3D Sensor Template Matching for Depth Map Distortion

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Optoelectronic 3D sensors face challenges in reliably detecting objects of minimum size within predetermined detection fields due to projective perspective distortions and shadows, leading to incomplete or inaccurate depth maps, which affects safety and availability in security and automation technologies.

Innovation Solution

The use of a configuration unit to create templates that account for projective perspective distortions and shadows, allowing for efficient object recognition by comparing these templates with depth map sections, ensuring accurate detection of objects of minimum size across the detection field.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional object detection methods are used in 3D sensors, then the detection process is simple, but detection accuracy decreases due to projective perspective distortions and shadows

Engineering Contradiction:
Improvedetection accuracyVSAvoidevaluation complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent creates detection templates in advance that account for projective perspective distortions and shadows. These templates are generated based on the geometric relationship between the sensor and detection fields, allowing the system to compensate for distortions before actual object detection occurs, thereby improving accuracy without adding complex real-time processing

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent creates virtual copies of detection fields in the form of templates that represent how objects of minimum size would appear in the depth map. By comparing actual depth map data against these pre-generated templates, the system can accurately detect objects despite projective distortions and shadows without requiring complex geometric calculations during detection

Inventive Principle:
Principle #26Copying

2Reliability

If conservative security approach is chosen to ensure safety, then detection reliability improves, but availability decreases due to frequent false alarms and shutdowns

Engineering Contradiction:
ImprovesafetyVSAvoidavailability
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent replaces traditional geometric calculation methods with template matching. Instead of performing complex real-time geometric transformations and comparisons, the system uses pre-generated templates that encode the expected appearance of minimum-size objects. This substitution simplifies the detection mechanism while maintaining high detection accuracy, reducing false alarms and improving availability without compromising safety

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

3Measurement precision

If high resolution depth mapping is used to improve object detection, then detection precision improves, but data completeness decreases due to shadows and projective distortions

Engineering Contradiction:
Improvedepth map resolutionVSAvoiddepth map completeness
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The patent introduces templates as an intermediary between the depth map data and object detection logic. These templates serve as reference patterns that account for projective distortions and shadows, allowing the system to interpret depth map data more accurately even when complete geometric information is not available. The templates bridge the gap between incomplete depth data and reliable object detection

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP2819109B1Optoelectronic 3D-sensor and method for recognising objects
Publication Date: 2015.05.27 SICK AG
  • EP2819109B1 patent drawingFigure 1
  • EP2819109B1 patent drawingFigure 2~3
  • EP2819109B1 patent drawingFigure 4~5b

AI summary

An optoelectronic 3D sensor (10) for detecting objects (42) above a minimum size within at least one predefined detection field (32) is described. The 3D sensor (10) comprises a 3D image sensor (14a-b, 26) for acquiring a depth map, a configuration unit (28) for defining detection fields (32), and an object recognition unit (30) for evaluating the depth map to detect object intrusions into the detection field (32). The object recognition unit (30) is configured to define a viewing direction (46) to be checked, to generate a template (48) suitable for the viewing direction and minimum size from the detection fields (32), and to compare the template (48) with a neighborhood derived from the depth map in the viewing direction (46).