Robot Head Face Tracking Error Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional human face detecting and tracking methods for robots suffer from high detection errors, leading to inaccurate rotation control of robot heads, as they often recognize non-human faces as human faces.
Innovation Solution
A face detecting and tracking method that uses the Viola-Jones algorithm for face detection, followed by a detection error removal process to identify and correct falsely detected face boxes, and employs the Camshift algorithm for tracking, ensuring accurate face detection and continuous tracking by determining overlapping rates and average values of face box coordinates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional face detecting and tracking methods are used, then the robot can detect and track human faces, but the detection accuracy is low and non-human faces are frequently misrecognized as human faces
Solution Approach 1:
The patent segments the face detection process into multiple independent stages: initial face detection to identify candidate regions, error removal processing to eliminate false detections, and tracking to confirm valid faces. This multi-stage segmentation allows each stage to focus on specific tasks, improving overall detection accuracy while reducing false positives from non-human faces
Solution Approach 2:
The patent performs preliminary error removal processing on detected face boxes before final tracking confirmation. By预先 removing obviously erroneous detections (such as non-human faces that fail basic validation criteria), the system prevents these false positives from progressing to the tracking stage, thereby improving detection reliability before final confirmation
2Measurement precision
If conventional face detecting and tracking methods are used, then the robot can control head rotation to track faces, but the rotation control is inaccurate due to high detection errors
Solution Approach 1:
The patent implements a feedback mechanism where the tracking results from previous frames are used to validate and refine current face detections. The system continuously compares detected face positions with tracked positions, and only accepts detections that are consistent with the tracking trajectory. This feedback loop ensures that rotation control commands are generated based on accurate, verified face positions rather than erroneous detections
Solution Approach 2:
The patent performs preliminary validation of detected face boxes against tracking information before using them for rotation control. By checking whether detected faces are consistent with previously tracked positions and motion patterns, the system filters out erroneous detections that would cause inaccurate rotation commands, ensuring only reliable face positions drive head rotation
Data Source
AI summary
A face detecting and tracking method includes: acquiring an image and performing a face detection to determine one or more face boxes of each acquired image; performing a detection error removing process to the image of current frame, and saving one of the one or more face boxes of the processed image of current frame as a first face box; and determining a tracking box of the image of current frame based on the first face box and a tracked face box, and tracking the face in the tracking box of the image of current frame.