This invention relates to interactive display systems and more specifically to a system for remotely displaying and sensing information using shadow parallox.

The device of this invention performs the functions heretofore requiring several separate devices and technologies. These are reviewed as they relate to display, data input, and remote contact sensing.

Referring first to display, it has been a goal in the display of information in an office environment to minimize the desk space required by a personal display. Prior art alternatives include flat display technologies, such as plasma discharge and liquid crystal, and integration of a display into the work table or desk. It is believed such a "non-invasive" display would facilitate acceptance of a terminal in areas such as principal support.

Referring next to data input, besides the keyboard for the input of coded information, there are diverse means of inputting data to a computer system. These will be examined by questioning : does the technology permit an output image to occur in the same plane as the input data ? Can an operator input data by tracing over a printed or photographic image such as a map ? Can-the operator input data in real time by writing on paper ? Can an ordinary writing instrument or even a finger be used to point, or is a special device needed ? Can the input device be used on a horizontal desk top ? Can images from photographs, books, or other prepared sources be entered ? Now the various input technologies are discussed. A photocell, usually mounted on the end of a rod, and used with. a CRT or similar refresh device., is called a light pen. In use, the photocell is pointed to a section of the image selected by a user. The photocell will then respond to the point of light produced as the scanned electron beam of the CRT.passes under the photocell. The time of this response in relation to the known position of the electron beam indicates the location of the light pen relative. to the image. This technology allows in-plane interaction with the image, and of course, possesses inherent image registration. It does not permit graphic tracing, finger manipulation, real time entry on paper, use on a horizontal desk top surface, or the inputting of charts, books, or photographs. A second technology to be discussed is the data pad. A data pad system usually consists of a flat, relatively thin "pad" of specific construction, and a pen, often of specific construction. Many technologies are used, and will not be discussed here as they are not relevant to this invention. Several types of data tablets are made of transparent material, and by placing them over a display such as a CRT, a nearly in-plane interaction is possible. However, there is no inherent image registration and parallax problems prevent accurate pointing. Most data tablets are opaque, and are used quite differently than transparent tablets. Opaque tablets permit graphic tracing, real time entry on paper, and the use of a horizontal desk top surface. A few provide finger manipulation at the expense of precision when using a pick such as pencil or pen. Many require a special pen to be used. They, of course, do not permit in-plane interaction with a computer image, and cannot input from charts, books, and photographs. The final input technology to be considered is the facsimile scanner, which converts a document into an electronic image. The scanner is the only technology-capabl of inputting charts, books and photographs, but possesses generally none of the other features being examined. The technology by which this is accomplished may be that of a television pickup tube such as a vidicon, a charge-coupled device with.an array of sensors and a shift register, a mechanically rotating drum, or a flying spot scanner. This device gets its name from a "flying-spot" of projected ligh that rapidly moves across an image to be scanned. A photocell aimed generally at the image will receive more light a the spot passes over a white surface, and-less light as the spot passes a black area, thus effectively scanning selected areas of the scene. The first television images were "seen" by a flying-spot scanner using a mechanically rotating disk called a Nipkow disk. Modern flying-spot scanners use a _CRT to generate a moving spot. They are popular for television pickup from film and are still used in facsimile scanners by focusing the flying-spot from a high-resolution CRT onto a document in a light-tight box.

Referring finally to remote contact sensing, it may be seen that contact sensing presents a distinct problem from positional sensing within a plane. With a light pen, for example, data-input normally is not desired until the image is touched. When using a data pad for real-time handwritten entry, it is of course vital that when the pen is lifted marginally above the paper that the computer be aware that the line is discontinued. This sensing is generally done either by a mechanical switch or electrical device in the pen, or with. a special electrical device in a data pad on which the input is to be written. As another means of contact sensing, when writing or generating artwork, a human will sense contact by tactile feedback and, in slow precision work, by visual parallax. Using parallax, the image including the pen is viewed simultaneously from two angles. Only when two objects are in the same plane their apparent separation appears equal from both angles. Parallax is also used to judge distance, angle, and separate distinct ob^÷ jects. Parallax is very difficult to perform with. a computer system because not only is high resolution required, but the techniques of pattern recognition necessary to find and separate. objects are not yet generally of adequate quality, despite. much. work in this field which. is considered a subject of artificial intelligence. The problem is that surface details must be used to derive correlations, and these details themselves, camouflage dimensional boundaries.

By combining. all the desired features of the display, data input, and contact sensor discussed above, the resulting device would form a touch sensitive image devoid of imaging or sensing hardware in the display area, permitting an ordinary paper on an ordinary desk to become a display and also to sense contact with an ordinary pencil or even a finger. In the prior art, such a device was not possible.

The object of this invention is therefore a system for remotely sensing contact of a pen, pencil, finger, other pick with a surface, using shadow parallax.

The system of the invention comprises a central processor with a storage, a display means for projecting an image representative of information stored in the storage onto a selected area, an introduction means for introducing an image of information not stored in the storage onto said area, a-flying spot scanner device for scanning the area and sensing the information not stored in the storage, and an input means for storing the sensed information in said storage.

In an improvement of the invention, the flying spot scannei device includes a plurality of non-concentric sensors such . that each senses an independent shadow on a common perspective of the introduction means indicative of said introduction means.

The invention is now described with reference to the following drawings

• FIG. 1 shows the environment in which the system operates.
• FIG. 2 diagrams the operation of the invention.
• FIG. 3 depicts shadowing perspectives with flying-spot scanning.
• _FIG. 4 shows a two-sensor arrangement for shadow parallax.
• FIG. 5 illustrates the use of shadow parallax in remote contact sensing.
• FIG. 6-7 portray alternate forms of the invention.
• FIG. 8-9 illustrate the use of shadow parallax in dimensional form sensing.

Referring now to FIG. 1, by way of illustration a typical operating environment is presented. An operator 1 is seated at work area 2, and views a projected image 3, which may be focused on a document, work piece, or any other material. The image is projected from device 4, which contains both a display system and sensors to be discussed in detail later. Device 4 is connected to Computer system 5 including a central processor and storage which controls what is displayed by device 4 and interprets the information received by the sensors in device 4 to form a complete system. This application will be concerned with the operation of device 4, its effect on image 3 and its capacity to input data from work area 2, and with the method used in computer 5 to initially interpret the data received by the sensors in device 4.

The device 4 of FIG. 1 is detailed in FIG. 2. The image produced by a projection kinescope, or high intensity CRT 10, is reflected by mirror 11 and focused by lens 12 onto work surface 13. The work. surface 13 is equivalent to the work area 2 in FIG. 1. The image formed by CRT 10 is controlled by position and intensity signals 14 and 15, respectively, input from conventional computer and drive circuitry (not shown). The high intensity CRT could, of course, be replaced by a laser and acoustic scanner, or equivalent technology. The dixection of projection along line 16 is not required to be perpendicular to the work surface 13, An angle other than perpendicular may be chosen to minimize the shadowing of the image by the operator. Unlike a lamp, specular reflection off the work pieces of the projected image improves image brightness and contrast. Geometric distortions of angled projection are correctable in the deflection control circuitry, and focus is corrected by a lens-image tilt as is common practice in view."camera photography. The image formed on work area 13 is completely non-invasive in that there is no display hardware in the area of the image. The image may be formed on many materials, for example, if projected onto a paper form, a forms fill-in function can be easily performed, with the projected image and printed image both easily seen and distinguished from one another.

Continuing now with FIG. 2, photocells 18 are positioned to view the image projected on work area 13, acting as flying-spot scanner type sensing elements. Photosensors 18 are identical assemblies that view the image from different angles. The use and positioning of each one will be discussed later. However, unlike a flying-spot scanner, the sensing from the projected image must operate in an environment of normal work area illumination, which typically in an office includes fluorescent lights that flicker at the 120 Hz power cycles of the 60 Hz power line.

Methods of suppressing the effects of ambient light will now be presented. The angle of view by photocell 18 is limited by lens assembly 19 to include only the projected image on plane 13. However even after this angle limiting the magnitude. of the ambient light reaching sensor 18 is typically several times greater than the magnitude of light originating from the projected image. A'colored filter 20 matched to the projected color and counter- matched to the ambient light color further reduces the effect of ambient light. If a narrow-band phosphor is used in CRT 10 and those bands are separate from the mercury bands in fluorescent lighting, a narrow band interference filter or spectral separator can be used for a greater effect. A context sensor 21 can be added to purposefully receive. primarily ambient light. This signal of ambient light is then subtracted from the signal from sensor 18, leaving in purer form a signal representing the magnitude of light originating from the projected image alone. Also the intensity control 15 to the CRT may be modulated which., in conjunction with a fast-decay phosphor in CRT 10, causes photocell 18 to receive an amplitude modulated signal, with. amplitude corresponding to the reflectance of the point being scanned on work area 13. Because the harmonics from a fluorescent tube typically fall off at about 6db/octave above 120 Hz, modulation raises the desired signal above the noisy base-band to a frequency with little interference. The modulation may be performed in hardware, or in software by alternately toggling the beam on and off and reading the light in each state. To summarize, the effect of ambient light is removed by a combination of spatial filtering, spectral filtering, context nulling, and modulation-demodulation.

In conjunction with the ambient filtering, the unit is able to perform as a flying-spot scanner, inputting images from documents, forms, photographs, and other facsimile sources laying on an ordinary work surface. It is also able to display information concurrent with inputting information from the same plane on which the display appears, with both input and output functions featuring inherently perfect spatial registration because the same "flying-spot" performs both. functions. An example of an application using this feature is interactive forms fill-in, allowing a computer to "highlight" lines to be filled in while projecting related data or suggested responses. Another example is an interactive assembly station, allowing a computer to find and "highlight" parts, to show by direct pointing exactly where a part is to go, and to check for visually correct assembly.

To permit general data entry, the system must be capable of remotely sensing contact between a pointer or pick, such as a.pencil or finger, and the work surface being Viewed. This will be accomplished with a technique which is given the name "shadow parallax" as is now described.

Referring now to FIG. 3, consider the flying-spot scanner consisting of photosensor 30 and projection CRT 31 with lens 32. A computer 33 viewing the scene 34 via the photosensor 30 would see objects in the perspective 35 of the projection scanner light source, CRT 31 and lens 32. The photosensor 30 through which the computer views the scene 34 appears in the image 35 as a light source, capable of casting shadows 36, but not affecting perspective. Note the reciprocal effect of the source-sensor arrangement or shadows and perspective compared to conventional imagery.

Next consider the two sensor arrangement of FIG. 4. Depending on which sensor 40 or 41 is selected through which to image scene 42, a deep shadow will appear either on the left or right of the pencil 43, cast by the irtual source, which is the selected one of the two sensors 40 or 41. These images are shown in FIG. 5 as a and b. Note that both shadow images can be viewed independently and simultaneously, without interference between them. Unlike binocular vision, which permits several perspectives but with the same shadows, the nature of a flying-spot scanner does not permit independent perspectives, but does permit a plurality of independent shadows. The latter effect, called here "shadow parallax", as opposed to perspective parallax, has strong advantages in the present invention. Shadow parallax is the complement of binocular vision.

The image formed by taking the mathematical maximum of the signal point by point from the left and right sensors of FI_G. 4 is shown in c of FIG. 5 to be free of all shadows. It may be used to capture the location of the pencil using conventional methods of pattern recognition. The image formed by taking the point by point minimum of the left and right sensors is shown in d of FIG. 5 to have very deep shadows on both sides of the pencil. As contact is made with. the surface, the two shadows merge, as in e of FIG. 5. After the pencil location is determined from c, the ability to distinguish between states contrasted in d and e is-simple. A triangle of three sensors, or more, would aid in sensing contact at arbitrary pick angles.

Common perspective parallax refers to distances relative to the sensor, and is unable to measure the distance to surfaces, such as white paper on which surface details are extremely faint, graph paper for which the correlation solution would be multi-valued. Shadow parallax, on the other hand, directly measures the third dimension relative to the writing surface, and thus performs equally as well with thin paper, paper pads, or an irregular surface such as a calculator keyboard or bound notebook, or with clean white paper, graph.paper, or documents. In addition to better function, shadow parallax also realizes a cost advantage, as only the sensor, not the scanner, is duplicated, and less computing power is needed as the difficult image ccrrelation of fferent visual perspectives is obviated..

Referring again to FIG. 5, illustration c represents the shadowless image which the computer could use to track the x-y position of the pencil. In practice, a distracting background would complicate tracking, as illustrated in c'. Dividing the image of d by c, with appropriate provision for the case of c approaching zero, cancels the pencil, text, and other distractions, leaving only the very black shadows of d' for easy computer tracking, free from camouflaging patterns.

Referring now to FIG. 6 and FIG. 7, alternate physical forms of the invention are portrayed. FIG. 6 shows a 5 system with the entire module ceiling mounted in a recessed box resembling a standard.2 x 4 feet fluorescent light fixture module. A similar alternative could form the assembly to.resemble a track light fixture which could be mounted from a suspended rail. In both cases, the system could draw power through the ceiling wiring as would a lamp, and could be connected to a computer also via hidden ceiling wiring. Mounting the system in the ceiling would provide the most non-obtrusive arrangement. A second alternative is shown in FIG. 7, in which an upward pointing CRT is contained in the vertical "neck" and a lens with two mirrors 30 is held in the head to redirect the image down at a desirable angle. This system, resembling a "desk lamp", would be portable and require no special installation. Other configurations are, of course, possible, and these are given as samples and not limits.

So far the description has used shadow parallax to remotely sense Z motion of a pick relative to a plane, and contact of the pick with the plane. Shadow parallax can be used to obtain an image containing only shadows and surface texture, devoid of surface albedo patterns, as though everything were painted gray, reminiscent of an electron micrograph. This idealized "all gray" image would aid the ability of - computer-driven industrial robots to locate and manipulate parts. By using three sensors arranged as in FIG. 8, and dividing the image from each by the maximum of all three, (analogous to dividing the images of FIG. 5, a by c and b .by c) to obtain the effect illustrated in FIG. 9, a three-dimensional mapping and true physical boundary sensing of the environment is easily obtained. Such a mapping has been an elusive quest for artificial intelligence and industrial robotics through decades of research attempting to use stereo parallax or pattern extraction of a single image. Shadow parallax is a key to using computer-driven robots, for wide-ranging applications from parts assembly to precision dentistry.

Microfiche is unlikely to replace books-in libraries, and displays are unlikely to replace papers in offices. People want something tangible to hold, to interpret without a machine, and to spread out on a desk. This invention provides the only system capable of turning ordinary paper into an interactive tablet and display, and conversely of turning an electronic document into a desk top image resembling ordinary paper in many functional ways. The method of this invention provides a device that is equally at home perfoming the jobs of a display, a tablet, or a scanner, or all three at once, and all on the flat, gadget free, surface of an ordinary desk top. The operator is free to use ordinary paper with an ordinary pencil on an ordinary desk top, seeing nothing in the display area but a touch-sensitive image devoid of harware.

It will be obvious to one skilled in the art that modifications to the specific examples used for illustrations may be made without departing from the spirit and scope of this invention.