FingerprintVisionLab

FingerprintVisionLab is a SwiftUI iOS prototype for experimenting with camera-based fingerprint image capture and simple local comparison. It uses the rear camera preview to help a user position a finger, captures the centered alignment area, enhances the image with Core Image filters, and compares lightweight feature vectors for enrollment and verification flows.

Features

• Live rear-camera preview for finger alignment.
• Torch-assisted alignment so close-up finger captures are visible.
• Manual enrollment of a reference fingerprint image.
• Manual verification against the enrolled reference.
• High-contrast preview of the latest processed capture.
• Local-only processing with AVFoundation and Core Image.

Requirements

• Xcode 16 or newer.
• iOS 18.6 target or compatible project settings.
• A physical iPhone or iPad with a rear camera and torch.

The simulator is not suitable for the capture flow because it does not provide the same live rear-camera and torch behavior as a device.

Running

1. Open FingerprintVisionLab.xcodeproj in Xcode.
2. Select the FingerprintVisionLab scheme.
3. Choose a physical iOS device.
4. Build and run.
5. Allow camera access when prompted.

Capture Flow

1. Place a finger inside the center alignment guide.
2. Keep the torch enabled if the preview is dark.
3. Tap 1. Capture Reference to enroll the current centered finger area.
4. Place the same or another finger in the guide.
5. Tap 2. Capture & Compare to compare the new capture with the reference.

Implementation Notes

• CameraPreviewView renders the camera using AVCaptureVideoPreviewLayer as the backing layer of a UIKit view embedded in SwiftUI.
• FingerprintScannerController owns the capture session, camera permission flow, torch state, image enhancement, and comparison logic.
• The capture step crops the centered alignment area instead of relying on hand-pose detection. This is better suited for close-up fingertip/fingerprint framing, where full hand joints may not be visible.

Limitations

This is a vision prototype, not a production biometric identity system. The current comparison uses a simple image-derived feature vector and should not be used for security-sensitive authentication decisions.