What are FAR (False Acceptance Rate) and FRR (False Rejection Rate) and what are the speeds of the technology provided?
FAR (False Acceptance Rate)
- A percentage that indicates how a biometric system identifies different features as similar.
- This is a safe level, the rate of false acceptances, and should be as low as possible, meaning that any user who has different features from those recorded manages to evade the system.
FRR (False Rejection Rate)
- Percentage that reflects the same features are seen by a biometric system as different.
- Represents a level of ease of use, the rate of false rejections, and must be as low as possible: means that a user who has its own characteristics recorded is always correctly recognized by the system.
FAR systems Eter: 1/1.000.000
FRR systems Eter: 1/1.000
What is the level of security?
The security level is measured by the score threshold obtained during recognition. If you set the level of security, ie points higher safety threshold will reduce the FAR but increase the FRR. Referring to the figure below, the area of intersection between the X-axis, the curve B and the threshold of the line (Area B) indicates the FAR, while the area of intersection between the X-axis, the curve A and the threshold of the line (area A) indicates the FRR. Then moving the line of threshold to the right increases the level of safety: the area B is reduced while the area A is expanded. The level of security will be set differently depending on the applications.
Security level - FAR and FRR
The following figure represents different levels of security for different needs of different market segments and curve FAR systems Eter compared with the curve 1/FRR.
Specifications of APIs for different market segments
This graph shows the specific FAR and FRR requested by various market sectors such as protection of financial transactions, medical applications, access control and timekeeping, identity card digital secure information exchange network, e-commerce, etc..
Some applications require higher levels of security, that is a FAR lower, while others require more ease of use, ie a lower FRR. For example, access control, ID card digital exchange of information in the network and secure e-commerce require FAR less to get greater security, while accepting a higher FRR; health and timekeeping applications require a more FRR low while also tolerate a higher FAR. The pink full line indicates the performance of the systems Eter to test for recordings of tracks made to the quality optimum one, namely to security level A. The performance meets the requirements of all applications listed.
Yes, our systems are already installed more than ten years in different contexts such as government agencies, military organizations, companies, hotels and clubs, private homes, for more detail go to the reference section.
Used by many users in real conditions of daily use and different application contexts, are very reliable and quick recognition, consisting only in a maintenance routine cleaning of the sensor.
Mainly FAR and FRR, along with the time of registration, identification and verification. It 'also important to the maximum number of users that are allowed to enter into the database, ie the maximum number of biometric features (fingerprints, irises, ...) between which the system recognizes the user to reliably and quickly, without the use of PIN codes, proximity cards or badges.
After the image has been captured, the process that extracts the minutiae digitally from an image and compares it with a similar database employed in less than a second. The longer time is spent on the extraction. If you also counts the time to capture the image, the system employs about 1 second to complete the verification. For a database of 1500 people with 2 fingerprints each registered (so a total of about 3000 fingerprints in the system), we can give less than 3 seconds to complete the identification. For databases that are larger than would take more time, then it is appropriate to use a keyboard to make groups of users or a proximity badge used in combination with the impression and thereby speed up the operation of identification.
The standard size varies from 256 bytes to 750 bytes per fingerprint, depending on the characteristics of the system, including data on singularities (minutiae) and description.
6000 fingerprint access control, recognizing one-to-many (identification) is carried out in a short time up to 3000 tracks, and it is advisable to use a keyboard to groups of users. To protect your PC and data logic there are no limits except the memory of the system, but the recognition one-to-many (identification) is performed reliably up to 100 fingerprints, and you must also use the user name for an indexed search on the database or a token (such as a smart card) that contains the template.
Also, given that the fingerprints of a person never vary from birth to death and always differ from individual to individual, even in the case of identical twins. It is necessary to repeat the recording only in the case where a wound permanently compromise the use of the finger or render unreadable the singularities detected during recording, erasing deeply.
Yes, via RS232 network (up to 20 mt.), RS485 (up to 1 km) or TCP / IP Ethernet network.
No, it can happen.
The information contained in the templates stored in the database are extracted from specific algorithms and only one set of numbers that can not play. You can not remove them if you do not have access to the system, and in any case the identification is carried out in an embedded system with dedicated functions and closed, so it is impossible to provide the system with only the template previously subtracted.
Furthermore it is impossible to reconstruct the image of the impression starting from the template, which is only a sequence of numbers formed by injecting functions, and therefore not reversible.
In fact, the logic of operation of the systems is as follows:
- Acquisition of the biometric feature
- Elaboration of the same instant by the software, and immediate transformation in numerical code
- Record the code number of a contact smart card or proximity that the user carries
- Use only the numeric code for the recognition of the person
The extraction process of the numeric code from the image of the fingerprint or of any other biometric feature can not be reversed, thereby making it impossible to reconstruct the image of the biometric insignias from the numeric code. Therefore, the protection of personal data of users of our systems for access control application is fully guaranteed.
3DES (Data Encryption Standard) algorithm with 128 bit keys.
When your finger is resting on the lens of the optical sensor, the ridges of the surface of the impression touch the optical prism, and is always a small distance between the valleys of the surface of the impression and the lens. The light is then reflected at different angles, as indicated in the figure above. Only the reflected light of 90 degrees is received by the CCD sensor, and then the CCD is capable of capturing the fingerprint image (the bright parts are the valleys of the impression, the black portions are the ridges).
Yes, the API of our systems are BioAPI-compliant and therefore are compatible, portable and can be interfaced with other biometric devices.
Image Capturing Module
Module for image capture
Supports the image capture from different input devices such as scanners and sensors chip
Captured images can be single retained in the memory buffer or file saved in bmp format or raw, grayscale or binary mode
Provides an "auto-snap", that is, the impression is captured automatically if the quality is considered good enough, otherwise the function returns the status of the impression (too dry, too wet, not well placed)
Provides a diagnostic function for controlling the device
Gain Control Module
Form gain control
Supports the function that adjusts the parameters of the image so as to obtain the best quality of different devices (the parameters are different depending on the Tracking Device): for example, it is possible to adjust the brightness and contrast differently, or the offset and the range, so as to obtain the desired image quality
Minutiae Extraction Module
Module for the extraction of minutiae
Supports the function that extracts the singularity (minutiae) from an image captured by different devices impression
Extracts only the significant details and eliminates the details of disurbo and edges
Create a template with a size between 100 ~ 800 bytes
Fingerprint Enrollment Module
Form for registration of fingerprints
Supports the function that provides recording from singularities extracted from Minutiae Extraction Module obtained from three images of the impression: This algorithm combines the singularities of three prints and creates a template final, which will contain the latest information on the fingerprint with size between 100 ~ 800 bytes
Fingerprint Verification Module
Module for fingerprint verification
Supports the function for verifying one by one of the registered fingerprints
Supports functions that allow the experience one to many, that is, if the impression matches a template stored in the database
The module provides a cross-check with high performance and high speed as long as the singularity is extracted from any detection device that supports 500 dpi
Provides different levels of security for applications that require different FAR / FRR
The architecture of the SDK is represented in the figure below:
Includes several examples of how to integrate technology with biometric access control systems Eter or timestamps. System integrators can use SDK to develop systems for specific applications such as ATM, POS, credit card terminals, automotive, authentication, networking, etc.