Ear Recognition
Identifies people by the unique shape and texture of the outer ear, often for surveillance or forensic support.
Overview
Ear recognition uses images of the ear’s contours and texture as a supplemental biometric, particularly in surveillance footage.
How it works
- Detect the ear region in images or video.
- Extract features (e.g., shape descriptors, deep embeddings).
- Compare with stored ear templates or multi-modal records.
Common use cases
- Surveillance in public spaces
- Forensic photo analysis
- Research on unobtrusive biometrics
Strengths and limitations
Strengths: Can work at a distance; stable shape over time.
Limitations: Susceptible to occlusion by hair or accessories; smaller dataset availability.
Key terms
- Auricle: External part of the ear captured for matching.
- Occlusion: Blocking or covering of the ear in images.
References
Frequently Asked Questions
Is ear recognition accurate?
Modern CNN-based approaches show promise, but occlusions and hairstyles can reduce reliability.
How is it used with face recognition?
As a supplemental cue in surveillance or re-identification when the face is partially visible or at oblique angles.