PTAB

IPR2014-01341

Google Inc v. Visual Real Estate Inc

Key Events

Petition

petition Intelligence

Title: Methods For and Apparatus For Generating a Continuum of Three Dimensional Image Data
Brief Description: The ’396 patent relates to methods for creating a "continuum of image data" by capturing multiple 2D image data sets of a subject from different points along a path, tracking features within the images to generate a point cloud array, converting the point cloud into a 3D polygon-based model, and associating the model with location data.

Prior Art Relied Upon: Fruh ("An Automated Method for Large-Scale, Ground-Based City Model Acquisition," a 2004 journal article).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner argued that Fruh discloses every limitation of the challenged claims. Fruh describes a vehicle-mounted system with a digital camera and 2D laser scanners that continuously captures geometry and texture data to generate large-scale 3D city models. This system inherently creates a "continuum of image data" by capturing images from disparate points along the vehicle's path. Fruh explicitly teaches generating a 3D point cloud from the laser scans, transforming the point cloud into a triangular mesh (a polygon-based model), and mapping image texture onto the model. The system also uses GPS to determine the vehicle's global pose, thereby receiving and associating location data (including Cartesian and latitude/longitude coordinates) with the generated models.

Prior Art Relied Upon: Fruh (a 2004 journal article) and Di Bernardo (Patent 6,895,126).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner contended that Fruh teaches the core system for 3D model generation from vehicle-based sensors. To the extent any claim elements are not expressly disclosed by Fruh, Di Bernardo supplies them. Specifically, Di Bernardo teaches creating a composite image by extracting and combining vertical columns of pixels from sequential video frames captured along a street. Di Bernardo also discloses overlaying metadata (e.g., icons linked to business information) on the composite image, satisfying the limitation of claim 16.
- Motivation to Combine: A person of ordinary skill in the art (POSITA) would combine Di Bernardo's teachings with Fruh's system to provide useful context to the 3D models by overlaying metadata. A POSITA would also recognize that generating a composite image as taught by Di Bernardo was a known technique for visualizing image data captured along a path and would have been motivated to apply it to a similar system like Fruh's to achieve predictable results.
- Expectation of Success: The combination involved applying known image processing techniques to an existing 3D modeling system, which would have presented a high expectation of success.

Prior Art Relied Upon: Akbarzadeh ("Towards Urban 3D Reconstruction From Video," a 2006 journal article).
Core Argument for this Ground:
- Prior Art Mapping: Petitioner asserted that Akbarzadeh discloses a method for the automatic 3D reconstruction of urban scenes from multiple video streams captured by a vehicle-mounted, multi-camera system. The system generates geo-registered 3D models by creating depth maps for video frames (which function as a point cloud), fusing the maps to ensure consistency, and generating a final triangulated, texture-mapped 3D model (a polygon-based model). The system uses GPS and INS data to track features, determine camera poses, and geo-register the final models, thus disclosing the claimed steps of receiving location data and associating it with the polygon model.
Additional Grounds: Petitioner asserted additional obviousness challenges based on Fruh alone, Akbarzadeh alone, and the combination of Akbarzadeh and Di Bernardo, relying on similar arguments and technical disclosures.

"a continuum of image data": Petitioner proposed this term means "two or more images captured along a path," arguing this construction is consistent with the specification's description of capturing image data from disparate points along a continuum, such as the path of a vehicle.
"point cloud array": Petitioner proposed this term means "a set of points corresponding to locations of features." This construction was argued to align with the patent’s disclosure that point clouds are generated using the location of feature points present in multiple image sets.
"polygon based model": Petitioner proposed this term means "a model representing the 3D structure of objects using polygons," consistent with the specification’s teaching that point clouds are processed to generate three-dimensional polygon-based models.

Priority Date Challenge: A central contention was that the ’396 patent is not entitled to the earlier filing dates of its parent applications (’465, ’707, and ’708 applications). Petitioner argued these prior applications lack written description support for the key claim limitations of "generating a point cloud array" and "converting the point cloud array to a polygon based model." Therefore, Petitioner asserted the challenged claims are entitled only to the patent’s actual filing date of February 21, 2008, making Fruh (2004) and Akbarzadeh (2006) valid §102 prior art references.

Petitioner requests institution of an inter partes review (IPR) and cancellation of claims 1, 2, 5, 8-13, and 16 of the ’396 patent as unpatentable.