Vehicle Obstacle Detection via Shape and Motion Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting obstacles in a vehicle's environment, particularly in the field of motor vehicles, face challenges in accurately detecting pedestrians due to difficulties in distinguishing them from other objects and background movements.
Innovation Solution
A method combining shape recognition and movement detection, using a combination of image processing techniques such as AdaBoost, Harris Corner, and Lucas Kanade methods, along with sensor data from ultrasonic and radar sensors, to identify and classify obstacles with confidence indices, enhancing the detection of pedestrians by minimizing errors in distance estimation and background noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single detection method (shape recognition or movement detection) is used, then the detection process is simpler, but the detection accuracy and reliability for pedestrians is insufficient
Solution Approach 1:
The patent combines shape recognition detection and movement detection into a unified detection system. The control unit integrates results from both detection methods, using shape recognition to identify pedestrian forms and movement detection to track motion patterns, thereby improving overall detection reliability while managing system complexity through coordinated integration.
2Measurement precision
If multiple detection methods are combined, then detection accuracy improves, but the complexity of the detection system increases
Solution Approach 1:
The system implements feedback mechanisms where detection results from shape recognition and movement detection are continuously evaluated and integrated. The control unit uses confidence indices from both methods to validate and refine obstacle detections, creating a feedback loop that improves measurement precision while managing complexity through systematic result integration.
3Measurement precision
If confidence indices are applied to classify obstacles, then the distinction between pedestrians and other objects improves, but the processing time increases
Solution Approach 1:
The system applies confidence indices selectively to classify obstacles, focusing computational resources on ambiguous cases where classification is uncertain. For clear-cut detections with high confidence indices, minimal processing is required, while borderline cases receive more intensive analysis, thereby improving classification accuracy without proportionally increasing overall processing time.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
The detection procedure involves classifying obstacle (O) detected with indexes of confidence applied to the regions of interest in relation to given characteristics. The classification of the detected object is validated in relation to the indexes of confidence and in relation to the regions of interest to obtain validated regions of interest. The first region of interest is defined by detecting obstacle by shape recognition in acquired image of environment. Second region of interest is defined by detecting obstacle by movement detection in relation to vehicle on sequence of acquired images. Independent claims are also included for the following: (1) an obstacle detection device; and (2) a computer program product.