This paper reviews briefly some of the previous work done with the EIKONIX Diffusion Model Transformation (DMT) as an image enhancement algorithm for electro-optical imagery. It also compares the technique to some other common forms of image enhancement, such as: - linear high pass filtering - homomorphic high pass filtering - histogram remapping The DMT produced results superior to the other enhancement methods. A combination of the DMT and histogram equalization proved still better.
A very high resolution and high contrast laser addressed liquid crystal color display has been developed for use in the acquisition and exploitation of high density alpha-numeric and graphic data. The display system consists of two smectic liquid crystal light valves (SLCLV) which are thermally addressed by a single laser selectively. Each light valve is capable of generating 2048x2048 addressable pixels within a 1"x1" data format. There are 8x106 pixels available for data presentation. The laser writing system consists of the laser, a modulator which enables the formation of discrete data elements on the SLCLV, a two axis galvanometer deflection unit and beam splitting optics designed to multiplex the writing laser on each SLCLV. A closed-loop galvo deflection system can position the focused laser beam to an accuracy of 0.1% across the screen. Laser focus lens is an f/8.0 telecentric design to produce a 1"x1" data format on the SLCLV with spot size less than .0004". Color images from each light valve are superimposed onto the screen by a wide angle f/8.0 projection lens with 48X magnification relays the combined color image to a 48"x48" screen with less than 0.1% of distortion. The possible contrast of the display is up to 60 to 1. The overall dimension of the rear projected color display is 60"x52"x31".
The objective of intensity mappings, functions from recorded intensity to displayed intensity, is to provide contrast in the display of features of interest in images where these features appear as regions of increased or decreased intensity. Three component objectives of such mappings can be identified: 1) device linearization, 2) emphasis of contrast in commonly occurring and accurately measured intensities, and 3) user-controlled mapping based on the importance of image features. Mappings of type 1, which immediately precede the display device, are designed to modify the device so that equal changes in the driving intensity value are equally perceivable. The importance of this mapping will be demonstrated, and the argument that by itself this mapping must improve displayed image quality will be refuted. A method of measuring the JND curve and of calculating this mapping from it will be discussed. A method for automatic determination of a mapping of type 2 consisting of a nonstationary modification of Cormack's mean pixel uncertainty minimization method [Phys. Med. Biol. 25(2)] will be presented. Methods for user control in the determination of mappings of type 3 will be briefly discussed.
Visual responses to briefly presented sequential displays were studied in two experiments exploring stimulus contrast variation influences on target acquisition. When two stimuli are presented in close temporal and spatial proximity, the target stimulus (TS) information is changed by the flanking or surrounding mask stimulus (MS). The observers' visual response characteristics or TS information bandwidth varies in a non-monotonic U-shaped fashion as the temporal difference between TS and MS onset (stimulus onset asynchrony or SOA) increases (Type B masking). Experiment 1 varied only TS con-trast, Experiment 2 varied only MS contrast. Observer responses to TS shape and location in the visual field were measured as a function of SOA and stimulus contrast. Data were evaluated by estimating the maximum of the regression functions describing the empirical masking functions. For Experiment 1, as TS contrast decreased, the SOA of maximum shape masking also decreased. SOA for maximum location masking was lower than that for shape responses. Experiment 2 revealed an attenuation of the Type B shape masking function as MS contrast was reduced. Location masking was minimal. In conclusion, variation of user visual response characteristics may prove useful in optimizing information bandwidth during human/computer display dialogues.
Three dimensional projected film displays without glasses are now possible as a result of recent developments in autostereoscopy of the parallax panoramagram type. These dis-plays have the features of: full color; excellent resolution; horizontal parallax; con-ventional illumination sources; recording with conventional filming techniques; and in some cases, full motion. The applications include: three dimensional aerial and satellite reconnaisance pictures; 3D computerized axial tomography displays for medicine, seismology and industrial purposes; scientific stereoscopic displays; and educational, advertizing and entertainment slides and motion pictures in depth. A Stereorama film projector has been developed, which back projects a sequence of slides on to a lenticular-type screen. A motion picture projector is being designed.
This paper will describe the implementation of a number of different image processing functions in the hardware of the Model 70 Image Display-Computer. Usually performed in host computer software, these algorithms run substantially faster. when converted to the hardware of a display-computer. The Model 70 architecture will be briefly outlined and its ability to perform operations globally will be discussed. Resulting speed improvements range from 3.25 to an estimated 230 times faster than typical software implementations. Use of the Model 70 to perform such intensive image processing operations allows the host computer to be smaller and less sophisticated.
The applicability of Synthetic Aperture Radar (SAR) imaging to many practical problems is severely limited by the excessive computational load associated with the technique. In this paper we describe a simple real-time, SAR processor which is implemented with acousto-optic and charge-coupled devices. This device will be useful in applications where real-time image formation is essential and in applications where present processors are not cost effective. It could also be used to preview large sets of imagery. The device is very compact and hence it has the potential to operate aboard the vehicle that collects the SAR data.
Historically, 8-bit/pixel (or less) imagery has predominated in most image processing applications. This is a result of a number of different factors, but is primarily due to psychovisual and cost considerations. However, a number of new imaging sensors and appli-cations involve data with as much as 16-bits/pixel in depth. Simply scaling up existing 8-bit image display/processor architectures has turned out to be both expensive and ineffi-cient in use of hardware resources. This paper presents a new approach to providing cost effective manipulation of 8 to 16-bit/pixel data using currently existing technology.
We present details of the EIKONIX Picture Processing Language, EPPL. The language was developed for use on Digital Equipment Corporation's PDP-ll and VAX computers to allow rapid acquisitions, manipulation, and display of digital images. We discuss the hardware that the language supports, the software architecture, the modes of data transfer and storage, the Parameter Set concept which simplifies user specifications of data and opera-tions, and typical image processing functions.
On December 19, 1980 a Los Angeles subscription TV system was the first to broadcast true stereoscopic (3D) programming. Using a unique electronic 3D Video* Process developed by the 3D Video Corporation, films originally shot in 3D are transferred to tape then chan-neled through a special electronic processor creating a 3D Master Tape. The resultant 3D tape can be broadcast conventionally or duplicated on video cassettes or video discs. The viewers at home wear anaglyph glasses to see true 3D on their color television sets. Sev-eral actual broadcasts have indicated considerable public interest and approval.
Omnispectravision is a real time electro-optic system that mimics and expands the ability of humans to receive radiant energy at multiple invisible wavelengths and convert it into visual stimuli thru the effect of display energy upon the photosensitive recePtors. A novel trichroic silicon mirror generates a composite plus two samples of three discrete infrared colors across a single InSb/PV detector array. The prototype digital scan con-verter with a recursive filter synchronously samples the multiplexed in-line detectors and transversal filters transform the λ3 > λ2 > λl infrared signals dispersed along the TV scan lines into -R-G-B (minus red, green, blue respectively).
An image chain analysis, utilizing a comprehensive computer program, has been gen-erated for the key elements of a digital image rectification system. System block dia-grams and analyses for three system configurations employing film scanner input have been formulated with a parametric specification of pertinent element modulation transfer functions and input film scene spectra. The major elements of the system for this analy-sis include a high-resolution, high-speed charge-coupled device film scanner, three candidate digital resampling option algorithms (i.e., nearest neighbor, bilinear inter-polation and cubic convolution methods), and two candidate printer reconstructor implemen-tations (solid-state light-emitting diode printer and laser beam recorder). Suitable metrics for the digital rectification system, incorporating the effects of interpolation and resolution error, were established, and the image chain analysis program was used to perform a quantitative comparison of the three resampling options with the two candi-date printer reconstructor implementations. The nearest neighbor digital resampling function is found to be a good compromise choice when cascaded with either a light-emit-ting diode printer or laser beam recorder. The resulting composite intensity point spread functions, including resampling, and both types of reconstruction are bilinear and quadratic, respectively.
intelligence templating is an analytical tool used to relate enemy activities to terrain and weather. On the rapidly changing modern battlefield, the commander needs more than just raw intelligence - hf needs quick answers. Templating, therefore, is an integral part of the commander's intelligence preparation of the battlefield. Three types of intelligence templating are doctrine, situation and event. A subset of event templating is equipment templating. The focus of this paper is equipment templating.
The basic requirements for combining imagery, graphic, and alphanumeric data into an operational system optimized for human needs are discussed. A data base system which contains all these data types is presented. Information display cells are defined and their use in handling images, graphics, and alphanumerics is explained. Scale, projection, overlay time, and multiple parameters are included in the complexity of the data base design and design constraints.
In visualizing objects in 3-D digital images based on shaded-surface displays, segmen-tation of the 3-D regions corresponding to the objects to be visualized is a basic processing operation. When object and non-object regions touch or overlap and have similar features, automatic segmentation methods fail. To overcome this problem, a method, based on creating an arbitrarily-shaped three-dimensional window that encloses only the object regions, is proposed in this paper. The window is created using the contours traced in a series of 'slices' comprising the 3-D image. The capability to create arbitrarily-shaped 3-D windows also permits 3-D visualization corresponding to arbitrary intersections of the object. Having identified the object regions in the 3-D image, the next important processing operation before the 3-D display of an object can be achieved is the formation of its boundary surfaces. A simple and fast algorithm to form boundary surfaces from a given set of contours is presented in this paper which makes use of the 'cuberille model' for the representation of discrete surfaces. The performance of the algorithms is illustrated using computerized tomography data.
Experimental data collected from a large number of transducers spatially distributed throughout a three-dimensional volume has typically posed a difficult interpretation task for the analyst. This paper describes one approach to alleviating this problem by present-ing color graphic displays of experimental data; specifically, data representing the dynamic three-dimensional distribution of cooling fluid collected during the reflood and refill of simulated nuclear reactor vessels. Color-coded binary data (wet/dry) are integrated with a graphic representation of the reactor vessel and displayed on a high-resolution color CRT. The display is updated with successive data sets and made into 16-mm movies for distribution and analysis. Specific display formats are presented and extension to other applications discussed.
Digital records of scene brightness, as recorded by Landsat's 4-band multispectral scanner (MSS), were enhanced and displayed in full color on a CRT and also in color print form. Digital enhancements, made through the use of appropriate algorithms, entailed corrections for (1) satellite roll, pitch, yaw and variations in heading; (2) scale variations caused by earth curvature, variations in satellite altitude and changing scan velocity of the MSS mirror; (3) radiometric variations resulting from variable gains within the MSS components; and (4) striping caused by an occasional repetitive malfunction of the MSS detectors. To facilitate the identification and mapping of wildland resource features, and thereby to provide resource managers with much needed information, several known image examples of each resource category were extracted from this enhanced, band-ratioed imagery and compiled into reference materials known as "image analysis keys". Through the proper visual display of these keys image analysts attempted to identify other examples of these same kinds of resource features on the digitally enhanced Landsat imagery. Conclusions regarding the increased identification accuracies achievable by such means are presented.