| 摘要 |
[0001] A system projects a user-viewable, computer-generated or -fed image, wherein a head-mounted projector is used to project an image onto a retro-reflective surface, so only the viewer can see the image. The projector is connected to a computer that contains software to create virtual 2-D and or 3-D images for viewing by the user. Further, one projector each is mounted on either side of the user's head, and, by choosing for example a retro angle of less than about 10 degrees, each eye can only see the image of one of the projectors at a given distance up to 3 meters, in this example, from the retro-reflective screen. The retro angle used may be reduced with larger viewing distance desired. These projectors use lasers to avoid the need for focusing, and in some cases these projectors use instead of lasers highly collimated LED light sources to avoid the need for focusing. The retro angle, as used herein shall mean the opening angle from the center axis of the reflected cone to a surface line on said cone. That is for practical purposes, widest angle a viewer might get a good quality image. At least one camera is mounted near a projector on the user's head and is used to adjust the image or used to observe user interaction with the projected image. A beam and sensor may be added in an invisible wavelength, and the retro-reflective surface may have fiduciary markings in color not visible to the human eye, but contrasting in the invisible wavelengths, and the sensor may be able to recognize the retro reflection or its absence, thus being able to read human invisible fiduciary markings, allowing the system to calculate the user's head position and orientation relative to the surface. A user, interjecting objects, may create a disruption of the reflected invisible beam, and detection of such interruptions can be interpreted by the system as commands for actions, including but not limited to navigation in a virtual environment, launching of programs, manipulation of data, and so forth. The user interface takes into account the natural use of hands and objects by means of "virtual" hands, simple auto calibration, and alignment with natural 3-D vision, without the need to "transpose" like most pointing devices. It offers personal adjustments to actual eye positions and actual intra-ocular distance, as well as correct horizontal and vertical disparity, correcting for inclination (for example, a user lying on a couch looking sideways) and changes during viewing and interaction. By allowing fiduciary marks to be read on each scan line, not just full image scans, and using fiduciary markings that contain full location information, a very fast feedback is provided, typically around 500-1000 times the frame speed currently used by conventional methods with full frame cameras, for example, including but not limited to, Kinect and Sony PS3 EyeToy and 3D Systems. Those typically use cameras that introduce frame, shutter, and/or frame buffer serial delays, creating human-noticeable latencies. By use of direct first-person view, the short trip to the screen and back can be computed instantly and accurately, using a 500-1000x faster and near zero latency instantaneous detection of first-person motion detection. The projector and screen observing sensors are at the same vantage point and in a fixed relationship to each eye. Thus the degree of uncertainty is greatly reduced because of not doing successive estimations, each with both an error and one or more frame delay. Additionally, the system adjusts for changed head and eye position; i.e., vantage point detection, making adjustments for both horizontal and vertical disparity, head rotation, head rotation in response to fast moving objects (such as following a ball, dodging or ducking a ball in a motion sport simulation or augmented reality games). The construction of the device is simplified with integration of only a few parts, low weight, and low power cost, enabling ultra-light, affordable Eye Stalks. By employing screen corner cube array diversity (pseudo randomness, deliberate avoidance of excessive periodicity) the notion of specie and interference patterns can be vastly reduced. Combinatorial efficiency of multiple primaries (four or more) with feedback from the screen, from observed response is different from prior art that typically uses diversity as a deterministic solution, regardless of actual image and artifacts occurring. The observed response can be used to switch between various schemes to find optimum trade off and varying them based on a combination of real time instantaneous observed screen response (e.g., speckle in certain areas, at certain sceen scan angles, with certain color mixes, at a certain brightness). Each primary can be adjusted for observed intensity versus intended intensity. Multi layer screens (two and three ways) can combine multiple retro-reflective functions with specular (for virtual panning) and diffusion (for illumination and pointing devices) and absorbtion (for high contrast suppression of ambient light). The system can be switched to allow any combination of 2-D and/or 3-D projection within a same field of view and based on vantage point, hi some cases, rather than use a full head-mount system, the parts that need to be placed near the eyes can be implemented as "parasitic" clip-on glasses frames and can connect to a mobile phone (smart phone) as a computing device. |
