Palm Vein Recognition Guide
X-Telcom

Palm Vein Recognition. Why original images matter.

Palm Vein Recognition and Palm Vein Payment Technology rely on biometric identity matching. Securely retaining RGB palm print and infrared palm vein images during registration supports future algorithm upgrades and biometric feature recovery after data loss.

Contactless palm reader capturing a hand
Contactless palm recognition Illustrative scene
01 / CAPTURE
RGBIR
Two images per hand

Palm print + palm vein

02 / RETAIN
Secure image storage

Linked to the correct user ID

03 / RE-EXTRACT
Rebuild or upgrade

Generate biometric features again

01 / THE REASON

Palm vein authentication starts with reliable source data.

Your current matching service may only need extracted features. Original palm images give you the source data to extract those features again when the algorithm or infrastructure changes.

Upgrade algorithms

Use retained images to support new server-side feature extraction without asking users to register again.

Recover lost features

Rebuild biometric features from retained images if the production server or its feature data is lost.

Preserve continuity

Keep the original capture associated with each user as your platform evolves.

02 / ALGORITHM MODELS

What each model needs

The original images and the extracted feature vectors serve different purposes.

One registered hand. Two complementary inputs.
RGB CHANNEL
RGBPalm print imageSurface patterns
Feature extractionEncode the capture
Palm print featuresUsed for matching
INFRARED CHANNEL
IRPalm vein imageVein patterns
Feature extractionEncode the capture
Palm vein featuresUsed for matching

Keep the source images and user associations so features can be extracted again.

Small model
Large model
End-side features

Matching uses palm print and palm vein feature vectors extracted on the device.

Original images are not required for current matching, but retaining them is recommended.

End-side + server-side features

Uses device-extracted features together with features extracted again on the algorithm server.

Retain both RGB and IR images, plus both feature vectors, to support this model.

03 / DISASTER RECOVERY

Disaster recovery for Palm Vein Recognition

If the production server suffers a complete physical failure, securely retained palm images can support rebuilding the biometric feature database on a replacement server.

From retained images to restored matching
01Image backupRGB + IR + user ID
02Replacement serverAlgorithm + license
03Rebuild featuresRe-extract and link
04Validate matchingThen resume service
  1. Retrieve retained RGB and IR images

    Restore the images and their user associations from your backup infrastructure.

  2. Prepare the replacement matching server

    Deploy the algorithm service and arrange the corresponding license.

  3. Re-extract biometric features

    Generate features from the retained images and rebuild their links to user records.

  4. Validate before resuming service

    Check restored records and matching through your backend recovery procedures.

Image retention supports biometric recovery. Your own backend manages the wider data, operations and disaster recovery process.
04 / SERVER LICENSING

Does a backup server need its own license?

It depends on whether the X-Telcom algorithm service is deployed on that server. The algorithm service performs palm ID matching and biometric authentication within your platform.

Store the images

RGB + IR images → Backup storage. No algorithm service deployed.
No additional algorithm license

Run biometric matching

Algorithm service → Palm ID matching. Deployed and ready for recognition.
Corresponding license required
Choose the backup server's role

No additional algorithm license

The server only stores original palm images for recovery. The algorithm service is not deployed on it.

A storage-only backup does not provide matching. Deploying a matching service during recovery requires the corresponding licensing arrangements.
05 / IMPLEMENTATION

Image retention for Palm Vein Payment Technology

For a palm vein payment system, this guide covers the biometric registration and matching layer. Payment processing, user data and disaster recovery remain part of your own backend infrastructure.

  • Save both capture types. Each registered hand has one RGB image and one IR image. Two-hand registration therefore produces four images.
  • Keep quality-approved captures. Retain images that pass the SDK's registerPalm quality check.
  • Associate images with the user. Store them with a unique user ID and identify the hand and image type.
  • Protect the retained data. Plan secure storage, controlled access and backup procedures within your backend infrastructure.
Where can developers find image-saving sample code?

Refer to the ImageUtils class in the SDK demo project. It demonstrates converting a data stream into a Bitmap and saving it locally. Integrate the image-saving process into your own registration and storage workflow.

Do we have to register both hands?

No. The demo supports two-hand registration, but your application can register one or both hands. Retain one RGB image and one IR image for each registered hand.

Who manages the disaster recovery process?

Your backend infrastructure manages operations, user data, storage and recovery procedures. The X-Telcom algorithm service provides the palm ID matching function within that platform.