System and method for 3-D object rendering of a moving object using structured light patterns and moving window imagery
AssigneeSpatial Integrated Systems, Inc.
InventorsPing Zhuang, Yunqi Zhang
Life of this patent
Estimating the term…
Abstract
A moving object to be subject to 3-dimensional rendering is illuminated by a repeated sequence of a number N different two-dimensional structured light patterns. For each illumination, the illuminated object is captured by two spaced apart cameras, thereby creating a series of pairs of two-dimensional images. A set of N adjacent pairs of the two-dimensional images is used to form a single three-dimensional image frame. A moving window of N adjacent pairs of the two-dimensional images provides a train of such three-dimensional image frames. The object may furthered be illuminated by non-structured light whose captured image reflection may be used as an overlay over the three-dimensional rendered object.
Claims23
A method of three-dimensional object rendering of a moving object within a field of view, the method comprising:
- illuminating a moving object to be rendered with a repeated sequence of a number N different 2D structured light patterns, wherein N is at least three;
- for each illumination, capturing a 2D image set reflected off of the moving object at a number K spaced apart camera locations to thereby create a train of captured 2D image sets Hieach captured 2D image set comprising K images, wherein i is a counting integer and K is at least two;
- and
- for each overlapping set of a number M adjacent captured 2D image sets Hm, Hm+1, . . . , Hm+M−1, obtaining a single 3D rendered object frame Fm, to thereby create a train of rendered three-dimensional image frames, each 3D rendered object frame Fm formed from a moving window of said captured 2D image sets Hm, Hm+1, . . . , Hm+M−1, wherein M is at least three and not greater than N;
- such that:
- successive 3D rendered object frames are obtained from overlapping sets of M adjacent captured 2D image sets.
The method according to clam 2, wherein the two-dimensional three structured light patterns are created by:
- (a) obtaining four triangular waveforms, each triangular waveform having a predetermined amplitude and predetermined periodicity;
- (b) phase shifting each triangular waveform by a predetermined amount;
- (c) summing the phase shifted triangular waveforms to thereby form an aperiodic summed waveform;
- (d) gamma correcting each pixel in the aperiodic summed waveform;
- by a gamma factor to form a gamma-corrected waveform;
- (e) rescaling the gamma-corrected waveform to conform to the pixel value range of the projection device, thereby forming a one-dimensional illuminating waveform;
- (f) replicating the one-dimensional illuminating waveform in a second dimension to form a two-dimensional structured light pattern;
- and
- (g) repeating steps (b)-(f) at least two more times with different combinations of phase shifts to form a total of at least three two-dimensional structured light patterns.
A structured light system for three-dimensional object rendering of a moving object within a field of view, comprising:
- a light projector configured to illuminate a moving object to be rendered with a repeated sequence of a number N different 2D structured light patterns, wherein N is at least three,
- a number K cameras configured to capture a 2D image set reflected off of the moving object in response to each illumination to thereby create a train of captured 2D image sets Hieach captured 2D image set comprising K images, wherein i is a counting integer and K is at least two;
- and
- a computer configured to:
- receive 2D image sets from the K cameras and
- for each overlapping set of a number M adjacent captured 2D image sets Hm, Hm+1, . . . , Hm+M−, obtain a single 3D rendered object frame Fm, to thereby create a train of rendered three-dimensional image frames, each 3D rendered object frame Fm formed from a moving window of said captured 2D image sets Hm, Hm+1, . . . , Hm+M+1, wherein M is at least three and not greater than N, such that:
- successive rendered three-dimensional image frames are obtained from overlapping sets of M adjacent captured 2D image sets.
Dependencies are inferred from the claim text.