Depth-Filter Overhang Detection for Complete 3D Reconstruction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current 3D object reconstruction methods are slow due to inadequate view planning in Next-Best-View (NBV) processes, particularly on image-capturing devices with limited processing power, leading to incomplete surface reconstructions from occlusions caused by overhangs.

Innovation Solution

An overhang detection method using three filters based on depth gradients and salient object detection to automatically identify overhangs, guiding users to capture images from hidden surfaces, implemented on low-power devices like mobile phones.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If Next-Best-View (NBV) iterative reconstruction is used to ensure complete surface capture, then reconstruction completeness is improved, but processing time and energy consumption increase significantly

Engineering Contradiction:
Improvereconstruction completenessVSAvoidprocessing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent applies preliminary action by detecting overhangs in advance using depth gradient filters before the reconstruction process begins. This allows the system to pre-identify regions that require additional views, eliminating the need for iterative NBV reconstruction and significantly reducing processing time while ensuring complete surface capture.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent extracts the overhang detection function as a separate preprocessing step using three filters (depth gradient filter, salient object detection, and overhang classification). This extraction allows the main reconstruction process to operate efficiently on already-identified critical regions, reducing overall processing time while maintaining completeness.

Inventive Principle:
Principle #2Taking out (Extraction)

2Reliability

If Next-Best-View (NBV) iterative reconstruction is used to capture occluded surfaces, then surface coverage is improved, but energy consumption increases due to repeated reconstruction cycles

Engineering Contradiction:
Improvesurface coverageVSAvoidenergy consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary overhang detection using lightweight depth gradient filters and salient object detection before reconstruction. This preliminary action identifies all occluded regions in advance, allowing the reconstruction to proceed in a single pass without energy-intensive iterative cycles, thus reducing energy consumption while maintaining complete surface coverage.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent replaces the mechanical iterative reconstruction process with an automated overhang detection system using computer vision filters. This substitution eliminates repeated reconstruction cycles by pre-identifying all occluded surfaces, significantly reducing energy consumption on mobile devices while maintaining complete surface coverage.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Adaptability or versatility

If manual view selection by humans is used for 3D scanning, then flexibility in handling different object shapes is improved, but scanning efficiency and consistency deteriorate

Engineering Contradiction:
Improveflexibility for different object shapesVSAvoidscanning efficiency
Core Design Contradiction:
Adaptability or versatilityVSProductivity

Solution Approach 1:

The patent implements self-service by enabling the system to automatically detect overhangs and determine required viewing angles using depth gradient filters and salient object detection. This automated approach maintains flexibility for different object shapes while dramatically improving scanning efficiency and consistency, eliminating dependence on manual view selection.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent introduces an intermediary overhang detection system that acts as a mediator between the object and the reconstruction process. This intermediary automatically analyzes object geometry using filters and provides optimized view planning, maintaining adaptability to various object shapes while improving scanning efficiency and consistency without manual intervention.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Ease of operation

If mobile phones with limited processing power are used for 3D scanning, then portability and accessibility are improved, but processing speed and reconstruction quality deteriorate

Engineering Contradiction:
Improveportability and accessibilityVSAvoidprocessing speed
Core Design Contradiction:
Ease of operationVSProductivity

Solution Approach 1:

The patent extracts the computationally intensive overhang detection as a separate preprocessing step using efficient filters (depth gradient, salient object detection). This extraction enables mobile phones with limited processing power to perform 3D scanning by breaking down the complex reconstruction into manageable stages, improving processing speed while maintaining reconstruction quality.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The patent replaces heavy iterative reconstruction algorithms with lightweight overhang detection filters that are computationally efficient for mobile devices. This substitution maintains portability and accessibility while significantly improving processing speed and reconstruction quality on devices with limited processing power.

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS12354215B1Overhang detection for use in three-dimensional object reconstruction
Publication Date: 2025.07.08 AMAZON TECH INC
  • US12354215B1 patent drawing
  • US12354215B1 patent drawing
  • US12354215B1 patent drawing

AI summary

A method is disclosed to automatically detect overhangs from images with depth taken around an object during a scan of the object. An overhang detector can use an intersection of three filters based on these images and depth data associated with the images. The first filter looks for negative depth gradients along a 2D projection of a gravity vector, which is generally a vertical axis for images taken using a portrait orientation. The second filter selects the depth gradients that are oriented towards the projection of the gravity vector. The third filter is a salient object detection mask computed from the image. An intersection of the three filters can then be used to obtain overhangs. The method can be implemented in real time with a User Interface (UI) directing a user of a location of the overhang so that an image below the overhang can be taken.