What's new in VeriLook Standard SDK 12.4
Oct 25, 2022
- Algorithms-related changes:
- Face liveness check:
- The new highly accurate algorithm has been certified for a level 1 PAD by iBeta Quality Assurance. The certification includes smartphone and PC with camera scenarios, and ensures compliance with ISO 30107-3 Biometric Presentation Attack Detection Standards.
- Specific parameters allow better algorithm adaptation for image and stream processing.
- Quality check was integrated into liveness processing so both checks can be performed simultaneously.
- Fingerprint quality check:
- New attributes which allow better control on biometric data quality and can tell if a finger is too wet, too dry, pressed too much or not enough, or if only tips of fingers are present.
- Iris quality check:
- New attribute for detecting iris rotation (roll) angle in an image.
- New features:
- Face to background ratio can now be specified for a token image via a parameter.
New in VeriLook Standard SDK 12.2 (Jul 13, 2021)
- Faces:
- Added the fastest extraction speed scenario for faces. It can be enabled using a parameter. This scenario is especially useful for devices with low compute power to improve extraction time. Obviously, this results in lower quality. Use this option only if default extraction speed was found unnaceptable.
- New face liveness detection mode optimized for high resolution images. Can be enabled using using a parameter.
- Face/not-face filtering for existing face templates. Latest face quality algorithm can be applied to existing templates generated even with previous versions of the SDK in order to accurately remove non-face images. This feature is enabled using a parameter. When enrolling face templates to engine, if a face does not pass quality threshold it means either there are no face or low quality face is present in template and ObjectNotFound NBiometricStatus is returned. It is possible to adjust the threshold.
- New high quality face photo background elimination algorithm. For mobile devices there is a special version which works a lot faster but is less accurate. User may choose between high quality and fast version using a parameter.
- More accurate face landmark detection.
- Irises:
- Possibility to automatically determine iris position. It is beneficial to set position to left or right for irises so that matching is done only with compatible position resulting in better performance. This feature can be enabled using a parameter.
- Improved iris extraction and matching algorithm. More accurate template is created when engine.TemplateSize=Large (default since this line). Iris templates are interoperable with the previous product lines.
- New features:
- Added parallel face extraction tutorial for C language which demonstrates how to perform tasks in parallel for maximum performance.
- Fixes:
- Fixed several issues in Activation Wizard.
- Fixed license generation bug on Android when using BouncyCastle as the security provider.
- Updates and improvements:
- Better interoperability with older SDK versions when using dongle licensing.
- Better issue diagnostics message on Android if Internet permission not present.
- Fingerprint scanners:
- Added DigitalPersona U.are.U support on ARM Linux.
- Added support for Jenetric LIVETOUCH QUATTRO Compact scanner. Updated Jeneteric's SDK which includes advanced spoof detection and some bug fixes.
- Added Mantra finger scanner module (MFS500 and MAPRO CX devices supported).
New in VeriLook Standard SDK 11.2 (Sep 5, 2019)
- Algorithms:
- Improved fingerprint recognition algorithm which was recognized by NIST as the most interoperable and accurate fingerprint template generator and matcher in the world (submission Neurotechnology+010A in the NIST MINEX III evaluation).
- Improved face recognition algorithm which provides even better accuracy of face recognition using medium template size: with almost 6 times smaller algorithm deployment files size and 2 times faster extraction speed it reaches the same accuracy as the large template.
- New features:
- Added CropRect property to NEAttributes for returning cropping position when cropping standard iris images.
- Fixes:
- Fixed a bug when EnrollWithDuplicateCheck could sometimes not detect duplicates when multiple operations are started concurrently on very slow hardware.
- Fixed an issue with NFingerExaminer not failing on images which have small number of minutiae.
- Fixed an issue with Suprema RealScan-F not listed as palmprint scanner.
- Fixed a number of small bugs and issues.
- Updates and improvements:
- Optimized reading and writing ANTemplate XML files.
- More robust JPEG 2000 file handling.
- Scanners support:
- Added support for Suprema BioMini Slim 3.
- Added support for Integrated Biometrics Kojak 3.0.
- Faster start of Integrated Biometrics scans after the first scan on Android.
- Added image preview for DigitalPersona U.are.U scanners on Microsoft Windows (for models which support it)
New in VeriLook Standard SDK 10.0 Revision 186632 (Dec 5, 2018)
- New features:
- Added support for Iris ID TD-100 device.
- Fixes:
- Fixed a voice extraction error when audio duration is 60 seconds.
- Fixed issues with licensing on Android 9.
- Fixed Mac crash during application quit.
- Fixed issues in Fujifilm camera module.
- Fixed FireWire operation issue with Cross Match L Scan.
- Fixed issues with ANTemplate and other standard record support.
- Fixed a number of other minor bugs.
- Updates and improvements:
- Improved palm print matching speed (up to 2,000 per second on Core i7-4771 processor).
- Green Bit palm scanners now support palm positions.
- Updated Cross Match L Scan support libraries.
New in VeriLook Standard SDK 10.0 Revision 182882 (Sep 19, 2018)
- New features:
- Added Iris ID iCAM T10 scanner support.
- Added Fujifilm camera support.
- Fixes:
- Fixed Hongda S500 fingerprint scanner support.
- Fixed an issue with GreenBit fingerprint scanners which continued scanning if captured image was of unacceptable quality.
- Fixed list of Futronic scanners which support spoof detection.
- Fixed issues in these device modules: Futronic (Android), NextBiometrics, Hongda, Integrated Biometrics, CrossMatch IScan, CrossMatch LScan, HFSecurity.
- Fixed a number of other minor bugs.
- Updates and improvements:
- Suprema Real Scan module now allows to configure spoof detection threshold.
- Updated Nikon camera module to support latest devices.
New in VeriLook Standard SDK 10.0 Revision 180563 (Jul 17, 2018)
- Fixed biometric capturing device support, including Cross Match, Dermalog, CredendeID, ZKSoftware, Startek, DigitalPersona, NeuBio, IntegratedBiometrics, ARH and Suprema.
- Fixed a number of other minor bugs.
- Added new licensing capabilities on iOS and Android.
- Added new functionality to ANTemplate (API, samples and tutorials) allowing to create ANType-X record from image buffer.
- Improved documentation.
New in VeriLook Standard SDK 10.0 Revision 178878 (May 31, 2018)
- Fixed issues with some activation dongles not working on Linux.
- Fixed fingerprint scanner support, including Integrated Biometrics Kojak, Green Bit, Suprema Real Scan.
- Fixed a number of issues with Android devices.
- Fixed a number of other minor bugs.
- Added face liveness check to iOS sample.
- Updated Canon SDK to support newest devices.
New in VeriLook Standard SDK 10.0 Revision 176566 (Mar 26, 2018)
- Fixed an issue with ANTemplate Type-8 record creation from NImage.
- Fixed issues with some media formats which were not properly converted under Windows Media framework.
- Fixed an issue with ABI incompatibility between BLAS libraries causing face extraction to fail on Android under some circumstances.
- Fixed a number of other minor bugs.
- Added liveness check to the iOS sample.
- Updated Integrated Biometrics support module on Android.
New in VeriLook Standard SDK 10.0 Revision 173239 (Jan 10, 2018)
- Fixed issues with face generalization.
- Fixed a bug with face detection when a face is rotated to 180 degree angle.
- Fixed ICAO validation fixes and attribute refresh in sample views.
- Fixed CbeffRecord TLV-encoded patron format Biometric type and subtype cross-validation.
- Fixed issues with ANTemplate.
- Fixed issues with NServer.
- Fixed issues with capturing from microphone on Windows 10.
- Fixed issues raised when IriTech IriShield device was not granted usage permissions on Android.
- Fixed issues related to IntegratedBiometrics scanner module.
- Fixed a number of other minor bugs.
- Updated FMRecord, FCRecord and IIRecord quality block validation/calculation according to 29794-1:2016 standard.
- Updated Android build tools and gradle version in sample and tutorial builds.
- Updated Suprema SDK in BioMini scanner module.
- Updated Lumidigm SDK in device module.
- Updated CrossMatch LScan SDK in device module.
- Improved documentation (standards, odbc usage instructions, licensing components, etc.).
New in VeriLook Standard SDK 10.0 Revision 171427 (Jan 10, 2018)
- Fixed bugs with ANTemplate record.
- Fixed bugs in Face Verification component.
- Fixed issues with NLRecord.
- Fixed issues in licensing on Android 8.
- Fixed issued in Aratek scanner module.
- Fixed a number of other minor bugs.
- Added ANTemplate XML encoding for ANSI/NIST-ITL 1-2011, ANSI/NIST-ITL 1-2011 Update:2013 Edition 2 and ANSI/NIST-ITL 1-2011 Update:2015 standards.
- Added new tutorials for new ANTemplate functionality.
- Updated Cbeff TLV-encoded patron format to ISO/IEC 19785-3:2015 standard.
- Updated face quality estimation.
- Updated to JNA 4.5.0 version.
- Updated jnidispatch libraries.
- Updated Delete operation example in Abis samples.
- Updated licensing component list in documentation.
- Optimized capturing from camera on Android.
New in VeriLook Standard SDK 10.0 Revision 169753 (Jan 10, 2018)
- Fixed bugs in Face Verification component.
- Fixed NServer port configuration.
- Added ARM linux 64-bit platform support.
New in VeriLook Standard SDK 10.0 Revision 168881 (Jan 10, 2018)
- Fixed voices NotActivated exception.
- Fixed a number of minor bugs.
- Updated NdmSupremaBioMini, NdmGreenBit, NdmLumidigm.
- Updated ANTemplate's non-ASCII fields parsing when no charset encoding is set in Type-1 record. By default non-ASCII characters are decoded as UTF-8.
New in VeriLook Standard SDK 10.0 Revision 167245 (Jan 10, 2018)
- Fingerprints:
- Improved algorithms:
- Fingerprint extraction, matching, and generalization algorithms have been improved.
- New fingerprint quality estimation is included.
- Note: new matching algorithm for small templates may use up to 3 times more memory to provide best recognition quality. If this is a concern, same memory usage as in previous line may be restored by setting NBiometricEngine.FingersMatchingSpeed parameter to High.
- Minimal minutia count:
- Fingerprint extractor reports an error status if fingerprint does not meet minimal quality requirements. In that case application should attempt to rescan the fingerprint. However, in some cases there is no possibility to rescan and it is preferable to decrease default requirements to extract lower quality template rather than have no template at all. NBiometricEngine now exposes a FingersMinimalMinutiaCount property for that purposes. Also, FingersQualityThreshold property is often used together.
- Faces:
- Face detection has been enhanced to detect even more faces and support higher degrees of rotation. The new algorithms are automatically enabled by increasing MaximalRotation parameters in NBiometricEngine. Also there is a "Faces.DetectMore" which detects even more faces the default detector cannot find.
- Profile faces images are now fully supported: profile faces can be detected, extracted, and matched. Please note that it is not possible to match profile faces with frontal.
- Previously face detection confidence was stored in NBiometricAttributes.Quality property. After template extraction the same Quality property was updated with extraction quality value. To avoid this confusion, a new NBiometricAttributes.DetectionConfidence property was introduced specifically for detection confidence. Quality property now only contains template extraction quality. This is a backwards incompatible change which may require changes in user application (if application uses the properties directly). NFaceView components supplied with the SDK will handle the new properties automatically.
- New age detection algorithm is included.
- ICAO:
- New checks:
- Face Darkness
- Pixelation
- Looking Away
- Skin Reflection
- Glasses Reflections
- Red Eye
- Skin Tone
- Washed Out Colors
- "ICAO" fixes: Red-eye correction.
- Liveness check: Custom liveness check steps support. Ported liveness actions from FaceVerification custom project.
- Irises:
- Irises extractor was enhanced to better handle complex extraction cases. This can greatly improve recognition accuracy for databases which contain many badly scanned images.
- A new irises detection confidence value is available which can be helpful to determine if images should to be rescanned.
- .NET:
- .NET libraries updated to use .NET framework 4.5 (from 2.0).
- Async versions for most of methods in NBiometricEngine and NBiometricClient are now available in addition to synchronous ones. Legacy Asynchronous Programming Model (APM) is still supported for backwards compatibility.
- Previously Neurotec.Media_FX35 and _FX45 libraries were distributed to allow NImage conversion to WPF images. Now this functionality is integrated into Neurotec.Media.dll and these extensions are no longer provided separately.
- System.Drawing.Point and System.Drawing.Rectangle have been replaced with the same named classes in Neurotec.Drawing namespace. This should help to remove unneeded dependencies for Windows or ASP.NET services.
- .NET Portable libraries have been replaced with .NET Standard 1.1. They can be useful if writing portable software which needs to support multiple platforms. Support for platform-specific features, such as image conversion to WinForms or WPF bitmaps is excluded. The libraries can be found in Bin/dotNET_Standard directory.
- Upgraded supported programming and development environments:
- Dropped armeabi (ARM v5) architecture support in Android. Minimal supported Android version raised accordingly to 19. Currently supported architectures include armeabi-v7a, arm64-v8a, x86.
- Dropped support for Visual Studio 2008. Currently supported versions include VS 2012, 2013, 2015, and 2017.
- Windows XP is no longer supported. If your application requires Windows XP support, please use NeurotecSDK product line 9.0.
- Android:
- New multimodal sample is included which demonstrates how to use multiple biometric modalities (fingerprints, faces, irises, etc.) in a single app.
- NCore.setContext() no longer requires Activity or Service context. ApplicationContext can be used instead. However, some fingerprint scanners will not work in such case.
- Standards:
- Implemented new ANSI/NIST-ITL 1 (Data Format for the Interchange of Fingerprint, Facial & Other Biometric Information) standards:
- New standard versions - ANSI/NIST-ITL 1-2011, ANSI/NIST-ITL 1-2011 Update:2013 Edition 2 and ANSI/NIST-ITL 1-2011 Update:2015.
- Implemented "Incits 378" vendor block, defined by INCITS Technical Committee M1 developed standard INCITS 378, for Minutiae data (Type-9) record.
- Added Type-20 (Source representaton) and Type-21 (Associated context) records.
- Deprecated Type-3, Type-5 and Type-6 records.
- Simplified ANTemplate/Type-X records API by adding empty constructors.
New in VeriLook Standard SDK 4.0 (Dec 9, 2009)
- Increased recognition accuracy
- full tolerance to face roll
- increased tolerance to face yaw and pitch
- improved facial template generalization