Distance Image Acquirer for Moving Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional moving-object recognition systems face challenges in accurately detecting and tracking moving objects, particularly in varying lighting conditions and dynamic scenarios, leading to inconsistent recognition and control issues in vehicles and other equipment.
Innovation Solution
A processing apparatus comprising a distance image acquirer, a moving-object detector, and a danger-level determining unit, which uses stereo camera technology and scene graph models to acquire distance images, detect moving objects, and determine danger levels, enabling robust object recognition and control adjustments based on priority ranking and situation prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional moving-object recognition methods using two-dimensional images and focal distance are used, then the system is simple to implement, but the measurement accuracy for distance and detection accuracy for changes in movement are low
Solution Approach 1:
The patent transitions from two-dimensional image analysis to three-dimensional distance image analysis. By acquiring distance information for each pixel and generating parallax images from stereo images, the system adds a depth dimension to object recognition, enabling accurate measurement of distance and movement changes without relying on complex focal distance calculations from 2D images.
Solution Approach 2:
The patent introduces distance images as an intermediary representation between the raw stereo images and the final object recognition results. The distance image acquirer generates parallax images that encode depth information, serving as a mediator that simplifies subsequent moving object detection and distance measurement while maintaining high accuracy.
2Reliability
If the system uses distance images and scene graph models for robust object recognition, then the reliability of moving-object recognition is improved, but the device complexity increases
Solution Approach 1:
The patent performs preliminary action by pre-defining scene graph models that represent expected spatial relationships between objects in typical driving scenarios. These pre-established models are used to guide the interpretation of distance images, enabling the system to reliably recognize moving objects by matching observed distance patterns against known scene configurations, thereby improving reliability without requiring complex real-time reasoning.
3Measurement precision
If the system acquires and processes distance images with parallax information, then the detection accuracy for moving objects is improved, but the processing time and computational load increase
Solution Approach 1:
The patent extracts only the essential distance information needed for moving object detection from the full stereo image data. The distance image acquirer generates parallax images that contain condensed depth information, and the moving object detector focuses specifically on changes in these distance values rather than processing complete image frames, thereby reducing computational load while maintaining detection accuracy.
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
The system enhances the accuracy and reliability of moving-object recognition, allowing for stable tracking and control of recognition targets, even in changing conditions, thereby improving safety and reducing errors in autonomous systems like vehicles and robots.
Implementation Method 1
obtains parallax images (also called distance images) by calculating parallax from stereo images obtained by a stereo camera apparatus
Data Source
AI summary
A processing apparatus includes a distance image acquirer, a moving-object detector, and a danger-level determining unit. The distance image acquirer acquires a distance image containing distance information of each pixel. The moving-object detector detects a moving object from the distance image. The danger-level determining unit determines a danger level of the moving object by use of the distance image, and outputs the danger level to a controller that controls a controlled unit in accordance with the danger level.