EN
RU

VOCORD FaceControl is a non-cooperative face capturing and recognition system.

Non-cooperativeness means that a system doesn’t require system-man cooperation in any way in order to facilitate recognition of that person. Therefore, VOCORD FaceControl system is used successfully for facial recognition purposes in crowded areas – subway stations, railroad stations, stadiums, concert halls and other public areas where people can not be forced to interact with the biometric identification system.

Facial recognition system VOCORD FaceControl:

  • Captures faces of persons in a video stream
  • Matches captured faces against watch lists and identifies them in real time
  • Alerts an operator upon successful facial recognition
  • Archives all photos of captured faces
  • Allows of searching particular subjects of interest or suspects in facial databases and watch lists.

Public safety ensuring

VOCORD FaceControl is successfully utilized for security purposes in public areas. For that end, the sites are equipped with specialized facial recognition cameras. The check points are deployed at the entries and exits or in narrow passages (underground and overhead walkways, corridors, etc).

Unlimited number of databases can be connected to the system. For example, databases of local police, Federal Security Service of Russia and court marshals. At appearance of the person in search, an operator receives a notice specifying the matched database and comments if any.

Using VOCORD FaceControl systems, operators can not only run searches in the existing databases but also populate them. In respect to safety of sports and other facilities, petty offenders often visit the same areas repeatedly. VOCORD FaceControl system allows to receive a portrait reference database containing high quality photos of all individuals passing the check point. Using this database you can on an operational basis and with a high probability detect an suspicious individual next time at his/her appearance at the site.

Using the above approach, you can make the best use of the system in large-scale retail chains for the purposes of operational notification to security guards of appearance of shop breakers.

This permits to search the photo of person of interest against portrait reference database . The system analyzes the archive automatically and detects all events of appearance of the person of interest at the facility.


Can the facial recognition system recognize a person with a painted face or looking under the angle in relation to the camera?
Yes, if it is new generation system VOCORD FaceControl 3D.

Access control

To ensure access control for secure facilities and special protected sites, the standard access control system based on personal proximity-cards can be insufficient. Using VOCORD FaceControl facial recognition system you can increase control level. Special facial recognition cameras are installed at the check points, and the system is integrating with a standard access control system.

The system will grant access upon receipt of positive results of matching against the “Whitelist”. Otherwise, the system will notify a security officer of presence of a stranger. “Reinforcement” of a security system by integrating with a facial recognition technology not only raises an access control level but also offers additional protection against unauthorized persons entering building using valid proximity cards owned by others.

VOCORD FaceControl system contributes to investigations of accidents as it stores the photos of all people which passed the check point. In case of an unexpected accident, you can always look through the portrait reference database over a particular period of time and find out persons who visited the facility.

VOCORD FaceControl system operation can be organized in an “interfaceless mode”. For example when integrated with an access control system, VOCORD FaceControl just confirms or declines access and registers all access facts in the access control system log; at that, an operator uses only the access control system interface.

System interface

> Advantages Up

Non-cooperative facial recognition

Unlike a number of biometric systems requiring accurate positioning of the person’s face in relation to the camera, VOCORD FaceControl works in a completely noncooperative mode. In other words, the subject should not take up a position in the camera view – the system captures faces of subjects in a crowd or pedestrian flow, the subjects may be completely unaware that they are under video control.

This allows VOCORD FaceControl to work effectively not only as an access control system but also as a security system in the areas difficult to ensure cooperation of people and recognition systems – subway stations, railroad stations, stadiums, cinemas, stores and other public areas.

Video cameras with an inbuilt face detection

VOCORD FaceControl system uses specialized video cameras VOCORD NetCam4 K-series with an inbuilt face detection and capturing algorithm. Such camera transfers video stream containing only captured facial images. This allows:

  • To reduce dramatically the data channel load. The camera broadcasts just only facial images, so using of a wideband channels is not necessary. The camera-server communications can be performed via Wi-Fi and even 3G networks.
  • To reduce load on servers through performing a number of operations directly inside the camera.
  • To obtain high-quality facial images in any conditions including difficult lightening – backlight or contrasting side lighting. Images quality can be improved through using dynamic ROI (Region of Interest), video data capacity conversion 12 bit to 8 bit and other automatic image processing algorithms.


Видеомодуль VOCORD NetCam4 с Wi-Fi для систем распознавания лиц


VOCORD NetCam4 cameras allow simultaneous capturing of up to 16 faces per frame.

VOCORD NetCam4 cameras are specifically developed for facial recognition systems. Except algorithms of adaptation to video analytics tasks, these cameras support multicast broadcasting and interfaces with peripheral equipment, for example, with access control system controllers.

High recognition accuracy

  • Only high-quality images can be used for recognition purposes
  • While the person passes through the control zone, the system does a series of facial snapshots from which the best are selected
  • The system is based on the most advanced recognition algorithms

It helps to achieve very high recognition accuracy. The average probability of the correct recognition by VOCORD FaceControl system in real conditions is above 90 %.


For estimation of biometric system accuracy, common interdependent factors FAR/FRR (False Acceptance Rate/False Rejection Rate) are used. But when declared FAR/FRR error is 0.1–0.2 %, it doesn’t mean the system will recognize people with 99.8–99.9 % confidence in real conditions. The figure will come out right for recognition in ideal conditions: recent high-quality photo in reference database,, the person is taken frontally in relation to the camera and his/her facial expression is neutral. Such conditions are possible when people “cooperate” with the system like in facial recognition systems used for access control purposes. In facial recognition systems used for public security at railroad and subway stations, the accuracy of 80–90 % is a very good rate because people don't behave ideally in front of the camera. In order to decrease the impact of such negative factors as non-frontal position in relation to the camera, facial expression, artifacts, Vocord Company has developed a new generation biometric identification system – VOCORD FaceControl 3D.

Compliance with biometric standards

VOCORD FaceControl is a one of a number of facial biometric systems which comes up to IEC 19794-5-2006 “Automatic identification. Biometric identification. Biometric data interchange formats. Facial image data.” for quality of digital video data suited for execution of automatic biometric identification by facial image.

< Description > Architecture Up

VOCORD FaceControl has a geographically distributed architecture. In small-scale systems the whole installation may be deployed on a single server. In case of a geographically distributed system, each software module have be installed on a separate server and there can be an unlimited number of servers inside the installation network.

The system is available in 2 variants: (1) variant based on VOCORD NetCam4, K-series cameras with inbuilt facial detection and capturing algorithm, and (2) variant based on VOCORD NetCam4, D-series cameras.

System architecture based on K-Series Cameras

shema_fc_new_K_site.jpg

System architecture based on D-Series Cameras


fc02.jpg

Components

1. VOCORD NetCam4 K-series
Video cameras, 1.4 Mps with an inbuilt face detection and capturing algorithm. They transfer captured facial images in a separate video stream. These VOCORD NetCam4 cameras support multicast broadcasting, for example, for video data archiving.

2. VOCORD NetCam4 D-series
Video cameras, 1.4 Mp. These VOCORD NetCam4 cameras support multicast broadcasting, for example, for video data archiving.

3. FC Catcher
Software module capturing facial images from a video stream. Up to 4 video cameras can be connected to each server equipped with FC Catcher software module.

4. FC Grinder
Software module defining biometric parameters of captured face images and calculating the biometric templates. Further these patterns will be used for matching against the watch list.

5. FC Matcher
Software module which compares biometric templates with images in the watch list and decides whether two images coincided and what is the recognition accuracy.

6. FC Archive
Software module for long-term video data archiving. It stores all images, biometric templates, system logs, etc.

7. FC Client
Software module for an operator’s workstation. It allows the operator to view live- video images from different sites, receiving successful face recognition alerts as well as performing archive retrieval and searching.

VOCORD NetCam4, K- and D-series – comparative specifications

Parameters

K-Series

D-Series

Inbuilt facial capture algorithm

Yes

No

Data transmission channel requirements

10 Mbit/s

300 Mbit/s

Server computational capability requirements

Lower
(6 cameras per CPU Core i7)

Higher
(4 cameras per CPU Core i7)

Price

Higher

Lower

< Advantages > Technical characteristics Up

Parameter

  Value

Architecture

Geographically distributed. Unlimited number of servers and video cameras.

Communication interface

TCP/IP

DBMS

For a single-server system: SQL Express (free version). For a multiserver system: commercial versions of SQL.

Video cameras

VOCORD NetCam4: 1.4 MP, 2/3”

Rate of facial capturing in a video sequence

About 60 ms per frame

Identification rate

100,000 comparison operations per second

Reference database size

Unlimited

Required illumination

At least 200 lx

Interpupillary distance requirements

At least 60 pixels (at face size of 140 pixels); at least 120 pixels (at face size of 280 pixels)

Operating temperature range

from 0 to +50 °С


Parameters recommended to improve recognition accuracy:

  • Video camera-to-object distance – maximum 8 m
  • Diffused light – at least 200 lx
  • No backlight, swings in light-and-shade
  • Maximum angle of deviation of the person’s face from the frontal position – 15˚
  • Requirements to photos: interpupillary distance – at least 60 pixels, recommended – at least 120 pixels

< Architecture Up
VOCORD FaceControl. Product Video
Loading Player
How to Choose Cameras for Facial Recognition
Loading Player

Reviews

Success Stories

Projects