Method and system for fast three-dimensional imaging using defocusing and feature recognition

• US 8,456,645 B2
• Filed: 12/01/2011
• Issued: 06/04/2013
• Est. Priority Date: 01/22/2007
• Status: Active Grant

• Current Assignee: California Institute of Technology
• Original Assignee: California Institute of Technology
• Inventors: Gharib, Morteza, Graff, Emilio, Pereira, Francisco
• Primary Examiner(s): NGUYEN, SANG H
• Application Number:
  US13/309,482
• Publication Number:
  US 20120076400A1
• Time in Patent Office: 551 Days
• Field of Search: 356600-601, 356/614, 356/620, 356/622, 356/625, 382/255, 382/274, 382/275, 382/278, 382/284, 382/285, 382/203, 382/173, 348/46, 348/335, 348/208.14
• US Class Current: 356/601
• CPC Class Codes: G01B 11/24 for measuring contours or curvatures

Abstract

Described is a method and system for fast three-dimensional imaging using defocusing and feature recognition is disclosed. The method comprises acts of capturing a plurality of defocused images of an object on a sensor, identifying segments of interest in each of the plurality of images using a feature recognition algorithm, and matching the segments with three-dimensional coordinates according to the positions of the images of the segments on the sensor to produce a three-dimensional position of each segment of interest. The disclosed imaging method is “aware” in that it uses a priori knowledge of a small number of object features to reduce computation time as compared with “dumb” methods known in the art which exhaustively calculate positions of a large number of marker points.

Claims

1. A method for fast three-dimensional imaging using defocusing and feature recognition in a fast three-dimensional imaging data processing system having a sensor and a processor, comprising:

capturing a plurality of defocused images of an object on the sensor by receiving signals transmitted from the object through a plurality of apertures of a mask and through a lens, where the data processing system understands the type of object being imaged;

using the data processing system to identify a priori known segments of interest in each of the plurality of defocused images with a feature recognition algorithm; and

using the data processing system to match the identified a priori known segments of interest with three-dimensional coordinates according to the positions of the images of the segments on the sensor to produce an output image containing the three-dimensional position of each segment of interest.

2. The method of claim 1, further comprising ascertaining sub-pixel positions of the a priori known segments of interest using a cross-correlation type algorithm.

3. A data processing system for fast three-dimensional imaging using defocusing and feature recognition, comprising:

a mask having a plurality of apertures;

a lens;

a sensor configured to capture a plurality of defocused images of an object substantially simultaneously, by receipt of signals transmitted from the object through the plurality of apertures and through the lens, wherein the data processing system understands the type of object being imaged; and

one or more processors configured to identify a priori known segments of interest in each of the plurality of defocused images with a feature recognition algorithm, and configured to match the identified a priori known segments of interest with three-dimensional coordinates, according to the positions of the images of the segments of interest on the sensor, to produce an output image containing the three-dimensional position of each segment of interest.

4. The data processing system of claim 3, wherein the one or more processors are further configured to ascertain sub-pixel positions of the a priori known segments of interest on the sensor using a cross-correlation type algorithm.