Target Recognition System for Accurate Virtual Image Detection
Here’s PatSnap Eureka !
Summary
Problems
Existing target recognition systems using millimeter-wave radar sensors face errors in detecting targets outside the basic detection area due to phase folding, leading to incorrect identification of virtual images as real images, which results in unstable recognition results.
Innovation solutions
A target recognition apparatus that defines a basic detection area, an additional detection area, and a folding area, using a radar sensor and an image sensor to differentiate between real and virtual images by determining the presence of a history connection and combination with image targets, and calculating the likelihood of virtual images, thereby improving accuracy in target recognition.
TRIZ Analysis
Specific contradictions:
General conflict description:
Principle concept:
If radar sensor is used for target detection, then detection range is extended, but phase folding causes erroneous detection in specific angular ranges
Why choose this principle:
The patent introduces an image sensor as an intermediary device to verify radar detection results. The image sensor captures visual information in the same angular range where phase folding occurs, allowing the system to cross-validate radar-detected targets and distinguish real targets from virtual images caused by phase folding errors.
Principle concept:
If radar sensor is used for target detection, then detection range is extended, but phase folding causes erroneous detection in specific angular ranges
Why choose this principle:
The system implements a feedback mechanism where image sensor data is used to verify and correct radar detection results. When the image sensor fails to detect a target in a region where the radar detects a target in the folding area, the system identifies this as a phase folding error and corrects the directional measurement accordingly.
Application Domain
Data Source
AI summary:
A target recognition apparatus that defines a basic detection area, an additional detection area, and a folding area, using a radar sensor and an image sensor to differentiate between real and virtual images by determining the presence of a history connection and combination with image targets, and calculating the likelihood of virtual images, thereby improving accuracy in target recognition.
Abstract
In a target recognition apparatus, a candidate detection section detects a target candidate, provided that a target exists in a basic detection area. A candidate addition section adds, regarding each target candidate detected in a folding area, a target candidate determined provided that the target candidate detected in the folding area is a virtual image, and a corresponding real image exists in an additional detection area. A tracking section determines, regarding each detected and added target candidate, presence/absence of a history connection with the target candidate detected in a past measurement cycle. A combination determination section determines, regarding each target candidate, presence/absence of a combination with an image target, based on whether or not an image target associated with the target candidate exists. A likelihood calculation section sets and updates a likelihood of a virtual image of the image target by using a determination result of the combination determination section.