Showing papers on "Aerial image published in 1984"

An optimized cubic interpolator for image resampling

Abstract: It is noted that the cubic resampling function is only one member of a family of functions, defined by the single parameter of the slope of the cubic function at its first zero crossing, whose other members are in some cases superior to the standard cubic. This superiority is especially noteworthy with respect to the extent of gray level overshoot induced by the resampling process at high contrast edges. It is shown that there is an optimum member of this 'parametric cubic convolution' family which minimizes the mean-squared radiometric error arising from interpolation. This interpolator requires no additional computation time over the conventional cubic one. These conclusions are supported and illustrated by resampling simulations with both a high resolution digitized aerial image and a Landsat Multispectral Scanner image.

Evidence accumulation for spatial reasoning in aerial image understanding (expert system, computer vision)

Abstract: Some image understanding tasks, such as locating culture structures in aerial photographs, are difficult to do automatically. Three major sources of difficulty are identified. A system based on an evidence accumulation strategy is proposed. This system is applied to the recognition of suburban housing developments in high resolution aerial photographs. First, the underlying image processing steps are unreliable. Errors are introduced in the segmentation process, and reprocessing of the image is often necessary. Second, the descriptions of objects are complex. Houses can have different shapes as well as different numbers of neighboring houses. Third, extensive use of spatial relations between objects is often necessary. The control of their usage is complex. The spatial relations between houses and roads can be used to establish relations between houses and roads which have been located in the image already. They can be used to indicate other houses and roads not yet discovered in the image. A paradigm based on test-hypothesize-act is used to represent the spatial relations between objects. All the hypotheses created are recorded in a database. Consistent hypotheses are combined and their composite hypotheses are constructed. A control strategy based on select-and-evaluate is used. A "context rich" composite hypothesis is selected to be verified by the system and those rules whose evaluation "depends" on the verification result are now evaluated. This evaluation process causes changes to the database of the system, and the process is repeated. A general conclusion from this work is that by accumulating relevant evidence, one can realize a flexible process which performs both top-down(goal-oriented) picture processing and bottom-up verification of consistent spatial relations among objects.

Assessment Of Image Quality Of Photographic Lenses By Edge Image Analysis

Abstract: ASSESSMENT OF IMAGE QUALITY OF PHOTOGRAPHIC LENSES BY EDGE IMAGE ANALYSISHarald ThomasErnst Leitz Wetzlar GmbH, Postfach 2020, D -6330 WetzlarPreliminaryIn this paper assessment of image quality is restricted to normal photography. Thatmeans that we should assess the capability of the lens to give a reproduction of thestructures, the colours, and the intensity levels of any object in our surroundings, likelandscapes, buildings, technical equipment, or people, animals, and so on.It means furthermore, that we do not look at an aerial image as in a telescope, butthat always a photochemical process is included, leading to a projected slide or to aprinted image. The final image is inspected subjectively by the human eye.If we test a lens by measurement of data of an aerial image, the evaluation of the datashould take the film and the eye into account at least in a simplified way.A second restriction is restriction to black and white photography. Of course it ispossible and desirable to incorporate the assessment of chromatic aberrations. But in thispaper we will regard intensities only to keep problems simple.Problems of the OTFNobody will doubt, that the OTF is describing the imaging properties of a lens correctly.But on the other hand up to now no method of interpretation has been developed, which maybe agreed by all experts world wide.One main problem of the OTF is, that it is a complex function. We may be accustomed tocomplex functions and may understand them mathematically. But it is difficult to visualisetheir practical meaning. Therefore usually the PTF ist put aside, and only the MTF isconsidered. That means that asymmetric line spread functions are simply taken as beingsymmetric, and even severe coma is neglected. Even in the case of a symmetric line spreadfunction there may exist a spatial frequency range for which the real term of the OTF isnegative. This means a decrease of image quality. But the MTF is positive, thus giving awrong result, if the phase shift of 180° is not taken into account.A further problem may be seen in the linear spatial frequency axis. In electronics andacoustics always a logarithmic frequency axis is used for transfer functions.A grating with linear frequency steps (Fig. 1) obviously shows, that we do notperceive these steps as equidistant, but that there should probably be a constant factorfor each step. Probably a logarithmic frequency axis would be better for OTF, too.

Image understanding systems

Abstract: This paper provides a brief summary of the goals, methods, and achievements of the Image Understanding field. Some details of work at USC are provided. The tasks described include automated mapping from aerial image, image to map correspondence and 3-D vision for robotics applications. This paper provides only a brief summary, more details can be found in other reports and publications.

Two-Dimensional Modeling Of Contrast-Enhanced Lithography

Abstract: The aerial image produced by projection mask aligners can be readily visualized using high resolution computer graphics. This paper describes a computer model that calculates the aerial image using a mask pattern and the optical system characteristics as input. The program converts the digital result into a grey scale image. This image is an accu-rate representation of the image the photoresist actually "sees." The model is applied to contrast-enhanced lithography (CEL).1120 By combining the aerial image model with the known bleaching behavior of CEL materials it is possible to calculate the image intensity transmitted by the bleachable layer as a function of time. This result is presented in the form a computer-generated movie, which makes apparent the high contrast of the transmitted image. A second application of the aerial image model is to two-dimensional resist pattern modeling. Although not as sophisticated as SAMPLE4 this model is capable of modeling com-plete structures, such as a dynamic RAM cell. The output of the model is a three-dimensional surface which is displayed using a computer-generated, shaded surface. Linewidth variation with exposure is easily explored with this model. It is a best case model in that it assumes ideal optics and resist development conditions. Resist thickness is calculated using an experimentally determined thickness transfer function. These assumptions are necessary in order to minimize the time necessary for performing the calculations. The model calculates a pattern on a 512 X 512 point array from an image in 1-2 min. on a VAX-780. Since ideal conditions are assumed, the utility of the model is primarily in its ability to predict when a structure is beyond the limits of a given optical system. Applications of the model to CEL will be presented.