Object Recognition Using Feature Point Speed for Cluster Identity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object detection devices struggle to accurately measure the movement speed of clusters, leading to inaccurate recognition of objects around a vehicle at different time points, especially when parts of the vehicle are shadowed or hidden by obstacles.
Innovation Solution
An object recognition method that clusters distance measurement points at fixed intervals, maps these clusters to images captured by a camera, extracts feature points, calculates their movement speed, and determines object identity based on position and speed differences between clusters at different times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If clusters are fitted to shape models using cluster movement speed, then object detection can be performed, but accurate measurement of cluster movement speed is difficult when different parts of the same object are clustered at different time points
Solution Approach 1:
The patent introduces an intermediary approach by mapping clusters to image regions and extracting feature points from these regions. Instead of directly measuring cluster movement speed (which is unreliable), the system uses feature points as intermediaries to track object movement. The feature point movement speed is calculated based on the movement of corresponding feature points in consecutive images, providing a more reliable measure of object movement that can be used to determine object identity across time points.
2Productivity
If clusters are generated at fixed periods by object detection sensor, then processing can be performed systematically, but different parts of the same vehicle are clustered at different time points when shadowed or hidden
Solution Approach 1:
The patent implements a feedback mechanism by determining whether clusters at different time points correspond to the same object based on comparing feature point positions and movement speeds. The system uses the calculated feature point movement speed and position differences as feedback to verify object identity. This feedback loop allows the system to maintain systematic processing while correcting for the issue of different vehicle parts being clustered at different times, as the feature point matching provides verification that ensures accurate object tracking.
Data Source
AI summary
An object recognition method includes generating a cluster at a fixed period by clustering distance measurement points acquired from an object detection sensor, mapping the cluster to an image captured by an imaging means, setting a region on the image to which the cluster is mapped as a processing region, extracting a feature point from the image in the set processing region, calculating a movement speed of the extracted feature point, and determining whether clusters at different time points are clusters of an identical object based on a difference between positions of two feature points corresponding to the clusters at the different time points and a difference between movement speeds of the two feature points corresponding to the clusters at the different time points.