Application Analysis of New Fingerprint Recognition Sensor

Single-touch sensors and swipe sensors are two new types of solid-state fingerprint sensors, both of which collect information through changes in capacitance during touch. This article analyzes the working principles and characteristics of the two types of sensors in detail, and introduces their applications in Internet security authentication, car keyless entry systems, etc.

There are two types of solid-state fingerprint sensors on the market: the first is a single-touch sensor that requires a finger to reliably touch the fingerprint collection area; the second requires a finger to rub across the sensor surface, and the sensor will collect a specific set of data, which will then be quickly analyzed and authenticated. These two types of fingerprint sensors will be increasingly widely used.

The working principle of the above two types of sensors is: when the raised part of the fingerprint is placed on the sensor capacitor pixel electrode, the capacitance will increase, and data collection is performed by detecting the increased capacitance. The pixel points in the sensor are 45 square microns, the interval is 50 microns, and the resolution of the capacitor pixel array is slightly higher than 500dpi. This type of sensor is based on a standard single-polysilicon three-layer metal CMOS process and is designed using a 0.5-micron process.

The third layer of metal interconnects, which forms the capacitor pixel layer, is made of titanium nitride and covered with a layer of silicon nitride, just 7,000 angstroms thick. This combination of hard metal electrodes and abrasion-resistant coatings makes for a very rugged sensor that can last for many years.

Fingerprint Detection

Human fingerprints are composed of closely adjacent concave and convex lines. By using a standard reference discharge current at each pixel, the fingerprint line condition can be detected. Each pixel is first pre-charged to a certain reference voltage and then discharged by the reference current. The rate of change of the voltage on the capacitor anode is proportional to the capacitance on it in the following relationship:

Iref=C×dv/dt

Pixels under the ridges of the fingerprint (high capacitance) discharge slowly, while pixels under the valleys of the fingerprint (low capacitance) discharge quickly. This different discharge rate can be detected by a sample-and-hold (S/H) circuit and converted into an 8-bit output. This detection method has high sensitivity to the ridges and valleys of the fingerprint and can form a very good original fingerprint image.

Fingerprint recognition can be performed using complex software algorithms through the process shown in Figure 1. This software collects the original fingerprint image, digitizes the image information and extracts the detail template, and then tests to determine whether the extracted detail template matches the reference template.

Comparison process

Single-touch sensors differ from swipe sensors in size and cost. Contact sensors are larger, typically with an effective contact surface of 15 x 15 mm, and can quickly capture the largest fingerprints or thumbprints. They are easy to use and can quickly transmit the entire fingerprint image at 500 dpi (the standard for automatic fingerprint recognition).

These sensors have been designed and are being used in fingerprint recognition by US government agencies and police departments. In the near future, they will also be gradually used in automobile one-touch keyless entry systems and emerging national security applications.

The sensor consists of 256 (columns) × 300 (rows) of micro-metal electrodes, each column connected to a pair of S/H circuits. The fingerprint image is collected row by row, and each metal electrode acts as one pole of the capacitor, and the finger in contact with it is the other pole of the capacitor. There is a passivation layer on the surface of the device, which acts as a dielectric layer between the two capacitor poles. When the finger is placed on the sensor, the bumps and depressions on the fingerprint will produce different capacitance values ​​on the array and form a complete image for authentication.

Swipe sensors are a new type of fingerprint acquisition device that requires the user to swipe their finger across the device. The advantages of swipe sensors are small size (e.g., Fujitsu's MBF300 measures only 3.6×13.3 mm2) and low cost. These devices are mainly used for embedded security identification applications in mobile devices such as mobile phones and PDAs. Sophisticated image reconstruction software quickly captures multiple images from the sensor at a speed of nearly 2000 frames/second and organizes the data details of each frame together.

[page]

Information and certification

There is no doubt that portable and low-cost fingerprint recognition technology has a profound impact on our lives. For example, in the future, the police can stop a suspect in a high-crime area and ask him to provide fingerprints instead of ID cards or driver's licenses. The person places the first, second or third finger of his right hand on a sensor connected to a wireless PDA, which can quickly compare the suspect with previous criminal records.

This identification technology is also beneficial for users of stolen mobile phones. When the mobile phone is turned on, the user is required to go through a quick authentication process. The user slides his finger across the sensor. If the authentication is passed, the user is authorized to use the various functions of the mobile phone. If the user is not an authorized user, the mobile phone will remain locked. If the authentication fails several times in a row, the mobile phone will delete the key information in the memory and shut down.

In the voicemail application, after dialing a voicemail number, the user only needs to swipe his finger across the sensor for the system to recognize it. With fingerprint recognition, there is no need to use email passwords or personal identification numbers.

In future automotive applications, users can input fingerprint samples of family members and drive after authentication. The registration process is very simple: each authorized driving member places his finger on the sensor and sets various parameters of the car according to personal preferences, and then stores these settings in the car's computer memory.

When the driver enters the car, he or she places his or her finger on the sensor to start the recognition process. In less than a second, the computer compares the detected fingerprint template with the stored template and establishes a relevant setting that matches the driver. The fingerprint template and matching software are stored in an embedded module in the car. When the fingerprint is matched successfully, the car controls the rearview mirror, car seat, wireless base station and air environment in the car according to the programmed internal parameters. In addition, the driving speed can be controlled, and if the driver is only a teenager, the speed is limited to 55 kilometers per hour. The implementation of these functions has many uses.

Making mobile Internet access more secure

With the development of semiconductor and software technology, mobile phones will gradually become a mobile terminal that can access personal and company data anytime and anywhere, so it is necessary to ensure the security of user access to prevent unauthorized access. The fingerprint recognition method used by law enforcement agencies only stores data on some specific points on the fingerprint rather than the entire image. Therefore, in comparison, the biometric fingerprint scanning system is more effective and reliable.

All processing steps of this type of detection are divided into the following steps: first, in the acquisition stage, the device collects biological samples from the finger; then, a pre-established mathematical formula or algorithm is used to extract unique data from the sample and convert it into a template; the registration and authentication program extracts at least seven feature matching points from the 30 to 40 feature points of the fingerprint for verification, including the bifurcation and termination points of the lines that constitute a fingerprint detail, and is defined as the distance between the feature points.

When registering, the information code is stored as a reference template for future user authentication. When the user enters the system, he/she swipes his/her finger across the sensor area, and the acquired live scan template is compared with the reference template. The whole process is completed within 1 or 2 seconds.

By comparison, the system will determine whether the live scan template contains enough biometric data to match the reference template and whether the two match. If not, the authentication fails and waits for the next recognition.

This fingerprint detection system has high performance. The probability of misjudging a valid fingerprint is less than 1%, and the possibility of misjudging an invalid fingerprint as a valid fingerprint is almost non-existent, with a probability of less than 0.01%.