Wheelchair following target identification method based on deep learning
The calculation of gradient and color contrast values through deep learning methods for light compensation and target feature extraction, which solves the problem of unstable target recognition and tracking in complex environments of wheelchairs, and achieves higher accuracy and stable target follow-up.
Patent Information
- Application Number
- CN202510469511.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-15
- Publication Date
- 2025-07-25
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has low accuracy in recognition and tracking of wheelchair follow-up targets in scenarios with severe light changes or complex backgrounds, making it difficult to deal with misjudgment of target positions and occlusion, resulting in unstable wheelchair follow-up.
The wheelchair follow-up target recognition method based on deep learning, by calculating the gradient statistics value and color contrast value of pixel neighborhood, adjusting the spatial coefficient and reflection components of Gaussian filtering, combining the lighting components for lighting compensation, segmenting the potential target area, calculating the color texture descriptor value, tracking the target position, and adjusting the processing parameters through feature comparison and attribute correlation performance metrics of continuous frames.
It improves the accuracy and stability of the target recognition of the wheelchair in complex environments, and enhances the reliability and robustness of the wheelchair's real-time follow-up.