These release notes are relevant for the following products:
• MegaMatcher 11.0 Standard / Extended SDK
• VeriFinger 11.0 Standard / Extended SDK
• VeriLook 11.0 Standard / Extended SDK
• VeriEye 11.0 Standard / Extended SDK
• VeriSpeak 11.0 Standard / Extended SDK
You may subscribe for newsletters to receive notifications about the released updates.
List of releases:
- January 7, 2019 (Revision 188367)
See also the list of release notes for previous versions.
January 7, 2019
Product revision: 188367
The first release of MegaMatcher 11.0 SDK line.
Line 11.0 includes outstanding in terms of PIV interoperability NIST MINEX III compliant Neurotechnology+0206 INCIST 378 template generator.
It leads to significantly improved cross-vendor two-way interoperability and native accuracy rates of both Neurotechnology's MINEX III leaderboard standard template matchers:
- top ranked mobile or large-scale systems dedicated Neurotechnology+0108 template matcher;
- smart cards dedicated Neurotechnology+0206 template matcher.
- Improved native recognition reliability evaluated within NIST PFTII as 4E Neurotechnology's fingerprint proprietary templates generator and matcher pair submission.
- New fingerprint classification algorithm is included which improves classification quality.
- New fingerprint slap segmentation algorithm is able to handle more complex segmentation cases which results in more successfully segmented slap images.
- Fingerprint examination scenario has been separated from template extractor. From 11.0 product line, fingerprint extraction creates only template by default, extra features such as generating binarized images are disabled by default. For such scenarios a new NFingerExaminer object is introduced which covers many use cases for fingerprint examining during adjudication process: skeletonized image, singular points (core, delta, double core), calculate number of ridge crossings between two points, find intersections points between specified line and fingerprint ridges.
- Line 11.0 includes outstanding in terms of PIV interoperability NIST MINEX III compliant Neurotechnology+0206 INCIST 378 template generator. It leads to significantly improved cross-vendor two-way interoperability and native accuracy rates of both Neurotechnology's MINEX III leaderboard standard template matchers:
- New face detection algorithm can detect many more faces in difficult conditions, up to full profile head rotation (requires fast CPU, optionally a simpler detection algorithm can be selected on weaker machines).
- New face recognition algorithm – about 5 times less recognition errors on low False Acceptance Rate.
- Improved facial feature points detection.
- New face image quality determination algorithm.
- New background uniformity check.
- New emotion classifier.
- New smaller face template.
- New face attributes/properties determination algorithm.
- Introduced iris detection confidence for determining if image actually contains iris.
- Detection of contact lenses which can distort iris image.
Palm print modality
- New palm print extraction algorithm.
- New text-independent speaker recognition algorithm.
- Trial and non-trial distributions merged. Previously trial distribution was a separate distribution, which required extra effort to migrate to paid license version. Starting from 11.0 line, there is a single distribution and trial/non-trial licensing switch can be made programmatically. NLicenseManager contains TrialMode property (which is false by default), and Activation Server (pgd) supports Trial = 1 in pgd.conf configuration file.
- Android per application sandbox licensing. Previously Android licenses were activated per device, which introduced lots of challenges during application deployment. From 11.0 line Android licenses are activated per application sandbox. This also enables to use license deactivation feature. Please note, that during development application should be updated instead of deleting and installing again, otherwise license instances could be lost. Also, other activation options such as using licenses from PC's dongle can be more convenient for development purposes.
- Android, iOS, Embedded linux licenses merged into Mobile licenses. Instead of specifying a particular operating system, now all licenses designed for mobile device usage are named as Mobile.
- Dongle licensing not supported in containers. If licenses contained in a dongle should be used from Docker (or other) containers, it is needed to launch License Activation Service (pgd) on host machine. For other licensing options License Activation service may be running on Docker. These restrictions apply only to License Activation Service, not on other Neurotechnology libraries and services.
- .NET Standard 1.1 libraries have been upgraded to .NET Standard 2.0. It is recommended to use the .NET Standard libraries for modern .NET applications. For backwards compatibility, .NET framework 4.5 libraries are provided as well.
- Android samples/tutorials now use only Gradle for project building, Maven is no longer supported. This allowed to simplify project structure to make the projects not dependent on various "common" folders and made easier to use from Android Studio.
- ANTemplate (ANSI/NIST ITL) was changed to tolerate lots of validation errors while reading standard templates from other systems. Validation checks are still performed but instead of failing with exception, any errors or warnings are added to ConformanceTestResult collection found on every type of ANRecord. To quickly find if record has any validation errors ANTemplate.IsValidated property can be used. It is up to user to decide whether strict validation is required in particular case. Also, it is possible to fix broken records by changing values of invalid fields to correct ones and calling ANTemplate.CheckValidation() to perform the validation again.
- Simplified adding new records to ANSI/NIST ITL template by introducing AddRecordTypeXXX methods for every record type. Most values are set to default and do not need to be specified manually. BDIF records (FMRecord, FIRecord, FCRecord, IIRecord) also updated in the same way. I.e. deprecated view constructors (e.g., FmrFingerViewCreate) and introduced view addition to owner record function (e.g., FMRecordAddFingerViewEx to FMRecord).
- ANTemplate auto-detects encoding of the read template (traditional or XML).
- WSQ reader to autocorrects broken NISTCOM blocks.
- Simplified any ANSI NIST ITL, BDIF standard record read by enabling ALLOW_OUT_OF_BOUNDS_FEATURES on read by default (even with no flags). Does not apply to write.
- Support creating records in ANTemplate from existing image data without repacking. FromImageData methods should be used in that case.
- Save ANTemplate or BDIF (FIRecord/FIRecord/FCRecord) image as data functionality in samples.
- Added ANType2Record tutorial
- Updated Java sample updated to list missing ANTemplate respective attributes.
- Extend ANSI NIST ITL ANTemplate 13 and 15 records to allow lower resolution value than found in standard specification using ANT_ALLOW_OUT_OF_BOUNDS_RESOLUTION flag. Applies by default with BDIF_NON_STRICT_READ while reading any standard ANSI NIST ITL template.
- Setup.exe is no longer provided. The SDK is designed to be used when unzipped to a writable folder, no additional installer steps are necessary.
- NCluster low-level communication library is no longer provided. Please use NBiometricClient with RemoteConnection functionality instead (refer to IdentifyOnServer tutorials for usage example).
- NdmVirtual previously deprecated API was removed, use ConnectTo functionality instead.