3D Head Mesh Alignment Using Iterative Closest Point

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for generating three-dimensional computer-generated imagery of human bodies from two-dimensional images are inefficient, as they fail to accurately align polygonal meshes and generate high-quality textures, leading to unsuitable models for interactive video games and other applications.

Innovation Solution

A processing workflow that acquires images of a person's head from various positions, creates a polygonal mesh, aligns it with a base mesh using iterative closest point and simulated annealing methods, and generates textures such as light maps and color textures, ensuring compatibility with target application topology and resolution.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If current methods are used to generate three-dimensional computer-generated imagery from two-dimensional images, then the process is simpler and faster, but the accuracy of polygonal mesh alignment and texture quality deteriorate

Engineering Contradiction:
Improvepolygonal mesh alignment accuracyVSAvoidprocessing workflow complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The processing workflow is segmented into distinct modules: image acquisition module, polygonal mesh generation module, mesh alignment module (using iterative closest point and simulated annealing), and texture generation module. Each module handles a specific task, improving overall precision while managing complexity through functional decomposition.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The base polygonal mesh is prepared in advance with predefined topology and resolution parameters before the actual processing begins. This preliminary preparation enables more accurate and efficient alignment operations during the main processing phase, as the target structure is already established.

Inventive Principle:
Principle #10Preliminary action

2Measurement precision

If advanced alignment methods like iterative closest point and simulated annealing are used, then polygonal mesh alignment accuracy improves, but processing time increases

Engineering Contradiction:
Improvemesh alignment precisionVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The iterative closest point method employs periodic iterations to progressively refine mesh alignment, while simulated annealing uses periodic temperature reduction to escape local minima and converge to optimal alignment. These periodic actions balance computational effort with alignment precision.

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

A rough initial alignment is performed before applying the computationally intensive iterative closest point and simulated annealing methods. This preliminary positioning reduces the search space and accelerates convergence of the advanced alignment algorithms.

Inventive Principle:
Principle #10Preliminary action

3Manufacturing precision

If high-resolution textures are generated for each polygon, then visual quality improves, but computational resources and processing time increase

Engineering Contradiction:
Improvetexture qualityVSAvoidprocessing efficiency
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

Different texture resolutions are applied to different regions of the polygonal mesh based on their visual importance. Areas requiring higher visual fidelity (such as facial features) receive high-resolution textures, while less critical areas use lower resolution textures, optimizing the balance between quality and processing efficiency.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The texture generation process dynamically adjusts parameters such as resolution, sampling density, and filtering levels based on the specific requirements of each polygon and its position in the scene. This adaptive parameter adjustment maintains high visual quality where needed while reducing computational load elsewhere.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentUS10403001B2Producing three-dimensional representation based on images of an object
Publication Date: 2019.09.03 ELECTRONIC ARTS INC
  • US10403001B2 patent drawing
  • US10403001B2 patent drawing
  • US10403001B2 patent drawing

AI summary

An example method of generating three-dimensional objects based on two-dimensional images of at least part of the object may include: receiving, from a mobile computing device, a notification indicating that a plurality of images depicting an object have been uploaded to a first storage location; transmitting, to a rendering application, a rendering request specifying the first storage location; receiving, from the rendering application, a rendering response indicating that a polygonal mesh representing the images of the object has been stored at a second storage location; transmitting, to an interactive video gaming device, a notification of comprising an identifier of the third second storage location; and making a file comprising the polygonal mesh available for download by the interactive video gaming device.