WO2001094966A2

WO2001094966A2 - Velocity measurement for a moving finger, when detecting fingerprints. velocity is determined by analysing flank shifts.

Info

Publication number: WO2001094966A2
Application number: PCT/NO2001/000242
Authority: WO
Inventors: Stig Mathiassen; Svein Mathiassen
Original assignee: Idex As
Priority date: 2000-06-09
Filing date: 2001-06-08
Publication date: 2001-12-13
Also published as: WO2001094966A3; AU2001264433A1; NO20003002D0; NO20003002L

Abstract

This invention relates to a method for determining the velocity of a structured surface, especially a finger surface, moving relative to at least two in the direction of movement positioned sensors, having a predetermined distance between them, the sensors measuring the variation of a predetermined characteristic of said surface, comprising repeated measurements of the finger surface characteristic and detection of a the presence of a predetermined value along to the time axis for each sensor, and based on the time lapse between the occurance of the predetermined value at each sensor, as well as the known distance between the sensors, calculation of the surface velocity relative to the sensors.

Description

VELOCITY MEASUREMENT -FLANK SHIFT

This invention relates generally to a method for determining the velocity of a structured surface, especially a finger surface, moving relative to at least two in the direction of movement positioned sensors, having a predetermined distance between them, the sensors measuring the variation of a predetermined characteristic of said surface .

The market of biometrics is evolving rapidly, and the industry is becomming more mainstreamed. However, for biometrics to penetrate the consumer market, requirements are strict in respect to both price and performance (eg. power consumption) .

Currently finger scanners are typically pure capturing devices, hence the responsibility of any signal processing is left to the host. Thus, for implementation in eg. a mobile phone, requirements are also strict in respect to the use of processing power and memory of the signal processing. Nevertheless, as these processes mature and can be implemented in the finger scanner device, the requirements will still apply.

The size of a matrix finger scanner is typically restricted by the size of the finger images to be captured, due to the number of interconnects if the sensor where to be separated from the IC. Since the price is typically very much proportional to the size of the silicon die, the price of matrix finger scanner can not be expected to decrease significantly. For a stripe finger scanner, however, where the user wipes a finger over one or more arrays of sensor elements, the number of channels is reduced dramatically and separation of sensor and IC is possible, and the size of the IC is not restricted to the size of finger images. The reduction in channels of course also contributes to reduce the size of the die substantially. Since the price and the size of the IC are so related, stripe finger scanners have the potential of breaking the consumer market.

International patent application no. PCT/NO98/00182 describes a finger print sensor operating as a scanner requiring that the finger is moved over the scanner, which then samples information about the finger surface and generates a twodimensional representation of the fingerprint .

For stripe finger scanners, the user must be allowed to pull the finger over the scanner at various velocity. If the finger scanner samples at a fixed rate, this causes for a method that detects the speed of a finger and remaps the rows that are sampled, so that the axes of the remaped finger image are equal and linear.

It is therefor an object of this invention to provide a method for determining the velocity changes of the finger print, og similar surfaces, in order to enable corrections in the sampled data representing the suface struktures.

This object is obtained with a method as described above being characterized by repeatative measurements of the finger surface and detection of a predetermined value relative to the time axis for each sensor, and based on the time lapse between the occurance of the predetrmined value at each sensor, as well as the known distance between the sensors, calculation of the surface velocity relative to the sensors .

The invention will be described below with reference to the accompanying drawings, illustrating the invention by way of examples . Figure 1 illustrates schematically a linear fingerprint sensor. Figure 2 illustrates a first method for velocity measurements.

Figure 3 further illustrates the method in figure 2. Figure 4 illustrates an alternative method for measuring velocity. Figure 5 illustrates a semi linear fingerprint sensor adapted for use as a pointer tool.

The typical method for determining the speed of a finger pulled over a stripe finger scanner would be to use auto correlation. This would require one or more additional arrays of sensor elements parallel to the image sensor elements, if the image sensor elements consists of only one array.

Important factors for a speed correction process to run on eg. a mobile phone, are that it should be fast, require litle processing power and use litle memory. Preferrably it should run in real-time, due to the amount of memory that would be required to buffer a finger image captured of fingers wiped at various velocities, if the speed correction is post-processed. Some facts about eg. a finger pulled over a finger scanner, can be used to simplify and create a more efficient process for determening the velocity of the finger. Figure 1 shows a possible layout of sensor elements for such a solution corresponding to the solution described in the abovementioned international patent application

PCT/NO98/00182. The layout consists of one array 1 of image sensor elements, and one array 2 of additional speed correction sensor elements. Each additional speed correction sensor element form a pair with a corresponding sensor 3 element in the image sensor element array, as indicated.

Since the direction a finger is pulled is known, such a pair can be used to detect the speed of a finger. For enhanced tolerance on the finger pulling direction, adjacent sensor elements 4 on the image array 1 can be used. The sensors may of course be active continuously, thus measuring any changes close to the sensors at any time, but for use in mobile phones or PDA' s a preferable additional features is that at least one touch sensitive sensor is provided which may activate the others automatically. Thus saving the power resources of the equipment. This touch activation may be based on any type of sensor, but is preferably based on the use of at least one of the scanner sensors, e.g. triggering the measurements at a predetermined change in the capacitance of the sensor sorroundings . An example of a digital readout according to the invention of a speed correction channel pair is shown in Figure 2. Calibration of any significant variation in offset and gain between the channels should be performed prior to speed correction. The two time series represents the same section of a finger, since the sensor elements are in line with the pulling direction of the finger, where the phase angle between the two signals represents the time dt the finger uses from the front channel to the back. However, due to variations in velocity of the finger, the sections may be distortet in the time scale.

The range of velocity T_f a user may pull a finger should be set in accordance with what is natural for a human being. When a finger is pulled at a natural speed, the velocity will not vary significantly within one cycle of the signal, ie. one ridge of the finger surface. Because of this, it is possible to only make a few measurements during one cycle and let these values be valid for a fixed time-frame as indicated in the figure. To get the most exact phase angle, measurements should be done on the flanks of the signal, ie. where the signal is increasing or decreasing rapidly.

As shown in Figure 3, a sample in the front channel 12 can be selected based on the signal level variation of the two adjecent samples. The pitch of the two sensor elements in a speed correction pair should be less than the smallest pitch expected of the ridges in a finger surface. Thus, whenever the signal in the back channel crosses the level of a sample selected in the front channel, it represents the same point on the finger surface. The calculation of the measurement should be at subpixel level for increased accuracy. When the number of samples is known for one point at the surface of the finger to move from the front channel 12 to the back 13, the speed of the finger can then be expressed as :

v = D * F / n , where v is the speed

D is the pitch of the two sensor elements F is the sampling frequency of one channel n is the number of samples The distance the finger has moved during one sampling cycle of a channel, can be express as: d = v / F , where d is the distance the finger has moved during one sampling cycle of a channel From the two equations we get: d = D / n

Since the pitch of the sensor elements of a pair is fixed, the distance the finger has moved during one sampling cycle of a channel, is inversely proportional to the number of samples one point on the surface of the finger uses to move from the front channel to the back. By accumulating the delta distances the finger moves from sample to sample of the same channel, selections can be made of which rows to discard, and wich rows to map to the finger image maintain the correct scale, or new values can be calculated from interpolation.

As shown in Figure 1 a certain number of speed correction pairs should be spread out along the image array. This would be to minimise the possibility that the ridges are parallel to the direction the finger is pulled for all the pairs, wich would cause the process to fail to detect any speeds. For those pairs that detects speeds, the results for the different pairs should be averadged for enhanced accuracy. If for some period no speeds are valid, the sampled history should be used.

For the solution described so far, the requirement to the signal quality is relatively strict in respect to signal/noise ratio. For finger scanners that don't provide this signal quality, a similar approach can be used, to the cost of some increase in processing power and memory usage. Figure 4 shows a sketch of such an approach. Line T_x marks the peak point of a valley 14 of a finger section from a back channel 13. When a successive sample exceeds a certain threshold T₂ from the peak point, that valley will be defined. The area A below the previous ridge is now ready to be defined. This can be done by setting a level T_& eg. % from the peak point of the ridge to the closer of the peak points of the two alongside valleys. Since the last defined valley does not yet have two ridges alongside, it is not ready to be fully defined until the next ridge is defined. To complete the definition of the ridge, the area A can be used to compute the vertical axis of gravity G_x . Line G₂ represents the axis of gravity of the correlating ridge in the front speed channel 12, as the figure is showing the back channel 13. The number of samples between G and G₂ represents the same phase angle as in the approach described above, and hence can be replaced in that description.

Since a finger scanner will not be active for much time in eg. a mobile phone or Personal Digital Assistant (PDA) , the value would be increased substantially if such a unit would include some kind of pointer functionality. The combination of two such modes in one unit is favorable since the modes are exclusively active. Since the mobile phone technology is moving towards graphical displays (opposed to character based displays) , with displays growing bigger and number of buttons being reduced, the need for a pointer device is becoming more important . For PDAs this is already the case. Typical pointing devices for mobile phones and PDAs today are navigation keys or pen. The touchpads typically used in portable computers are too large to be convenient for such devices as mobile phones and PDAs . This invention describes a method for adding the functionality of a pointer into a finger scanner by utilising the structure of the finger. While a touchpad detects a finger being moved over its surface, this method detects the movement of the sensor area over the finger surface. For a stripe finger scanner this requires som additional sensor elements, while for matrix finger scanners a selection of the already existing sensor elements is sufficient. As this mode will be active for a substantial amount of time during the use of a mobile phone or PDA, it is of importance that the number of sensor elements being used is reduced to a minimum, to keep the consumption of power, processor and memory of the host low. The method for measuring velocity according to this invention may easily be used to detect the direction of the finger, e.g. by using two perpendicular fingerprint scanners and as suggested above comparing the signals from the speed correection sensor not only with the signals from the closest sensor in the array but also with the sensors being close to this, and thus providing a possibility for calculating the angle of the movement.

An example of a layout of the sensor elements for this method is shown in Figure 5. The principle of the pointer is function is to determine the direction of the finger being moved and then calculate the distance on the basis of the speed of the finger. More detailed descriptions of the finger print scanner as pointer tools are disclosed in the simultaneously filed Internastional patent applications No. [to be disclosed later - claiming priority from Norwegian patent applications no. 2000.3006 and 2000.3001] .

This invention is described as primarily based on a finger print scanner of the type described in the abovementioned International patent application no.

PCT/NO98/00182 thus primarily on capacitance measurements on a finger surface. It is, however, clear that the invention may used different sensor types, such as thermal sensors, within the scope of this invention.

Claims

C l a i m s

1. Method for determining the velocity of a structured surface, especially a finger surface, moving relative to at least two in the direction of movement positioned sensors, having a predetermined distance between them, the sensors measuring the variation of a predetermined characteristic of said surface, c h a r a c t e r i z e d in repeated measurements of the finger surface characteristic and detection of a the presence of a predetermined value along to the time axis for each sensor, and based on the time lapse between the occurance of the predetermined value at each sensor, as well as the known distance between the sensors, calculation of the surface velocity relative to the sensors.

2. Method according to claim 1, wherein a time windows is defined, requiring the occurance of the predetermined values from each sensor to be within said time window.

3. Method according to claim 1, wherein an interpolation is performed between the measured values, and the time lapse is calculated in the basis of the values of the interpolated curve .

4. Method according to claim 1, wherein the sign of derivative of the varying values is determined, so as to differ between front or back edges of the passing surface features.