Smart PassportReader

2016: PassportReader 2.0 is support Elbrus

On March 29, 2016 the Smart Engines company announced release of the version of a system of recognition of the Russian Federation citizen passport of Smart PassportReader 2.0 and support in the products of processor architecture Elbrus from MCST company.

In the version of PassportReader 2.0 are implemented recognition of the passport, as in portrait, and landscape orientation for mobile devices, recognition of a turn of the passport using webcams. Feature of version 2.0 - support of architecture of the Elbrus microprocessor, along with ARMv7-v8 (AArch32 and AArch64), x86 and x86_64.

The solution based on Elbrus will allow to use a Russian Federation citizen passport sensing technology for problems of creation of hardware and software systems of personal data processing with increased requirements on information security and technology independence.

Development of algorithms of recognition and syntax/geometrical interframe integration allowed to increase quality and speed of recognition. Considerably the quality of recognition of Russian Federation citizen passports with faintly printed data increased. The carried-out works on optimization of architecture and a program code of a core of recognition provided improvement of support of multinuclear configurations for multi-core mobile and desktop processors.

In version 2.0 functions of extraction of text data, the images of the document entirely given about location of fields, images of text and graphic boxes (the photo, the signature) are implemented. At the same time for each text box in a system there is a possibility of a request of reliability assessment of its recognition.

For increase in convenience of integration of functionality of recognition of the Russian Federation citizen passport into a set of SDK Smart PassportReader 2.0 are included a component of work with the video camera for Windows/Linux and an ActiveX-component for integration into different information systems 1C (for example).

Updating is provided for users of the previous generation of a system free of charge on demand.

2015

Basic information

Smart PassportReader is the program interface (API) allowing software developers to embed functions of recognition and processing of passports in desktop and mobile applications.

API processes scans, photocopies and photos of the passport. Key technology difference of Smart PassportReader - an opportunity to distinguish passport details in a video flow of the camera of the mobile device or webcam in real time.

Smart PassportReader is intended for verification of the personality in services of service of mobile payments and the systems of mobile registration.

Creation based on the Smart PassportReader tools of new user services will help integrators and software developers to implement essentially new level of convenience of using the Internet / mobile applications.

For example, for purchase of the ticket for the train/airplane or registrations of armor in hotel, in mobile application it is necessary to enter passport data - a full name, a series and the passport number, date of a rozhedeniye, etc. that on the mobile device to do long and inconveniently. Using Smart PassportReader it is possible to enter passport details in 1-3 seconds.

Control and management systems for access (ACS) on the Smart PassportReader platform it is possible to refuse use of the passport (tablet) scanner having replaced it with  the normal modern smartphone or having connected the webcam to the computer. Besides, it is possible to exclude scanning and manual entry of passport details in points of service or at departure of agents to clients.

High quality of recognition and speed of work on mobile devices became possible thanks to the advanced scientific achievements of researchers of Smart Engines Ltd. in the field of image processing, recognition and training of neural networks, and also deep optimization of architecture and a program code of a core of recognition.

In the course of recognition of the passport search and selection of limits of the document in a video flow is performed. The algorithms which are based on syntax and geometrical integration of personnel information allow to achieve qualitative result already to the fifth recognizable frame.

In the presence on the passport of a machine-readable zone (MChZ) of Smart PassportReader makes its reading and correspondence between results of recognition of normal fields and MChZ. For recognition of MChZ the core of own development is used.

Representation of Smart PassportReader, 2015

Recognition of the scan of one page of the passport on the personal computer with the Intel Core i7-2600 CPU configuration @ 3.40 GHz, 8 GB of the RAM, Windows 7 SP1 64bit occupies on average 0.087 with, and iPhone 5s copes with this task for 0.23 pages. At the same time the quality of recognition of a series and number for the scanned images is more than 99%, and the Full Name - 95%, the quality of recognition of the 3-second video clip containing the passport is ^[1].

Smart PassportReader sensing technology multiplatform also works at the majority of widespread operating systems for mobile devices, desktops and servers: iOS, Mac OS X, Android, Linux, Windows, Windows Phone and main hardware platforms of ARMv7-v8 (AArch32 and AArch64), x86 and x86_64.

Screenshot of Smart PassportReader, 2015

Smart PassportReader 1.2

The Smart Engines company announced in September, 2015 release of Smart PassportReader 1.2 (API) which allows software developers to embed functions of recognition and processing of passports in desktop and mobile applications.

Key features of this version — recognition of the 2nd page of the passport on mobile and stationary devices and also processing of photos, scans of a turn and separate pages of the Russian Federation citizen passport in one flow.

Now using Smart PassportReader it is possible on the image (received by scanning or photography) or in a video flow (mobile devices, webcams) to take not only fields: "series", "number", "full name", "floor", "date of birth" and "birthplace", but also fields: "by whom it is issued", "date of issue" and "subdivision code".

The Passport AnyWhere technology developed by the company which is able to find the Russian Federation citizen passport located at any angle then recognition of data is made is applied to data entry of the Russian Federation citizen passport from the images received by scanning or photography in version 1.2.

In the new version it was succeeded to increase quality of recognition of fields and a machine-readable zone (MChZ) of the Russian Federation citizen passport by 20% due to increase in accuracy of search algorithms and character recognition. At the same time it was succeeded to reduce time of recognition of one image for 10% both by mobile, and on stationary devices.

Smart PassportReader can be demanded for increase in convenience of data entry of the Russian Federation citizen passport in payment services and recording systems in airlines, banks, insurance companies, hotels, the e-government systems, etc.

Technology difference of Smart PassportReader is an opportunity to distinguish passport details not only on the image, but also in a video flow. In case of use of API in mobile applications, all calculations are performed directly on the device (on-devise) in real time. Data are not transferred to recognition to "cloud" or on the server that not only ensures safety of personal data processing, but also gives an opportunity for work of mobile application at low quality of communication or in lack of the Internet.

All technologies which are the cornerstone of Smart PassportReader, since identification of the document and finishing with recognition of the character, are own development of Smart Engines company.

The Smart PassportReader tools provide multi-platform architecture and work at the majority of widespread operating systems for mobile devices, desktops and servers: iOS, Mac OS X, Android, Linux, Windows, Windows Phone and main hardware platforms of ARMv7-v8 (AArch32 and AArch64), x86 and x86_64.

Notes