Three-dimensional surface contour information is important for selective patients requiring radiation therapy. The patient contour is used to calculate the radiation dose distribution within the irradiated volume. A Moire photographic contouring device has been constructed to obtain these surface contours. The device consists of two light sources to project a grid pattern on the skin of the patient, and a camera with Polaroid back to photograph the image of the interference pattern on the skin. The interference fringes, as well as the patient's surface, is recorded on the photograph. Tissue thickness can be determined within 2.5 mm. It takes only a few minutes to get the contouring photograph. In addition to using the surface contour data for calculations of dose distributions, it can also facilitate construction of tissue compensators.
The technique of machine perception of three-dimensional objects implies the generation of a computational model of a solid in a computer. The topic of Computational Geometry deals with the representation, manipulation and generation of geometric data in computational systems. Four principle classes of representational schema are used: point sets, facetted models, functional and procedural. An abstract data type is defined by the values it may take on and by the operations which are defined over those objects. The operator classes associated with three dimensional solids are: solid transformations, visualization, continuation (or interpolation), mensuration and set operationsions. The utility of the various mathematical and computational formulations of computational geometry to the problems of machine perception of three-dimensional objects will be analyzed.
We describe a "stereo camera" which rapidly acquires the (x,y,z) locations of a large number of sample points on a surface. The device works on the same triangulation principle as an ordinary stereo camera, but involves an active camera (coded-laser-beam-array projector), a passive camera (TV), and the software for pattern synchronization and analysis. The system involves (1) a single laser beam; (2) interference optics to transform the single beam into an array of laser beams; (3) a programmable electro-optic shutter which can block or transmit various subsets of the laser beams in the array; (4) a TV camera to view the environmental surfaces illuminated by the laser beam array; and (5) a processor which controls the shutter, synchronizes the shutter with the TV camera, and decodes the contents of the TV images of the different coded patterns projected on the environment. The system is being developed for rapid biomedical surface mapping but can also be used in vision aid for the blind, robot vision, assembly-line inspection, data acquisition for 3-D pattern recognition, and the monitoring of large surfaces and volumes (dams, construction processes, security areas, etc.).
Coherent optical methods are described for application in the artificial visual sensing function of robot systems. Prior methods for scene measurements are reviewed. A new approach based on the coherent properties of laser radiation for the quantization of the scene is presented. The results from a prototype optical system are presented, and several potential features for further development are discussed.
This paper surveys some issues confronting the conception of models for general purpose vision systems. We draw parallels to requirements of human performance under visual transformations naturally occurring in the ecological environment. We argue that successful real world vision systems require a strong component of analogical reasoning. We propose a course of investigation into appropriate models, and illustrate some of these proposals by a simple example. Our study emphasizes the potential importance of isomorphic representations - models of image and scene which embed a metric of their respective spaces, and whose topological structure facilitates identification of scene descriptors that are invariant under viewing transformations.
This paper describes design and development of two types of 3-D ranging devices for robotic applications: both are noncontact, optical devices based on active illumination and triangulation. One is a laser scanning ranging device for medium-range (50cm), and the other is a proximity range sensor for short range (5 cm). The features of the devices are that both use analog area position sensor chips and that they provide simple, fast, accurate, non-contact visual sensing of range information.
The National Bureau of Standards is developing a vision system for use in an automated factory environment. The emphasis of the project is on the real-time acquisition of three-dimensional parts using visual feedback. The system employs multiple light sources in con-junction with object models to establish the position and orientation of an object in the camera's field of view. A flood flash enables shape information to be obtained from an image, while a plane of light can be used to find the three-dimensional positions of points on the object. Because there are only a small number of object types and the objects all have pre-defined nominal locations, a model can be used to predict how the scene should look from a given viewpoint using a particular light source. This prediction can be compared with the actual image, and the differences used to establish position information. Models are expected to be particularly useful in reducing the number of views of an object necessary to calculate its three-dimensional position.
Some of you attended the workshop on "Replication Technology" which was sponsored by M.I.T. in January, 1979. On that occasion, I reviewed the early work which led up to the recent, renewed interest in photo-sculpture. I was pleasantly surprised at the favorable response to that presentation and for today's seminar I thought I would take a similar approach to a closely related subject. My topic is somewhat narrower, but I hope it will be interesting and instructive to trace the origins of a simple but rather elegant technique for 3-D sensing--light beam profiling--and examine the current status and future pros-pects for this relatively inexpensive, but quite versatile technology.
Anthropometric data on reach and mobility have traditionally been collected by time consuming and relatively inaccurate manual methods. Three dimensional digital image acquisition promises to radically increase the speed and ease of data collection and analysis. A three-camera video anthropometric system for collecting position, velocity, and force data in real time is under development for the Anthropometric Measurement Laboratory at NASA's Johnson Space Center. The use of a prototype of this system for collecting data on reach capabilities and on lateral stability is described. Two extensions of this system are planned.
The diagnosis and treatment of malocclusion, and the proper design of restorations and prostheses, requires the determination of surface topography of the teeth and related oral structures. Surface contour measurements involve not only affected teeth, but adjacent and opposing surface contours composing a complexly interacting occlusal system. No a priori knowledge is predictable as dental structures are largely asymmetrical, non-repetitive, and non-uniform curvatures in 3-D space. Present diagnosis, treatment planning, and fabrication relies entirely on the generation of physical replicas during each stage of treatment. Fabrication is limited to materials that lend themselves to casting or coating, and to hand fitting and finishing. Inspection is primarily by vision and patient perceptual feedback. Production methods are time-consuming. Prostheses are entirely custom designed by manual methods, require costly skilled technical labor, and do not lend themselves to centralization. The potential improvement in diagnostic techniques, improved patient care, increased productivity, and cost-savings in material and man-hours that could result, if rapid and accurate remote measurement and numerical (automated) fabrication methods were devised, would be significant. The unique problems of mapping oral structures, and specific limitations in materials and methods, are reviewed.
Creating machines that will see in human-like fashion requires an understanding of human perception. This paper summarizes certain advances in visual science that suggests perception may be structured from a hierarchy of filtered images. It will be shown that a small numbered set of images created from filters based on biological data can provide a rich array of information about any object: contrast, general form, identification, textures and edges. It is argued that machine perception will require similar parallel processing of an array of filtered images if human-like visual performance is required. Some visual problems, such as certain visual illusions, multi-stable objects and masking are analyzed in terms of limitations of biological filtering. Machine solutions to those problems will be discussed.
This overview is the first presentation in a Session devoted to 3-D display of internal structures. In the following presentations different modalities for 3-D display of internal structures are discussed. For each modality a leading worker in the field gives an oveview of that modality with an emphasis of his own practical work. In just about all cases to be discussed, the data which we input to the 3-D display process have been acquired and processed by some sort of computerized tomography device. This overview describes the data acquisition and processing by such devices. While the medical application is used for illustration, emphasis is placed on the fact that nearly identical pro-cedures are applicable in many other fields, including non-destructive testing.
The currently-existing 3-D imaging technologies provide 3-D digital images of scenes in the form of 3-D arrays of numbers. In the representation of an object as a set of voxels, a boundary surface of the object is a set of faces of voxels. In spite of the large number of faces that form a boundary surface of a typical object, fast algorithms for hidden part suppression and shading of such surfaces are made possible by the simplicity of the geometry of the 3-D voxel environment. However, because of the very limited number (only three) of orientations of the faces, the shading rule based on the direction cosine of the face normals and distance of the faces from the observer sometimes produces rough display images of originally smooth surfaces. This causes an undesirable change of smoothness of the display image from one view to another in a dynamic mode of display. We propose a contextual shading scheme which assigns shading to a face based on the local shape of the surface in the neighborhood of the face. The number of computations required per face is kept to a minimum by precomputing and storing all the possible direction cosines used for shading. The new shading algorithm has speeds comparable to that of the algorithm based on three face orientations, and produces far better display images.
Three-dimensional display algorithms using computer graphics are evaluated for potential clinical utilization. The authors are utilizing triangular tiles for organ surface reconstruction based on contour information generated from computed tomography (CT), ultrasound examinations, and nuclear medicine examinations. Comparisons have been established among several triangular tiling algorithms which include the number of points required to adequately specify the contours, computer execution time, and the errors in reconstruction. The triangular tiling algorithm provides a means for accurately estimating the volume and surface area of the desired anatomic sites. Examples of studies include brain lesions, computed tomography radiation treatment plans, spleen, and nuclear medicine slant hole camera studies.
Volume images made up of "stacks" of parallel computed tomographic (CT) cross-sectional images are displayed in three dimensions utilizing the method of projection imaging. This technique involves the mathematical projection of the volume picture elements (voxels) of the 3-D image onto a plane to form a two-dimensional projection image which, for x-ray CT volume images, resemble conventional radiographs. Projection images formed at two angles of view, 2° to 8° apart, are utilized as stereo-pair projections to view the volume image in three dimensions. Before projection, selected regions of the volume image are partially dissolved or totally removed from the volume to enhance the visibility of remaining struc-tures. These processes, referred to as numerical tissue "dissolution" and "dissection", are utilized to overcome the undesired effects of superposition which occur as natural consequence of displaying a stack of cross sections as a volume image, i.e., deeper image regions are obscured by overlying structure. Examples are shown where overlying regions of the volume image have been "cut" from the volume to more clearly visualize deeper anatomy. Particular emphasis is given to the use of these methods in identifying two-and three-dimensional subregions of interest within the volume for further detailed view-ing and quantitative analysis. As an example, the use of the 3-D display of volume images to guide the process of identifying the optimal orientation of oblique section images through internal organs of the body is illustrated.
A new 3-D viewing system will be described which allows a physician to simultaneously examine multiple CT or B-scan ultrasound scans in their proper orientation in all three dimensions. Several persons can view the display at the same time and there is no need for special glasses. Test images and line drawings displayed on a prototype viewing device exhibit both stereoscopic and motion parallax depth cues characteristic of a real three dimensional object. Operating principles of the device will be outlined and a new 3-D figure/ground illusion will be described.
A software for developing, editing and displaying a 3-D computerized anatomic atlas of a human brain is described. The objective of this atlas is to serve as a reference in identifying various structures in CT scans.