Automatic Conversion from Monoscopic Video to Stereoscopic Video

• US 20080278569A1
• Filed: 01/12/2006
• Published: 11/13/2008
• Est. Priority Date: 01/14/2005
• Status: Active Grant

• Current Assignee: Rafael - Armament Development Authority Ltd.
• Original Assignee: Rafael - Armament Development Authority Ltd.
• Inventors: Rotem, Efrat, Pelz, David, Wolowelsky, Karni
• Granted Patent: US 8,063,930 B2
• Time in Patent Office: Days
• US Class Current: 348/43
• CPC Class Codes: H04N 13/221 using the relative movement between cameras and objects
• CPC Class Codes: H04N 13/261 with monoscopic-to-stereoscopic image conversion
• CPC Class Codes: H04N 13/275 from 3D object models, e.g. computer-generated stereoscopic image signals

Abstract

The invention is a method for converting a monoscopic video movie passively acquired using a single camera to a stereoscopic video movie. The method comprises generating stereoscopic pairs comprising, for each frame in the original sequence, a transformed original frame selected from the monoscopic video movie and a transformed alternate frame. The transformed frames are generated from the original and the alternate frame.

Claims

1. A method for converting a monoscopic video movie passively acquired using a single camera to a stereoscopic video movie, said method comprising generating stereoscopic pairs comprising, for each frame in the original sequence, a transformed original frame selected from said monoscopic video movie and a transformed alternate frame;

wherein said transformed frames are generated from said original and said alternate frame in said monoscopic video movie by the following steps;

a.) select said original frame;

b.) select said alternate frame such that a considerable, approximately constant parallax is developed between it and said original frame;

c.) calculate the fundamental matrix between said original frame and said alternate frame based on point correspondences between them;

d.) calculate from said fundamental matrix two homography transformations that may stratified said original and said alternate frames;

e.) apply one of said homography transformations to said original frame and the other of said homography transformations to said alternate frame, thereby obtaining two new frames, which are stereoscopic partners.

2. A method according to claim 1, wherein said method is carried out using a PC based real time system.

3. A method according to claim 1, wherein the point correspondences between the original and the alternate frames are created using an automatic process based on a known correspondence algorithm.

4. A method according to claim 3, wherein the known correspondence algorithm is selected from the group comprising:

a.) optical flow; and

b.) feature based registration.

5. A method according to claim 1, wherein additional information may be used to calculate the relative camera'"'"'s states in steps c) and d) either:

a.) together with the use of the fundamental matrix or

b.) instead of use of the fundamental matrix

6. A method according to claim 5, wherein the additional information is GPS information or other information from a navigational system.

7. A method according to claim 1, wherein in steps c) and d) the fundamental matrix is replaced by any other matrix that is capable of estimating the relative position of the two cameras that produced the pair of images.

8. A stereoscopic video movie created from a monoscopic video movie using the method of claim 1, wherein the sequence of the images in said monoscopic video movie is preserved in the sequence of stereo pairs in said stereoscopic video movie.