Method and device for the generation of 3-D images

• US 7,321,374 B2
• Filed: 05/29/2003
• Issued: 01/22/2008
• Est. Priority Date: 11/24/2001
• Status: Expired due to Fees

• Current Assignee: Phoenix 3D, Inc.
• Original Assignee: NewSight Corp. (Hong Kong Winalite Group, Inc.)
• Inventors: Naske, Rolf-Dieter
• Primary Examiner(s): Brier, Jeffery A
• Application Number:
  US10/447,463
• Publication Number:
  US 20040027346A1
• Time in Patent Office: 1,699 Days
• Field of Search: 345/647, 345/648, 345/649, 345/672
• US Class Current: 345/647
• CPC Class Codes: H04N 13/261 with monoscopic-to-stereoscopic image conversion

Abstract

A method for generating three-dimensional images based on a sequence of two-dimensional images is characterized by the steps of: analyzing a two-dimensional image with respect to its scene type, selecting a deformation assigned to the defined scene type with which the two-dimensional image is deformed, deforming the two-dimensional image and transmitting the deformed two-dimensional image to at least one viewing channel. The method also contemplates applying various transition functions to the two-dimensional image to provide a continuous and smooth transition from one scene type to another. The method can be carried out by an apparatus for performing these functions.

Claims

1. An apparatus for implementing a method for generating three-dimensional (3-D) images on a sequence of two-dimensional (2-D) images, the apparatus including a device for scene analysis that performs the following steps:

analyzing a two-dimensional image to determine its scene type;

assigning a deformation to the determined scene type;

deforming the two-dimensional image using the deformation assigned to the determined scene type; and

transmitting the deformed two-dimensional image to at least one viewing channel;

wherein the analyzing step includes defining a partial image with variable size near the center of the whole two-dimensional image and calculating a root mean square deviation of the partial image and of the whole two-dimensional image in order to define a scene type as a close-up or a normal shot based on this calculation.

2. An apparatus according to claim 1, including a synthesis device for interpolating a predetermined number of sequential images from at least one input 2-D image to produce at least a portion of the image sequences.

3. An apparatus for implementing a method for generating three-dimensional (3-D) images based on a sequence of two-dimensional (2D) images, the apparatus including a device for scene analysis that performs the following steps:

analyzing a two-dimensional image to determine its scene type;

assigning a deformation to the determined scene type;

deforming the two-dimensional image using the deformation assigned to the determined scene type; and

transmitting the deformed two-dimensional image to at least one viewing channel;

wherein the analyzing step includes defining a plurality of partial images near the edge of the whole two-dimensional image and calculating an absolute amount of the cross-correlation between different areas of the whole two-dimensional image in order to define a scene type as a wide angle shot based on this calculation.

4. An apparatus according to claim 3, including a synthesis device for interpolating a predetermined number of sequential images from at least one input 2-D image to produce at least a portion of the image sequences.

5. A method for generating three-dimensional (3-D) images based on a sequence of two-dimensional (2-D) images, comprising the steps of:

analyzing a two-dimensional image to determine its scene type;

assigning a deformation to the determined scene type;

deforming the two-dimensional image using the deformation assigned to the determined scene type; and

transmitting the deformed two-dimensional image to at least one viewing channel,

wherein the analyzing step includes defining a partial image with variable size near a central portion of the whole two-dimensional image and calculating a root mean square deviation of the partial image and of the whole two-dimensional image, and determining the scene type as a close-up shot or a normal shot based on this calculation.

6. A method according to claim 5, further comprising interpolating a predetermined number of sequential images from at least one two-dimensional image to produce at least a portion of the sequence of two-dimensional images.

7. A method for generating three-dimensional (3-D) images based on a sequence of two-dimensional (2-D) images, comprising the steps of:

analyzing a two-dimensional image to determine its scene type;

assigned a deformation to the determined scene type;

deforming the two-dimensional image using the deformation assigned to the determined scene type; and

transmitting the deformed two-dimensional image to at least one viewing channel,

wherein the analyzing step includes defining a plurality of partial images near an edge of two-dimensional image and calculating an absolute amount of cross-correlation between different partial images and determining a scene type as a wide-angle shot based on this calculation.

8. A method according to claim 7, further comprising interpolating a predetermined number of sequential images from at least one two-dimensional image to produce at least a portion of the sequence of two-dimensional images.

9. A method for generating three-dimensional (3-D) images based on a sequence of two-dimensional (2-D) images, comprising the steps of:

analyzing a two-dimensional image to determine its scene type from a plurality of scene types;

assigning a deformation to each scene type;

deforming the two-dimensional image using the deformation assigned to the determined scene type; and

transmitting the deformed two-dimensional image to at least one viewing channel, and repeating the analyzing, deforming and transmitting steps;

wherein the assigning step includes assigning a transition deformation for a transition between every pair of scene types; and

when the scene type of a current image is different than that of a previous image, the deforming step includes applying the assigned transition deformation to at least the current image to adjust the deformation from the deformation asigned to the previous scene type to the deformation assigned to the current scene type to prevent or minimize an unnatural appearance of the image.

10. A method according to claim 9, wherein applying the transition deformation includes applying a predetermined number of transition deformations to successive images, starting with the current image, whereby the transition deformations are calculated by interpolation between the deformation for the previous scene type and the deformation for the current scene type.

11. A method according to claim 9, further comprising interpolating a predetermined number of sequential images from at least one two-dimensional image to produce at least a portion of the sequence of two-dimensional images.