which includes the objective measurement of several quality-determining features. Then the realisation of the test device is described in some detail. Finally results of measurements are given and consequences for testing other projecting devices are discussed.
The Landsat Thematic Mapper (TM) represents significant improvements in spatial, spectral, and radiometric resolution over the older Multispectral Scanner System (MSS). For the last three years, NASA has conducted the Landsat Image Data Quality Analysis (LIDQA) program to quantify the performance of the TM on the Landsat-4 and 5 spacecraft. As part of this program, we have performed analysis of the TM imagery to extract the overall system modulation transfer function (MTF). In this paper, we describe the use of the San Mateo Bridge in San Francisco as a target for calculation of the line spread function and MTF. The analysis of two TM scenes, one from 12/31/82 and one from 8/12/83, yielded effective-instantaneous-field-of-views (EIFOVs) of 40.8 meters and 48.6 meters, respectively. These values are compared with the 33.8 meter EIFOV predicted by component modelling of the TM sensor and the differences discussed.
Airborne measurements of atmospheric-path transmission and atmospheric-path (up-welled) radiance in the 8 to 14 μm band were obtained by applying a multiple-altitude and a dual view angle calibration technique to thermal infrared line scanner data. Spectrally corrected LOWTRAN was used to generate path transmission and up-welled radiance values corresponding to the empirical measurements. Using LOWTRAN and the multiple-altitude method, calibration of the thermograms to account for atmospheric effects yields computed surface temperatures within 0.7°C of concurrent kinetic temperature readings. The angular calibration method results in similar computed surface temperature errors for 1000 ft altitude data and increasing by 1.2°C per 1000 ft up to a 6000 ft altitude. This paper contains the results of a comparative analysis of these approaches for atmospheric calibration.
More spatial resolution it not the necessarily better and spatial resolution is not the only measure of resolution that matters for crop monitoring. Temporal resolution is important if the target can change quickly as crops can, and perceptual resolution of the color axes of the display is important. Satellite data can be displayed as color images for operational crop monitoring. The atmospheric effects on radiation detected by the satellite, and rapid changes in the target crop make quantitative analysis impractical. This leaves qualtitative analysis of information displayed as color patterns within the images. A system is presented for displaying changes in intensity, hue and saturation (IHS) which is optimally suited for visual (qualitative) analysis. The color images use an IHS system to represent changes in crop health. Crops are monitored by comparing images and their patterns throughout the growing season. The system has been used successfully for monitoring the Sahel region of Africa, and is being tested for Southeast Asia.
By analogy with optics, the spatial resolution of image sensors is generally characterized by the Modulation Transfer Function (MTF). This notion assumes the system being a linear filter, which is not the case in integrated image sensors, since they have a discrete photoelement structure. These sensors must in fact be considered as integral samplers. Their response to any irradiance distribution can thus be computed, knowing the pitch of photoelements and using a characteristic function. This function is more or less similar to the MTF. Once exact theoretical foundations have been defined, a computer simulation enables the various MTF measuring methods to be compared this makes it possible to rule out er-rors inherent to experiments. The most accurate and reliable method appears to be the knife edge method, applied with a relative displacement of the sensor and of the image. This avoids the occurence of aliasing phenomenon. Experimentation of this method for measurement of the CCD sensors characteristic function, which we call MTF as agreed, is described. This method also makes it possible to evaluate the transfer inefficiency of shift registers.
Quite frequently, an otherwise rational individual will produce an optical system performance specification that is virtually unachievable due to either diffraction effects, manufacturing limitations or both. This paper presents several examples of perfect and near perfect optical systems, demonstrating the effect that small residual wavefront errors have on system image quality. Also examined is the effect of diffraction phenomena when an aperture obscuration of significant dimensions is present. Image quality has been evaluated in two ways. First, the Line Spread Function is examined in order to demonstrate the effect in terms of basic image sharpness. Second, the Modulation Transfer Function has been generated for each case in order to show the effect in terms of image contrast and overall system resolution. Results indicate that diffraction effects, particularly when an aperture obscuration is present, and small residual wavefront errors will have a significant effect on final system performance. In order to realistically evaluate the impact of this image quality degradation, one must carefully consider the ultimate function of the system.
Some of the techniques which can be used for measuring the MTF of video imaging systems are described, including a recently developed scanning microphotometer based on a linear CCD array. The relative merits of the different techniques are discussed.
An approach for evaluation of the Radiometric Quality of Landsat 4/5 Thematic Mapper Band 6 data is presented. The approach involves comparison of measured surface temperatures with surface temperatures predicted from observed satellite radiances propagated to the ground using the LOWTRAN 5A model. The atmospheric propagation data and surface temperatures are also compared with atmospheric propagation measurements and surface temperatures measured during an aircraft undertlight of the satellite.