Document Reader SDK
Document Reader SDK
Home
Website
GitHub
npm
Getting Started with Ionic
Troubleshoot license issues
Changelog
Requirements
Core
Scenarios
How to add the SDK to the project?
Database
Initialization
Custom settings
About SDK and license
Results
Enumerations
FAQ
Powered by GitBook

Changelog

This page contains notable changes made to the SDK

April 26, 2021 - v5.5

API:

  • Added the ability to save a video of the document recognition process.

  • Added the ability to use the FullAuth scenario in multipage mode.

  • Added the ability to set the mode via checkRequiredTextFields parameter where field validity would be set to failed if the required field value is missing in Visual OCR.

  • Completely redesigned camera focus control implementation

  • Added new values to the eVisualFieldTypeparameter

    • FT_DLCLASSCODE_CA_NOTES

    • FT_CITIZENSHIP_STATUS

    • FT_MILITARY_SERVICE_FROM

    • FT_MILITARY_SERVICE_TO

  • ​DocumentReaderResults now contains an array of DocumentPosition, BarcodePosition, MRZPosition, and ImageQualityGroup items with information for each document page that was read, where available

  • Added interactive documentation inside the SDK (Option + tap in XCode)

  • ​getTextFieldValueByType function call with LCID parameter returns nil if the value with such LCID is not presented in the results

  • RFID chip reading works correctly without first scanning the document, if the MRZ or CAN key is pre-set

  • Fixed the operation of the completion block in the stopRFIDReader function

  • Fixed the rendering position of the help animation and the multipage mode animations​

  • Fixed the position of custom statuses (status and resultStatus), now they are displayed in the center

  • Fixed issue with Capture scenario when using the 1120 device

Core:

  • Document detection and crop

    • Fixed issue with locating already cropped documents in FullAuth scenario

    • Fixed issue with locating documents on backgrounds with multiple lines aligned with the document edges

    • Fixed issue with locating two sides of ID card on one image

    • Fixed issue with locating documents on non-contrast backgrounds

  • OCR

    • Greatly improved bank cards OCR

    • Improved OCR of perforated fields

  • MRZ

    • Added new parsers:

      • Pakistan ID card

    • Updated parsers:

      • Austrian residence permit

      • EU residence permit

      • France ID card

      • Kazakhstan Residence permit

      • Malta ID card

      • Norway ID card

      • Poland ID card

    • Fixed issue with incorrect MRZ detection on some images

    • Fixed issue with incorrect symbol probabilities in some cases

    • Fixed issue with 1-line MRZ in DL parsing

  • Barcode

    • Added new parsers:

      • Belarus ID card

      • Colombia DL

      • France ID card

      • Ghana Voter card

      • Nigeria Voter card

      • Pakistan ID card

    • Updated parsers:

      • Korea Alien registration card

      • Moldova ID card

    • Fixed issue with code types for recognition filter

  • Text data parsing and validation

    • Added capability to verify date of issue validity based on document template activity term

    • Added capability for Italy tax code validation

    • Added capability for comparison of the field value from one source with abbreviature of this value from another source

    • Added XXK nationality code for Kosovo

    • Fixed issue with comparison between MRZ and Visual for document without expiration date

    • Fixed issue with probabilities for transliterated fields: they will have probability of original field

    • Fixed issue with German Ü letter transliteration

    • Fixed issue with composite surnames comparison where last part is missing in MRZ

    • Fixed issue with measure units on locale change

    • Fixed issue with FT_DATE_FIRST_RENEWAL and FT_DATE_SECOND_RENEWALfields validity

  • Image QA

    • Improved focus detection NN

    • Improved algorithm for detection of colorless images

  • Authenticity

    • Added new algorithm for obtaining correct IR image on 7310 devices

    • Implemented correct processing of all authenticity checks in FullAuth scenario

    • Improved IR visibility check for photo area, where grayscale photo should look the same in visible and IR

    • Improved variable patterns check for documents with a lot of UV luminescence

    • Fixed issue with hologram checks on series of frames

  • Added support of reading data from both sides of bank cards

  • Added support of single page (back page) data bank cards

  • Improved processing of bank cards: now it is payment system agnostic

  • Fixed issue with input images page indexes. Now they will be set automatically and correctly if missing

  • Code quality improved with lots of small issues fixed

  • RFID

    • Fixed issue with different .ldif format support

    • Fixed issue with parsing certificates on different platforms

Database:

  • Added to templates in whole database where applicable:

    • Perforated text fields

    • Hologram areas and checks

    • Activity term for template

    • Extended description for visible patterns

    • Barcode format check parameters

    • IR visibility for portraits, where they are the same in visible and IR

  • 248 countries and territories / 9331 documents included

  • 179 new documents added

January 27, 2021 - v5.4

API:

  • Added document authentication based on the detection of the presence of a hologram​

  • Updated document image for NFC chip reading activity animation for USA passport which can be read only when passport book is open

  • Added the ability to disable checking for the presence of a flashlight on the device

  • Added new values to the eCheckDiagnose parameter:

    • PHOTO_WHITE_IR_DONT_MATCH

    • FIELD_POS_CORRECTOR_GLARES_IN_PHOTO_AREA

    • HOLOGRAM_FRAMES_IS_ABSENT

    • HOLOGRAM_HOLO_FIELD_IS_ABSENT

    • FINGERPRINTS_COMPARISON_MISMATCH

    • HOLO_PHOTO_FACE_NOT_DETECTED

    • HOLO_PHOTO_FACE_COMPARISON_FAILED

    • HOLO_PHOTO_FACE_GLARE_IN_CENTER_ABSENT

    • HOLO_ELEMENT_SHAPE_ERROR

    • ALGORITHM_STEPS_ERROR

    • HOLO_AREAS_NOT_LOADED

    • FINISHED_BY_TIMEOUT

  • Updated enumeration values of eImageQualityCheckType​

  • Updated DocumentReaderException enumeration values related to licensing errors that may occur during initialization

  • Fixed an issue with incorrect camera focus in camera activity

  • Fixed an issue in the zoomFactor property. Now when setting a value above the maximum allowed value, the camera will be zoomed to the maximum available value for this camera

Core:

  • Document detection and crop

    • Added Boolean flag processParam.fastDocDetect to shorten the list of candidates to process during document detection in a single image process mode. Enabled by default. This solves the issue introduced in 5.2, that lead to significantly longer processing times on the images with specific backgrounds

    • Improved ID2 MRZ documents cropping in some cases

  • OCR

    • Added support for Amharic (fidäl / Ephiopian)

    • Added support for perforated numbers OCR (different fonts, straight and mirror)

    • Added Boolean flag processParam.checkRequiredTextFields. When enabled, each field in template will be checked for value presence and if the field is marked as required, but has no value, it will have "error" in validity status. Disabled by default

    • Added Boolean flag processParam.updateOCRValidityByGlare. When enabled, OCR field validity will be updated with | tr_GlaresOnField value, if there is a glare over the text field on the image. Disabled by default

    • Improved Latin NN

    • Improved Hebrew NN

  • Barcode

    • Updated Peru ID Card parser

    • Fixed issue with multiple pages processing

  • Text data parsing and validation

    • Implemented conversion from ROC dates (Minguo) into Gregorian

    • Fixed issue with DL class field comparison between barcode and visual

    • Fixed issue with incorrect validity for dates in Hijri

  • Image QA

    • Added new Resolution assessment check. Should be above threshold (150 ppi), otherwise the check will fail. Can be set via processParam.imageQA.dpiThreshold

    • Added new Perspective angle assessment check. Should be below threshold (5 degrees), otherwise the check will fail. Can be set via processParam.imageQA.angleThreshold

    • Added new Documents bounds inside the image check. All document corners should be inside the image, otherwise the check will fail

  • Authenticity

    • Added support for hologram presence detection by analysis of multiple frames

    • Improved patterns check for visible and IR light images

    • Improved patterns check for Russian Federation domestic passport

    • Fixed issue with AV on specific images in IR visibility check

  • Fixed issue with 1- and 8-bit images processing

  • Code quality improved with lots of small issues fixed

Database:

  • 246 countries and territories / 9152 documents included

  • 193 new documents added

  • For details see Supported documents list​

December 3, 2020 - v5.3

API:

  • Added the ability to change the zoom on the scanning screen (isZoomEnabled)

  • Added the ability to set the starting zoom factor for the scanning screen (zoomFactor)

  • Added the ability to change the size of the toolbar (toolbarSize)

  • Added the ability to get cut-out images of barcodes when working without a document description (returnCroppedBarcode)

  • Added new localizations: Bengali, Dutch, Finnish, Greek, Hebrew, Hindi, Japanese, Korean, Malay, Swedish

  • The jsonResult property has been replaced with the rawResult string property in DocumentReaderResults, which contains the original JSON string

  • Added new values to the eVisualFieldType enumeration

  • Deprecated eVisualFieldType enumeration value

  • Added new values to the eCheckDiagnose enumeration

  • Fixed a bug with the incorrect return of an error when processing an RFID chip.

Core:

  • Document detection and crop

    • Improvements in single image process algorithm with more precise cropping

    • Fixed issues with incorrect cropping by inner document lines instead of outer in some cases

    • Fixed issue with cropping large document formats like birth/marriage certificates

    • Fixed issue with incorrect cropping documents with MRZ in some cases

  • Document type recognition

    • New functionality implemented in template description that allows more precise document cropping and resolution determination

    • Fixed issue with MRZ filter for IDL

    • Fixed issue with documentIDList filter working incorrectly

  • OCR

    • Added support for Farsi, Kashmiri, Konkani, Marathi, Sanskrit, Sindhi, Punjabi languages

    • Arabic, Hindi, and Thai languages reading quality greatly improved

    • Added Arabic and Persian numbers correct support

    • Added capability to ignore field mask on recognition when matchTextFieldMask is set to false. By default, we are trying to read a field value with maximum efforts to match the mask and provide a correctly formatted value, making assumptions based on the provided field mask. In case this parameter is disabled, OCR will be done as is and then the recognized value will be matched to the field mask for validity

    • Updated Cyrillic NN with better recognition probability

    • Fixed issue with Latin characters in right-to-left strings (Hebrew, Arabic, etc.) leading to incorrect text direction (RTL, LTR markers)

    • Fixed issue with incorrect recognition probability (over 100)

  • MRZ

    • Added new parsers:

      • Malaysia visa

      • South Sudan ID car

    • Updated parsers

      • Azerbaijan Permanent Resident Card

      • Iceland Residence permit

      • Kenia ID card

      • Mexico ID card

      • Mexico Voting card

      • New Zealand passport

      • Peru ID card

      • USA & CAN Enhanced Driving License

      • USA Permanent Resident Card

      • USA Visa

    • Improved reading of images with color noise over MRZ

    • In IDL with 1-line MRZ, the first character “D” is now always parsed as Ft_Identifier instead of Ft_Document_Class_Code

  • Barcode

    • Added output of the Gf_BarCode graphic field with barcode image for unknown documents, where the barcode was detected and read

    • Updated New York Enhanced Driving License parser

    • Updated US DL parsers – zip code is output with trailing zeros

    • Updated US DL parsers – full names are divided into parts using spaces and commas correctly

    • Fixed issue with Peru ID card data parsing

    • Fixed Brazil ID card consular parser

    • Fixed issue with Code 128 decoding on some samples

    • Fixed rare issue with an extra character added on PDF417 decoding

    • Fixed issue with data parsing of a specific boarding pass leading to an exception

    • Fixed issue with reading multiple codes in series processing, when one code read on one frame, and the other on a different frame

    • Removed Ft_PDF417Codec field generation

    • Optimized processing in series and with multiple-page documents

  • Text data parsing and validation

    • Added correct LCID for fields from RFID DG11 and DG12 for all countries instead of always Latin

    • Added check digit calculation for the Swiss health insurance card

    • Added parsing possibility for Australia DL categories that are dynamically printed on the document

    • Fixed issue with visual vs MRZ comparison for field values that contain Ü. For some countries, MRZ contains UE and for some U

    • Fixed issue with Slovenian Ć transliteration for MRZ comparison

    • Fixed issue with Russian Domestic passport date of expiry validation if the passport was renewed just before 45 years of age

    • Reverted ^ symbol removal in Ft_Surname, Ft_Given_Names and Ft_Surname_And_Given_Names in lexical analysis results. If you need to remove this symbol from a field value, please do this on the client’s side, as this symbol helps to distinguish, for example, between the first and second surnames in some documents

  • Authenticity

    • Fixed issue with Barcode format check result was not available in case barcode should be present in document but cannot be detected or read

  • Added parameter imageDpiOutMin for input resolution adjustment in case, it is too small. By default, the image will be resized up to 200 dpi if it has less resolution. This parameter cannot be less than 100

  • Improved PNG saving speed

  • Fixed issue with incorrect rotation after incorrect face detection

  • Fixed input JSON validation for the correctness

  • Fixed issue with process timeout when processing single images. Now it is ignored

  • Fixed issue with no graphic results available in output if timeout reached

  • Fixed issue with the calculation of PA status by also checking ReadingStatus of the files

Database:

  • 246 countries and territories / 8983 documents included

  • 185 new documents added

  • For details see Supported documents list​

Previous
Troubleshoot license issues
Next
Requirements
Last updated 2 weeks ago
Contents
April 26, 2021 - v5.5
January 27, 2021 - v5.4
December 3, 2020 - v5.3