Feature-to-Depth Mapping for 2D to 3D Image Conversion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional methods for converting two-dimensional (2D) images and video to stereoscopic 3D are time-consuming, costly, and not scalable for large volumes, with limited applicability to general types of content.

Innovation Solution

An image converter uses a feature-to-depth mapping function based on an approximate depth map from a visually similar 3D image to determine depth values for each pixel in a 2D image, enabling automatic conversion of 2D visual content to 3D, leveraging a large database of images for accurate prediction and realistic rendering.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If conventional manual methods are used to convert 2D images to 3D, then the conversion quality can be controlled, but the process is time-consuming and not scalable

Engineering Contradiction:
Improveconversion speedVSAvoidmanual processing time
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The system pre-computes and stores depth maps for example images in a database before actual conversion is needed. When converting a 2D image to 3D, the system retrieves pre-computed depth information from the database rather than calculating it in real-time, dramatically reducing conversion time while maintaining quality

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates a database of example 3D images with their depth maps stored beforehand. These pre-computed depth maps are copied and reused for similar images, avoiding redundant computation and enabling rapid conversion of multiple images without manual intervention for each one

Inventive Principle:
Principle #26Copying

2Adaptability or versatility

If conventional techniques are used for 2D to 3D conversion, then specific types of images can be converted, but the method is limited to certain image types and not generalizable

Engineering Contradiction:
Improveimage type compatibilityVSAvoidconversion accuracy
Core Design Contradiction:
Adaptability or versatilityVSReliability

Solution Approach 1:

The system uses a universal feature-to-depth mapping approach that works across different image types. By extracting features like edges, corners, and textures and mapping them to depth values using learned relationships from example images, the system can convert various types of 2D images (landscapes, portraits, objects) to 3D without requiring type-specific algorithms

Inventive Principle:
Principle #6Universality (Multi-functionality)

Solution Approach 2:

The system introduces an intermediate feature extraction stage that converts diverse image types into a common feature representation. This intermediate feature space serves as a mediator that allows different image types to be processed through the same depth mapping mechanism, enabling generalization while maintaining accuracy

Inventive Principle:
Principle #24Intermediary (Mediator)

3Manufacturing precision

If manual 3D content creation is used, then accurate 3D effects can be achieved, but the process is costly and resource-intensive

Engineering Contradiction:
Improve3D effect accuracyVSAvoidprocessing system complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The system automatically extracts features from 2D images and computes depth maps using pre-trained mapping functions without requiring manual 3D modeling or expert intervention. The automated feature-to-depth mapping process achieves accurate 3D effects while eliminating the need for complex manual processing workflows and specialized equipment

Inventive Principle:
Principle #25Self-service

Data Source

PatentEP2774124B1Depth-map generation for an input image using an example approximate depth-map associated with an example similar image
Publication Date: 2020.01.08 GOOGLE LLC
  • EP2774124B1 patent drawingFigure 1
  • EP2774124B1 patent drawingFigure 2A
  • EP2774124B1 patent drawingFigure 2B

AI summary

An image converter receives a two-dimensional image to be converted to a first three- dimensional image. The image converter computes a feature-to-depth mapping function based on an approximate depth map associated with a second three-dimensional image. The image converter applies the feature-to-depth mapping function to a plurality of pixels of the two- dimensional image to determine a depth value for each of the plurality of pixels and generates the first three-dimensional image based on the depth values for the plurality of pixels of the two- dimensional image.