3D Mesh Sculpting via Ray Casting for Depth Camera Noise Reduction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for generating 3D models using depth cameras, such as point cloud alignment via Iterative Closest Point (ICP), are prone to noise and errors, resulting in unsatisfactory depth information alignment and 3D model generation.

Innovation Solution

A method involving a computing device that captures images from a depth camera, generates an initial sculpture as a mesh polygon, performs a ray casting process, and deforms the sculpture based on depth information to create a 3D model, incorporating techniques like vertex force calculation, tessellation, and edge handling to improve model accuracy and detail.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If point cloud alignment via ICP is used for generating 3D models, then the process is simple and fast, but the alignment accuracy and model quality deteriorate due to noise and errors

Engineering Contradiction:
Improve3D model generation speedVSAvoiddepth information alignment accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The patent segments the 3D model generation process into distinct stages: creating an initial sculpture from depth data, performing ray casting to identify intersection points, calculating forces at these points, and progressively deforming the mesh. This segmentation allows each stage to be optimized independently, improving overall accuracy without sacrificing too much productivity.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions by first creating an initial sculpture and performing ray casting to identify intersection points before applying forces and deforming the mesh. This preliminary setup ensures that subsequent deformation operations are based on accurate geometric relationships, improving alignment accuracy before the main modeling process begins.

Inventive Principle:
Principle #10Preliminary action

2Manufacturing precision

If traditional mesh deformation methods are used, then the process is computationally efficient, but the model detail and accuracy deteriorate

Engineering Contradiction:
Improve3D model accuracyVSAvoidprocessing complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent applies local quality by calculating forces at specific intersection points where rays meet the mesh, rather than applying uniform deformation across the entire model. This localized approach concentrates computational effort where it is most needed (at intersection points) while maintaining lower overall complexity compared to global deformation methods.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent changes parameters by introducing force magnitudes and directions at intersection points, then progressively deforming the mesh based on these parameter changes. This allows precise control over deformation behavior, improving model accuracy while maintaining computational efficiency through parameter-based control rather than complex geometric calculations.

Inventive Principle:
Principle #35Parameter changes

3Manufacturing precision

If force vectors are calculated for all vertices, then the model deformation is comprehensive, but the computational cost and processing time increase

Engineering Contradiction:
Improvevertex deformation accuracyVSAvoidcomputational energy consumption
Core Design Contradiction:
Manufacturing precisionVSUse of energy by moving object

Solution Approach 1:

The patent extracts and applies forces only at specific intersection points where rays meet the mesh, rather than calculating forces for all vertices. This extraction approach focuses computational energy on critical points that most influence the final model shape, achieving comprehensive deformation accuracy with reduced computational cost.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent applies partial action by calculating forces only at intersection points rather than all vertices. This partial approach is sufficient to achieve accurate model deformation because intersection points represent the critical geometric relationships between the depth data and the mesh, without wasting computational energy on less important vertices.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3155596B13D scanning with depth cameras using mesh sculpting
Publication Date: 2020.03.11 SONY GROUP CORP
  • EP3155596B1 patent drawingFigure 1A~1B
  • EP3155596B1 patent drawingFigure 2A~2B
  • EP3155596B1 patent drawingFigure 3

AI summary

A method, performed by a mobile communication device, may include capturing one or more images from a depth camera, wherein the one or more images include depth information. The method may further include generating an initial sculpture for an entity in the captured one or more images; performing a ray casting process on the generated initial sculpture using the depth information; and deforming the initial sculpture based on the ray casting process to generate a sculpture for the entity.