Multiple improvements that allow more precise cropping of the documents from input image and better recognition.
Added "processParam.documentAreaMin" parameter to set area percent that document should fill on the image to be considered as document. Allowed values from 0 to 1. Default value 0.3.
Added "processParam.forceDocFormat” parameter to force use of specified document format when locating and recognizing document to reduce the number of candidates.
Fixed issue with detection of ID card on A4 flatbed scanner produced image with incorrect resolution set.
OCR
Significantly improved recognition quality for Korean, Hebrew, and Cyrillic alphabets.
MRZ
Added new parsers:
Switzerland DL
Updated parsers:
Bulgaria ID card
Croatia ID card
Ecuador ID card
Netherlands passport
Implemented 2x30 MRZ detection and reading for Switzerland DL.
Fixed issue with coordinates for rotated MRZ.
Barcode
Added new parsers:
France ID card
Updated parsers:
Guatemala passport
Implemented mask check for data from 1D barcodes.
Fixed issue with base64 decoding from barcodes data.
Fixed issue with absence of data parsing results from images with low resolution.
Fixed issue with parsing data from multiple barcodes in one document.
Text data parsing and validation
Fixed issue with document number validation for Ireland DL.
Fixed issue with generating full name when same parts of the name are present on different sides of the ID card.
Image QA
Added screen captured images detection as a separate check.
Added portrait occlusion detection as a separate check.
Improved algorithm for detection of colorless images.
Authenticity
Added new check of the halo around the portrait in UV that can be result of photo substitution.
Improved fibers detection and validation algorithms.
Fixed issue with luminescence mask validation for some cases.
Other
Added new eImageQualityCheckType enum members:
IQC_ScreenCapture
IQC_Portrait
Added new eVisualFieldType enum members:
ft_DLClassCode_NT_From
ft_DLClassCode_NT_To
ft_DLClassCode_NT_Notes
ft_DLClassCode_TN_From
ft_DLClassCode_TN_To
ft_DLClassCode_TN_Notes
ft_DLClassCode_D3_From
ft_DLClassCode_D3_To
ft_DLClassCode_D3_Notes
Implemented rotation of output results so the document image is correctly oriented.
Fixed issue with some PDF files processing led to crash in some rare cases.
Fixed issue with processing of multiple pages led to different results based on page order.
Fixed issue with symbols visibility on mobile platforms, that lead to incompatibility with other frameworks that use same dependencies as OpenSSL.
Code quality improved with lots of small issues fixed.
Documentation updated.
Core RFID SDK
Implemented re-processing of the previously obtained session from mobile or desktop station for validating PA and further processing data on the server side.
Fixed issue with TA status calculation based on EF.CVCA ReadingStatus for both ePassport and eDL.
Fixed issue with duplicated nodes instead of the array in JSON results generation for eRFID_ResultType.RFID_ResultType_RFID_BinaryData.
Desktop API
Fixed issue with device thread mode setting on Linux.
Fixed issue with rotated ID card with MRZ cropping when “Quick MRZ reading” is enabled in desktop version working with device scanner.
Hardware support
18MP camera support across all 70x4M, 70x9 devices. Devices with 18MP camera work only with x64 version of SDK.
IMPORTANT! New firmware version 22.06. It is recommended to update your 21.x or 22.x firmware series RFID devices to this new version. Fixed minor issues.
Mobile API
Implemented ability to read multiple pages of a document in manual mode (manualMultipageMode).
Implemented ability to change the background for the status message on the scanning screen.
iOS
Fixed stopRFIDReader function.
Fixed an issue with saving video of the scanning process. Now the video is automatically saved to a separate subfolder.
Android
Added the source code for the extended DocumentReader SDK usage example on GitHub.
Web Service API
Demo site
Added capability of saving request JSON after processing has happened. External images may be processed and then either copy to clipboard request.json or download it to disk from Request tab.
Added capability to download all processing data including request.json, response.json, input and output images in zip archive to disk.
Added capability of drag & drop of JSON with request for processing. Previously processed and saved requests may be used for reprocessing/reproducing results.
Added capability of drag & drop of JSON with results for display. Previously processed and saved responses may be used for displaying results in detail for review.
Added display of the identified document category.
Added capability to store logs into files with daily rotation.
Logging format unified across all platforms.
App logs are stored into rotating "logs/app/document-reader-app.log" file by default.
App logs are sent to stdout instead of stderr on Linux.
Transaction processing log archive now contains files request.json_, _result.json.
JSON log formatting (useful for log collectors).
All API clients now have set api.regulaforensics.com as default server and may be trialed instantly by just downloading the client and running the sample.
When logging of results is enabled via setting DOCREADER_PROCESS_RESULTS_LOG_PATH environment variable LOGS_PROCESS_SAVE_RESULT environment variable may be used for setting whether processing result should be saved into zip file with transaction id as its name. Request input data will be saved regardless of LOGS_PROCESS_SAVE_RESULT value.
Fixed issue with new SDK updated package downloads for x64 version.
Fixed issue with mutex name that allowed to run multiple copies of the application with different names simultaneously.
COM API
Added property BSITR03135Results to set/get whether BSI TR-03135 compliant processing log should be generated. Disabled by default (boolean, read/write).
Documentation updated.
Installation packages
Docker
Red Hat Container Certified images with all details are now available here. It is useful when containers are deployed in OpenShift. regulaforensics/docreadermay be obtained from native Red Hat repositories. This can be considered as a security quality mark of the container.
Migrated to ubi8-minimal base image.
Migrated to Python 3.8.
Licensing
Online transaction-based license usage implemented on mobile. Now it is possible to use same online license on mobile and on the server.
Application ID/platform/OS agnostic license implemented on mobile. This special kind of license allows running SDK on mobile for capturing data as usual but delivers encrypted results package including document images and chip reading data for further processing on the server side with separate online transaction-based license. This allows easy redistribution of the SDK inside custom mobile SDKs without licensing management efforts and issues.
Regula Reader device-based application ID agnostic offline license for Android use implemented. In this case device is storing the license, and license is checked by SDK querying device presence and reading the license from it on initialization.
PKD
ICAO PKD master list version 175 included in RFID SDK component.
Database
Templates updated in whole database where applicable:
Rotation angle to get correct document display position.
Fibers description with regards to updated algorithms.
Nationality field output enabled for ID cards.
Supported documents list now contains details on image patterns in which lights are checked for each document template.
248 countries and territories / 9432 documents included.