US 6,028,672 AGranted

High speed three dimensional imaging method

AssigneeZheng J. Geng

InventorZheng Jason Geng

Life of this patent

Estimating the term…

Loading patent…

Abstract

The target of the present invention is to provide a high speed, low-cost, multi-mode three-dimensional (3D) surface profile measurement method. The proposed Rainbow Stereo 3D Camera exploits the projected color light with a spatially distributed wavelength spectrum on the surface of objects in the scene. Multiple color imaging sensors separated by a baseline distance are used to capture stereo pair images of the scene at camera's frame rate. The 3D depth values are calculated using triangulation principle by finding pixels corresponding to a common color feature in both images. Unlike conventional stereo correspondence matching algorithms, which requires the feature extraction from a group of pixels, the proposed method utilizes a projected rainbow color pattern as unique landmarks of each pixel for correspondence registration. Essentially, the colors of pixels in a pair of stereo images are used as a "token" to perform the stereo match. Searching corresponding point in a pair of stereo images becomes a straightforward pixel-to-pixel color matching. A simple and efficient 3D triangulation algorithm can be formulated to generate full frames of 3D images in high speed.

Claims6

Dependent claims

An apparatus for obtaining 3D image of objects in a scene, comprising

(a) a light projection means for generating a bundle of light rays with spatially varying wavelengths to illuminate surface of objects in the scene such that spatially varying color patterns are formed on the surface of objects through the light projection, and
(b) a sensor means for acquiring images of objects in the scene from multiple viewing locations, based on which the wavelengths can be detected of light rays reflected from objects' surface that is illuminated by said light projection means, and
(c) a calibration means for determining the geometric and optical parameters of the said sensor means at multiple viewing locations, and
(d) a data comparison means for matching the correspondence pixels in multiple data sets obtained from the said sensor means based on their wavelength values and/or local color patterns, and
(e) a data extraction means for computing distances in the 3D space between sensor means and surface points in the scene, and therefore determining 3D coordinates of surface points of objects in the scene, based on information of (i) locations of the matched correspondence points in multiple images obtained by the said sensor means based on color comparison, and (ii) geometric and optical parameters of the said sensor means obtained by the said calibration means,
whereby 3D images, in which each pixel value represents a 3D distance between the said sensor means and a surface point on objects in the scene that corresponds to the pixel, can be obtained using triangulation principle.

3 more elements

An apparatus for obtaining 3D image of objects in a scene, comprising

(a) a light projection means for generating a bundle of light rays with spatially varying wavelengths to illuminate surface of objects in the scene such that spatially varying wavelength patterns are formed on the surface of objects through the light projection, and
(b) a sensor means comprises a set of multiple color CCD cameras that provide a set of multiple stereo color images of objects in the scene from multiple viewing locations, based on which the wavelengths can be detected of light rays reflected from objects' surface that is illuminated by said light projection means, and
(c) a calibration means for determining the geometric and optical parameters of the said sensor means at multiple viewing locations, and
(d) a data comparison means for matching the correspondence pixels in multiple data sets obtained from the said sensor means based on their wavelength values and/or local color patterns, and
(e) a data extraction means for computing distances in the 3D space between the sensor means and surface points in the scene, and therefore determining 3D coordinates of surface points of objects in the scene, based on information of (i) locations of the matched correspondence points in multiple images obtained by the said sensor means based on color comparison, and (ii) geometric and optical parameters of the said sensor means obtained by the said calibration means,
whereby 3D images, in which each pixel value represents a 3D distance between the said sensor means and a surface point of objects in the scene that corresponds to the pixel, can be obtained using triangulation principle.

3 more elements

An apparatus for obtaining 3D image of objects in a scene, comprising

(a) a light projection means for generating a bundle of light rays with spatially varying wavelengths to illuminate surface of objects in the scene such that spatially varying wavelength patterns are formed on the surface of objects through the light projection, and
(b) a sensor means comprises single color CCD camera with a multiple image collecting means, which enables the single camera to capture color images of objects in the scene from multiple viewing points to form stereo image pairs, based on which the wavelengths can be detected of light rays reflected from objects' surface that is illuminated by said light projection means, and
(c) a calibration means for determining the geometric and optical parameters of the said sensor means at multiple viewing locations, and
(d) a data comparison means for matching the correspondence pixels in multiple data sets obtained from the said sensor means based on their wavelength values and/or local color patterns, and
(e) a data extraction means for computing distances in the 3D space between the sensor means and surface points in the scene, and therefore determining 3D coordinates of surface points of objects in the scene, based on information of (i) locations of the matched correspondence points in multiple images obtained by the said sensor means based on color comparison, and (ii) geometric and optical parameters of the said sensor means obtained by the said calibration means,
whereby 3D images, in which each pixel value represents a 3D distance between the said sensor means and a surface point of objects in the scene that corresponds to the pixel, can be obtained using triangulation principle.

3 more elements

Dependencies are inferred from the claim text.