Capture SDK 

Biometric Liveness Verification 

Biometric Capture SDK for automatic face and/or fingerprint capture allows service providers to integrate biometric capabilities within their own smartphone application (APP).

IDEMIA provides an “example of application” (in source code) in addition to the Capture SDK. The goal is to help service providers speed up the SDK integration into their own application.

Biometric Capture SDK uses the embedded smartphone’s camera sensor. It can use either front or back camera. It does not require the use of dedicated biometric sensors.

Biometric Capture SDK offers the following services:

  • Face/fingerprint auto-capture from the live video performed with smartphone’s camera
  • Liveness detection for both face and fingerprint
  • Biometric template creation and comparison

Automatic Capture 

Biometric Capture SDK provides an easy face/fingerprint acquisition, either through self-acquisition, or through an assisted acquisition (face/fingerprint acquisition performed by a human operator).

The SDK checks that the end-user’s face/fingerprint is correctly positioned. Interactive messages are provided to guide the end-user until acquisition conditions are met. When acquisition conditions are good enough (position, illumination) the best image is automatically selected from the video.

Face Liveness Detection 

Biometric information can be found within the public area. Indeed, face images could be available on the internet.

The goal of the liveness feature is to provide mechanisms that prevent/fight fraud attempts, for example, through the use of still images/photos and/or videos of a given person.

Biometric Capture SDK offers three anti-spoofing measures:

  • Artefact detection: The SDK detects artefacts that occur when a fraudster tries to perform face acquisition by placing the smartphone’s camera in front of a video feed.
  • 3D model analysis: The SDK detects that the face is in three dimensions. To do this, the SDK asks the end-user to move their head or their smartphone.
  • Challenge/response: The SDK asks the end-user to perform head elementary movements. The sequence of elementary movements that the end-user has to execute is random.

Characteristics of Fingerprint Capture 

  • Simultaneous four fingerprint auto capture
  • Fingerprint image is processed and segmented in four fingerprint images, for later matching
  • Contactless, non-intrusive fingerprint auto-capture
  • Robust to external lightning conditions
  • Robust to wet and dry fingers
  • Thumb capture is not operational
  • No liveness detection
  • At least three fingers are required. Amputees without at least three fingers will not be able to perform fingerprint acquisition.
  • Compatible only with the rear camera
  • Torch is required, or is automatically triggered by Capture SDK
  • Rear camera must have the capability to focus on short distance

Document Acquisition 

Document Capture SDK is a SDK for automatic ICAO identity documents capture (ID1, ID2, ID3) and MRZ Optical Character Recognition (OCR) from live video performed by the embedded smartphone’s camera. It allows service providers to integrate identity document automatic capture capabilities, especially OCR, within their own smartphone application (APP) or to develop their own applications upon the SDK. It does this by providing an “example of application” (in source code) in addition to Document Capture SDK. The goal is to help service providers to speed up the SDK integration into their own application.

Document Capture Functionalities 

The functions below are directly performed on the smartphone:

Automatic Capture

For end-user convenience, Document Capture SDK guides the end-user throughout the acquisition phase (document is too far, reflections detected, etc.), detects automatically the identity document (ID) document and performs automatic ID acquisition.

More specifically, the SDK checks that the ID is correctly positioned in the capture zone in order to maximize the quality of the image. It maximizes the performances of OCR and potentially the later biometric matching (photo of the ID vs a live selfie).

Optical Character Recognition of MRZ

The OCR of the MRZ is done directly within the smartphone.

NOTE: The OCR limit is that that only IDs with two or three MRZ lines are readable by the SDK (not just one MRZ line).

  • Photo cropping & MRZ analysis
  • Document Capture SDK extracts the data of the end-user from the MRZ and the photo of the document.
  • Document Capture SDK interprets MRZ data

Generic Technical Considerations 

Capture SDK can be used in conjunction with or integrated in other IDEMIA Identity Solutions or ID Proofing solutions.

Capture SDK is compatible with common mobile operating systems iOS (10.0+) and Android (5.0+).