Object Detection Device Using Shoulder Structure Height Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image processing devices face difficulties in accurately distinguishing between objects and road surfaces when environment conditions are unsuitable for generating disparity information, leading to decreased position detection accuracy and potential erroneous object detection.
Innovation Solution
An object detection device equipped with a stereo camera, three-dimensional data generation unit, shoulder structure detection unit, structure ground-contact position derivation unit, road surface position estimation unit, and object detection unit, which calculates disparity data and estimates road surface height using ground-contact positions of shoulder structures to separate road surface and object disparity data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If disparity information is used to detect objects on the road surface, then object detection capability is improved, but detection accuracy decreases when the environment is unsuitable for generating disparity information
Solution Approach 1:
The patent introduces shoulder structures (guardrails, barriers, signs) as intermediary reference objects to estimate road surface height. These structures provide stable reference points that do not depend on disparity information quality, acting as mediators between the detection system and the road surface when disparity-based detection fails
Solution Approach 2:
The system switches from using disparity information as the primary parameter for road surface detection to using height parameters derived from shoulder structure positions. This parameter change allows accurate road surface estimation in environments where disparity information is unreliable
2Reliability
If conventional disparity-based detection is used, then object detection is possible, but confusion between road surface and object occurs leading to erroneous detection
Solution Approach 1:
The patent segments the detection space by first identifying shoulder structures and using them to define the road surface boundary. This segmentation separates the road surface detection task from general object detection, preventing confusion between the two by establishing clear spatial boundaries derived from shoulder structure positions
Data Source
Figure 1(a)~2
Figure 3
Figure 4(a)~4(b)
AI summary
The purpose of the present invention is to provide an object detection device which is capable of accurately estimating the height of a road surface and is capable of reliably detecting an object present on the road surface. This object detection device 1 detects objects upon a road surface, and is equipped with: a stereo camera 110 for capturing images of the road surface 101a and the road shoulder 101b and generating image data; a three-dimensional data generation unit 210 for calculating disparity data for the pixels in the image data; a shoulder structure detection unit 310 for using the disparity data and/or the image data to detect shoulder structures; a structure ground-contact position derivation unit 410 for deriving the ground-contact position of a shoulder structure; a road surface position estimation unit 510 for estimating the height of the road surface 101a from the ground-contact position of the shoulder structure; and an object detection unit 610 for detecting an object 105 upon the road surface 101a by using the disparity data and the road surface height to separate disparity data corresponding to the road surface 101a and disparity data for the object.