3D User Model Generation via Angle-Based Image Selection

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current computer systems lack the ability to accurately generate three-dimensional models of users for virtual try-on applications, as existing methods are not robust enough to handle variations in pose and lighting, leading to inaccurate depictions of how products would look on an individual.

Innovation Solution

A computer-implemented method that calculates the angle of view in images of a user, selects images matching predetermined viewing angles, and uses cross-correlation algorithms to generate a 3-D model, combining texture coordinate information and geometry files to create a realistic polygon mesh of the user, allowing for accurate virtual try-on of products.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If existing methods are used to generate 3-D models, then the process is simpler, but the accuracy and realism of the user depiction deteriorates

Engineering Contradiction:
Improveaccuracy of 3-D modelVSAvoidcomplexity of image processing system
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the image processing into multiple distinct stages: image acquisition with controlled viewing angles, feature detection using cross-correlation algorithms, 3-D model generation, and texture mapping. This segmentation allows each stage to be optimized independently, achieving high measurement precision through systematic multi-stage processing while managing overall system complexity through modular organization of processing steps.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If multiple images are processed to generate accurate 3-D models, then the model accuracy improves, but the processing time increases

Engineering Contradiction:
Improveaccuracy of user depictionVSAvoidtime to generate 3-D model
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by pre-defining a set of predetermined viewing angles before image acquisition. During image capture, the system selectively acquires images only at these predetermined angles, rather than processing all possible angles. This preliminary preparation of angle specifications enables efficient selection and processing of only the necessary images, reducing processing time while maintaining accurate 3-D model generation through multi-angle data.

Inventive Principle:
Principle #10Preliminary action

3Manufacturing precision

If images at predetermined viewing angles are selected, then the 3-D model accuracy improves, but the image selection process becomes more complex

Engineering Contradiction:
Improveprecision of 3-D structureVSAvoidcomplexity of angle calculation system
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent applies parameter changes by establishing a discrete set of predetermined viewing angle parameters (specific orientations and rotations) that optimize 3-D model reconstruction. Instead of continuously varying angles, the system uses specific angular parameters that have been predetermined to provide the most informative views for 3-D reconstruction. This parameterization simplifies the angle selection process while ensuring high precision in the resulting 3-D structure through the strategic choice of viewing angles.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP2852935B1Systems and methods for generating a 3-d model of a user for a virtual try-on product
Publication Date: 2020.08.19 LUXOTTICA OF AMERICA INC
  • EP2852935B1 patent drawingFigure 1
  • EP2852935B1 patent drawingFigure 2
  • EP2852935B1 patent drawingFigure 3

AI summary

A computer-implemented method for generating a three-dimensional (3-D) model of a user. A plurality of images of a user are obtained. An angle of view relative to the user pictured in at least one of the plurality of images is calculated. It is determined whether the calculated angle of view matches a predetermined viewing angle. Upon determining the calculated angle of view matches the predetermined viewing angle, at least one of the plurality of images is selected.