A system for use in conjunction with a stereomapper for determining conjugate points on stereoscopically generated images along a plurality of parallel lines during each mechanical translation of the images across the area being mapped. The system operates in the digital domain and incorporates control logic, either hardware or software, for modifying the data to correct the geometrical distortions, for selecting the data to be correlated in accordance with the parallax data computed from the preceding correlation, for correlating the selected data, and for generating parallax data. The conversion of the image data to digital form and its attendant storage capabilities permits the data generation to proceed independent of the generation of parallax data. The use of the special purpose minicomputers for data shaping, correlation, and the computation of parallax data permits the data generated by scanning the many parallel lines to be converted to parallax data without exceeding the computation capacity of existing electronics. Further, the scanning of a plurality of parallel lines during each translation of the images reduces the number of translations required to scan the desired area on each image and significantly increases the speed at which the area may be mapped.
Two-dimensional images of a three-dimensional body photographed by two video cameras are parallactically compensated prior performing the detection of the corresponding point necessary to recognize the body. Namely, the parallax between both cameras for the object distance measured by a range finder is calculated by a parallax calculator, and one image is electrically shifted by an image shifting circuit by only the number of pixels in accordance with the parallax value, thereby performing the parallax compensation of the images. In a calculation processing unit, the correlation values for the corresponding points between both images are calculated, and the distance images of the body are produced on the basis of these correlation values.
In order to assure sufficient precision for both binary image and variable density image, a corresponding point extraction method for a plurality of images includes the steps of taking in two binary images, then extracting an arbitrary set of epipolar lines, inserting a picture element having the picture element value of a virtual picture element between adjacent picture elements on said two epipolar lines as extracted, creating an initial screen having the initial values on a parallax screen, using two epipolar lines into which the picture element having the picture element value of virtual picture element is inserted, performing in parallel local calculation having excitatory combination and inhibitory combination based on actual picutre element and local calculation having excitatory combination and inhibitory combination based on virtual picture element for each black point on the created initial screen to create a new parallax screen.
The invention is directed to a method and an apparatus for scanning the terrain surface and other objects transversely to the flight direction or direction of motion of the scanner with a detector row arranged perpendicularly to the direction of scanning and containing several individual detectors of equal size, wherein the terrain- or object strip to be scanned is divided into various distance ranges, within which the fundamental detector signals within the detector row as well as of consecutive exposure periods are collated respectively in such a way by averaging, that the collated individual signals form or cover respectively one object pixel of approximately constant magnitude independent of the scanning distance, and wherein these object pixels stand always at right angles with respect to the scanning beam and where each scanning sweep produces a terrain- or object strip with approximately equal object pixel size as well as constant object pixel number along the width and length of the scanned strip. This method and apparatus eliminates the known deficiency of opto-mechanical scanners, whose scanned strips have changeable strip widths as a function of the scanning distance.
Three-dimensional measurement of a surface of an object under inspection is performed by projecting a light pattern upon the surface, having a periodic, smoothly varying light intensity in the X direction and having a non-changeable intensity in the Y direction in the manner of a grid, and thereafter electronically producing a pair of stereo images of the surface so illuminated by a pair of CCD arrays; measuring the parallax phase offsets between correlated corresponding homologous points in the stereo images, and converting the resulting parallax data to an array of Z-depth data, indicative of the shape of the surface of the object. A single left and right stereo snapshot is simultaneously obtained, so that relative motion between the object and the field of view has no adverse effect on accurate measurement, and the solid state light detection arrays are fully offset in an outboard direction with respect to the principal optical viewing axes, so that a stereo-based distance is provided which may be maximized to obtain more accurate measurements.
A parallax panoramagram has increased depth and sharpness when a sharpness filter is applied after interdigitation of multiple image portions. An optical path of wave train (204) (on-axis) and wave train (205) (off-axis) intersect a single lenticle (201). The lenticle has a focal length (208) and the on-axis (204) and off-axis (205) wave trains correspond to different stripes. These wave trains (204 and 205) comme to focus at points (206 and 207) respectively. The surface (202) has a cylindrical curvature, and bracket (203) denotes the width of the lenticle (201). Each eye of the observer sees its own perspective view when looking at a lenticular stereogram.
