Object Detection Device Using Dual Resolution U-Maps

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional object detection systems using stereo cameras face instability in detecting distant objects due to insufficient information from captured images, leading to unreliable detection.

Innovation Solution

The system generates and utilizes both high-resolution and low-resolution U-maps from parallax images to stabilize distant object detection, with the high-resolution map for close objects and low-resolution map for distant objects, allowing for accurate detection by merging data from both maps based on distance thresholds.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a single high-resolution U-map is used for object detection, then close objects can be detected with high precision, but distant objects become unstable due to sparse parallax value distribution

Engineering Contradiction:
Improvedetection precisionVSAvoiddetection stability
Core Design Contradiction:
Measurement precisionVSReliability

Solution Approach 1:

The patent segments the detection space into two distinct zones: a first detection zone for close objects using a first U-map with first resolution, and a second detection zone for distant objects using a second U-map with second resolution. This segmentation allows each zone to use the appropriate resolution level, ensuring both close and distant objects are detected with suitable precision and stability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent applies local quality by assigning different resolution characteristics to different spatial regions. The first U-map maintains higher resolution for the first detection zone where close objects require precise detection, while the second U-map uses appropriate resolution for the second detection zone where distant objects benefit from denser parallax value distribution. Each region's detection quality is optimized according to its specific requirements.

Inventive Principle:
Principle #3Local quality

2Reliability

If a single low-resolution U-map is used for object detection, then distant objects can be detected stably, but close objects lose detection precision

Engineering Contradiction:
Improvedetection stabilityVSAvoiddetection precision
Core Design Contradiction:
ReliabilityVSMeasurement precision

Solution Approach 1:

The detection system is segmented into two zones with different resolution requirements. The second detection zone for distant objects uses the second U-map with second resolution to achieve stable detection through denser parallax distribution, while the first detection zone for close objects uses the first U-map to maintain high precision.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Different resolution qualities are applied locally to different detection zones. The second U-map provides appropriate resolution for distant objects where stability is the priority, while the first U-map provides higher resolution for close objects where precision is the priority. Each zone receives the quality level appropriate to its detection needs.

Inventive Principle:
Principle #3Local quality

3Measurement precision

If multiple U-maps with different resolutions are generated and merged, then both close and distant objects can be detected accurately, but system complexity increases

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

Solution Approach 1:

The system segments the detection task into two parallel processing streams: one for generating and processing the first U-map for close objects, and another for generating and processing the second U-map for distant objects. This segmentation allows independent optimization of each stream while maintaining overall system organization and manageability.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent merges the detection results from the first U-map and second U-map by determining object positions in both detection zones and combining these results. This merging approach consolidates the benefits of both high-resolution close object detection and stable distant object detection into a unified detection system, achieving comprehensive accuracy without requiring a completely new complex system.

Inventive Principle:
Principle #5Merging (Combining)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach enables stable detection of distant objects by preventing the sparse distribution of parallax values, ensuring accurate recognition and control of vehicles or other moving objects, improving safety features like automatic braking.

Implementation Method 1

a parallax image generator to generate a parallax image from a captured image of a stereo camera

Methodology Applied
Scientific EffectParallax: Parallax

Implementation Method 2

information obtained from a captured image

Methodology Applied
Scientific EffectLight reflection: Reflection

Data Source

PatentEP3413267B1Object detection device, device control system, objection detection method, and program
Publication Date: 2023.06.28 RICOH CO LTD
  • EP3413267B1 patent drawingFigure 1
  • EP3413267B1 patent drawingFigure 2
  • EP3413267B1 patent drawingFigure 3A~3B

AI summary

An object detection device according to the present invention includes a map generator configured to generate two types of maps having different resolutions based on a distance image that includes distance information for each coordinate; and an object detector configured to detect an object based on the two types of maps generated by the map generator.