JP2006146343A - Iris information sensor and iris identifying device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To extract a feature of the iris of an eye with a simple structure to improve iris identification accuracy in an iris information sensor in which pixels are arranged in a polar coordinate state. <P>SOLUTION: Around an imaging sensor part having a photoelectric conversion pixel group arranged in a polar coordinate state, the iris information sensor comprises a read voltage holding part for temporarily storing image signals by one cycle of pixels sequentially at every radius position of the pixel group; a weighting moving average circuit sequentially weighting-averaging to output signals corresponding to image signals from a prescribed number of pixels consecutively in a circumferential direction among image signals stored in the holding part; and a comparator comparing an output from the average circuit with a signal corresponding to an image signal from a prescribed single pixel among the prescribed number of pixels consecutive in the circumferential direction among the image signals stored in the holding part. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an iris information sensor and an iris discriminating apparatus, and in particular, in an area sensor in which pixels are arranged in polar coordinates, that is, a polar coordinate sensor, a weighted moving average circuit having a novel configuration is used for feature extraction of eye iris information. The present invention relates to a technique for improving iris identification accuracy with a simple configuration.

  The applicant of the present application previously proposed to use a polar coordinate sensor in which photoelectric conversion pixels are arranged in polar coordinates for use in an iris information acquisition device and an iris identification device for performing personal identification using the iris pattern of the eye. is doing.

  The polar coordinate sensor includes an imaging sensor including photoelectric conversion pixel groups arranged in a polar coordinate pattern, and an error direction between a center of an iris image formed on the imaging sensor and a polar coordinate pole of the imaging sensor. The center position error detecting means for obtaining the size is integrated and formed on one substrate.

  In such an iris information sensor, the iris pattern characteristic extraction specific to the iris of each individual's eye is performed to obtain an iris code. This iris code can be personally authenticated by comparing with an iris code registered in advance.

  In order to perform feature extraction based on the iris pattern of the eye, for example, the imaging signal of the polar coordinate sensor, that is, the pixel output signal is input to the bandpass filter, and the change point of the pixel output signal by the bandpass filter, that is, θ of the iris pattern Extract changes in direction. Then, the extracted signal is compared with a predetermined threshold value to generate an iris code. Alternatively, the feature extraction of the iris pattern may include, for example, a configuration including a band pass filter and a comparator by obtaining an average value of outputs of consecutive N pixels and comparing the average value with the output of the target pixel. Equivalent functions can be achieved and this method can be referred to as a moving average comparison method.

  When performing such feature extraction of iris information by the moving average comparison method, an FIR filter is required. FIG. 5 shows a configuration example of a circuit that performs feature extraction using a conventional FIR filter.

5 includes an A / D converter 42 that converts an output of the image sensor, that is, a pixel output (Sout) into a digital signal, and delay elements (Z ⁻¹ ) 41a, 41b, 41c,. .., 41h, and digitized pixel outputs (Sout) and amplifiers (or attenuators) 43a, 43b, 43c, which perform predetermined weighting on the outputs of the delay elements 41a, 41b, 41c,. .., 43i and a linear phase FIR filter provided with an adder 49 for adding the outputs of the amplifiers 43a, 43b, 43c,. 5 also includes, for example, an amplifier 45 that performs gain adjustment in response to an output of any one of the delay elements 41a, 41b, 41c,..., 41h, that is, a target pixel output (SDout), A comparator 47 is provided which receives the output of the adder 49 at the other input and the output applied to one input.

In the configuration of FIG. 5, the pixel output (Sout) from the polar sensor is digitized by the A / D converter 42 and then sequentially delayed by the delay elements 41a, 41b, 41c,. Sout) and the outputs of these delay elements are weighted and then added in an adder 49. As a result, a weighted moving average output of the pixel output (Sout) is obtained from the adder 49. Then, the comparator 47 compares the output of the adder 49 and the signal weighted by the amplifier 45 to the target pixel output (SDout) to obtain a feature code (Cout). In this case, it is necessary to appropriately set the gains of the amplifiers 43a, 43b, 43c,..., 43i, 45 according to the number of taps of the FIR filter, and to adjust the filter characteristics and the level at the time of comparison.
JP 2000-36036 A

  In such a feature extraction structure using the conventional FIR filter, since the pixel output is digitally processed, it is necessary to use a complicated delay element, etc., which complicates the circuit configuration and increases the occupied area on the integrated circuit chip. In addition, power consumption increases.

  Further, as will be described later, in the present invention, when a configuration that uses a plurality of weighted moving average circuits and simultaneously encodes and outputs features in a plurality of frequency bands to improve the identification accuracy, the circuit The configuration becomes more complicated and the amount of hardware increases. Therefore, the area occupied by the circuit on the integrated circuit increases and the power consumption also increases.

  Accordingly, an object of the present invention is to provide an iris information sensor including photoelectric conversion pixel groups arranged in polar coordinates in view of such problems in the prior art, and an eye based on a weighted moving average comparison method with a very simple circuit configuration. It is intended to enable feature extraction of iris information, and to achieve simplification of the device configuration, reduction of the occupied area on the integrated circuit, reduction of power consumption, and the like.

  Another object of the present invention is to perform feature extraction of eye iris information using a plurality of frequency bands and at the same time, with a simple circuit configuration, in an iris information sensor having photoelectric conversion pixel groups arranged in polar coordinates. By enabling encoding, the accuracy of personal identification using iris information is greatly improved.

  According to one aspect of the present invention, a photoelectric conversion pixel group arranged in a polar coordinate form is provided, and image light from an iris region of an eye is converted to a center of an iris image formed on the pixel group and a polar coordinate pole. And an imaging sensor unit that forms an image on the photoelectric conversion pixel group so as to match and reads an iris image signal, and is arranged around the imaging sensor unit, and sequentially for each radial position of the photoelectric conversion pixel group Corresponds to image signals from a predetermined number of pixels continuous in the circumferential direction out of the image signals stored in the read potential holding unit and the read potential holding unit for temporarily storing image signals of pixels for one round A weighted moving average circuit that sequentially weights and outputs signals; an output of the weighted moving average circuit; and a predetermined number of pixels that are continuous in the circumferential direction among the image signals stored in the read potential holding unit From a given pixel Iris information sensor, characterized by comprising a comparator for comparing the signal corresponding to the image signal.

  In this case, the weighted moving average circuit outputs one of the signals corresponding to the image signals from a predetermined number of pixels each having one end continuous in the circumferential direction among the image signals stored in the read potential holding unit. A plurality of resistors may be provided that constitute an output end that is connected together and has the other end connected together to provide a weighted average output.

  Further, a signal corresponding to an image signal from a predetermined number of pixels that are arranged around the imaging sensor unit and is continuous in the circumferential direction among the image signals of one round of pixels stored in the readout potential holding unit. Read lines, a switch circuit connected between an output of an image signal corresponding to each pixel of the read potential holding unit and the read line, and operating the switch circuit to read the read potential Of the image signals for one round of pixels stored in the holding unit, a predetermined number of pixels consecutive in the circumferential direction are sequentially selected, and signals corresponding to the image signals from the selected predetermined number of pixels are respectively corresponded. A circumferential shifter for supplying to the readout line, and a signal corresponding to an image signal from a predetermined number of pixels continuous in the circumferential direction from the readout line, respectively. It is convenient to supply to the plurality of input.

  The predetermined one pixel may be a pixel at a central position among a predetermined number of pixels continuous in the circumferential direction.

  In addition, the read potential holding unit may include at least a number of capacitors that accumulate signal charges of the number of pixels corresponding to one round.

  More specifically, the read potential holding unit captures an iris image after resetting the first capacitor that accumulates signal charge immediately after resetting the pixel corresponding to each of at least one round of pixels and the pixel. It is advantageous to provide a second capacitor for storing the signal charge obtained in this way.

  According to another aspect of the present invention, each of the plurality of weighted moving average circuits and the plurality of comparators is provided, and the plurality of weighted moving average circuits have a different number of inputs from each other, and are continuously predetermined in the circumferential direction. An iris information sensor characterized by receiving and encoding a different number of signals corresponding to image signals from a number of pixels, and simultaneously encoding and outputting features in a plurality of frequency bands. Provided.

  According to still another aspect of the present invention, the iris information sensor configured as described above is used to simultaneously encode and output features in a plurality of frequency bands, and a plurality of pre-registered feature codes in the plurality of frequency bands that have been output. An iris identification device is provided that performs iris identification by comparing with the feature code.

  According to the present invention, the image signal of the pixels for the entire circumference is obtained for each radius of the photoelectric conversion pixel group by the readout potential holding unit provided around the polar coordinate sensor including the photoelectric conversion pixel group arranged in the polar coordinate form. Temporary storage is sequentially performed, and a weighted moving average can be obtained with a simple circuit configuration using the image signal. Therefore, the same processing as that of the FIR filter can be performed without using a complicated delay element as in the prior art, and the configuration of the feature extraction portion of the iris information sensor can be greatly simplified. This simplifies the device configuration of the iris information sensor, reduces the occupied area on the integrated circuit chip, and reduces power consumption.

  Furthermore, since the weighted moving average can be performed with a simple circuit configuration as described above, a plurality of weighted moving average circuits are prepared, and a configuration for simultaneously encoding and outputting features in a plurality of frequency bands can be easily realized. This can greatly improve the accuracy of personal identification using iris information.

Hereinafter, preferred embodiments of the present invention will be described with reference to the drawings.
FIG. 1 shows an overall configuration of an iris recognition system using an iris information sensor 10 according to an embodiment of the present invention, that is, an iris information acquisition apparatus. In the figure, this iris recognition system first includes two wide-angle cameras 1a and 1b and a zoom camera 7. The wide-angle cameras 1a and 1b pick up images of the eyes of the person to be recognized by the CCDs 2a and 2b, and send image signals obtained by reducing noise of the picked-up signals to the iris acquisition controller 6 by the CDS circuits 3a and 3b. To do. The wide-angle cameras 1a and 1b control the CCD drivers 5a and 5b that drive and control the CCDs 2a and 2b based on the synchronization control signals generated by the control units 4a and 4b, and the processing operations of the CDS circuits 3a and 3b. Is controlling.

  The system of FIG. 1 also includes an iris acquisition controller 6. The iris acquisition controller 6 specifies the three-dimensional position of the eye of the person to be recognized based on the image signals input from the wide-angle cameras 1a and 1b, and sends a signal or data indicating the specified result to the zoom camera 7. Output.

  Based on this three-dimensional position, the zoom camera 7 drives and controls the steering motor 7b to move the mirror 8c so that the imaging direction of the zoom camera 7 is directed toward the human eye. The zoom camera 7 controls the zoom / focus motor 7a that drives the zoom lens optical system 8 to focus on the three-dimensional position of the human eye, and the iris image to be imaged is captured by the iris information sensor 10. Control is performed to adjust the zoom amount so as to match the region. The iris acquisition unit 9 in the zoom camera 7 sends the position error and focus error information between the iris image captured by the sensor 10 and the sensor to the iris acquisition controller 6.

  The iris acquisition controller 6 servo-controls each part of the camera system described above until information on the iris part is obtained. When the information of the iris part is obtained, this information, that is, the iris code, is collated with a built-in or pre-registered iris code held by a host computer (not shown) or is collated with the host computer. The host computer may control each part of the camera system, or the sensor part itself may control by each error signal of the iris acquisition part 9.

  Note that the configuration of FIG. 1 includes a zoom lens optical system 8 that adjusts the size of the captured image so that the iris image to be captured matches the imaging region of the sensor 10. However, it is obvious that the present invention does not use such a zoom lens optical system 8 and may use, for example, a single focus lens. For example, in a configuration in which a peephole is provided at a predetermined position on the front surface of the camera and the identification target person is instructed to view the lens portion of the camera from this peephole, a so-called peeping method, the distance between the camera and the eyes Can be set to a predetermined value. Therefore, in such a case, the iris image to be imaged can be made substantially coincident with the imaging area of the sensor 10 without using a zoom lens. In this case, the position of the eyes with respect to the camera is also regulated to a predetermined positional relationship, so that a configuration in which the wide-angle cameras 1a and 1b in FIG. 1 are not used is possible.

  In the configuration of FIG. 1, the mirror 8c is moved so that the imaging direction of the zoom camera 7 is directed toward the eyes. However, in the present invention, instead of the mirror, the eye can be aligned by shifting a VAP (variable vertical prism), an imaging lens or a part thereof in a direction perpendicular to the optical axis.

  Here, the iris information received by the iris acquisition controller 6 from the iris acquisition unit is an iris code binarized by the iris acquisition unit 9. The detailed configuration of the iris acquisition unit 9 will be described with reference to FIG.

  FIG. 2 shows a schematic configuration of the iris acquisition unit 9 in the iris information acquisition apparatus shown in FIG. The iris acquisition unit 9 shown in the figure includes an iris information sensor 10 including a polar coordinate sensor unit 10a including a plurality of photoelectric conversion pixels arranged in a polar coordinate shape. Each photoelectric conversion pixel of the polar coordinate sensor unit 10a may be constituted by a photodiode, a CCD image pickup element, or a MOS type image pickup element or other photoelectric conversion elements. Further, a position detection unit 10c for detecting a position error between the center of the iris image and the polar coordinate pole of the polar coordinate sensor unit is disposed in the central area 10b of the polar coordinate sensor unit 10a. The position error detection unit 10c can be constituted by, for example, a photodiode divided into four with the polar coordinate pole as the center.

  The apparatus of FIG. 2 also has an LED (light emitting diode) 13 for illumination that is provided on the optical axis of the sensor 10 and illuminates the human eye with visible light. It is advantageous that the light emitting surface of the LED 13 is optically shifted from the light receiving surface of the sensor 10 so that the LED 13 can accurately illuminate human eyes. As the illumination means, in addition to the LED 13, an EL (electroluminescence) element and other arbitrary ones can be used.

  The apparatus of FIG. 2 further includes an iris inner diameter detection circuit 11. The iris inner diameter detection circuit 11 detects the boundary between the pupil and the iris in the radial direction in the captured iris image by level change, that is, performs edge detection, and obtains the inner diameter of the iris image from the result. In this case, although not shown in FIG. 2, for the iris image capturing of the polar sensor unit 10 a, for example, eight ring bands are formed. The edge detection pixels are arranged in a ring shape. The iris inner diameter detection circuit 11 detects the boundary between the pupil and the iris by changing the level of the detection signal between the edge detection pixel at the inner periphery and the pixel arranged outside the pixel. Specifically, for example, the level change of the detection signal can be obtained by a difference in values obtained by averaging each pixel output for one round.

  The apparatus of FIG. 2 further includes an iris outer diameter detection circuit 12, and the iris outer diameter detection circuit 12 similarly detects the outer periphery of the iris in the picked-up iris image by level change, that is, performs edge detection. Find the outer diameter. That is, similarly to the iris inner diameter detection circuit, the iris outer diameter detection circuit 12 outputs a detection signal between the pixel for edge detection on the outer peripheral portion of the polar coordinate sensor unit 10a and a pixel arranged inside the pixel. Edge detection can be performed by changing the level.

  The iris outer diameter detection circuit 12 outputs a zoom error signal to the iris acquisition controller 6. The iris inner diameter detection circuit 11 detects the inner diameter of the iris as described above, but generates a deviation between the detected inner diameter and the inner peripheral edge of the sensor 10, that is, a pupil diameter error signal, and sends it to the iris acquisition controller 6. To do.

  The iris acquisition controller 6 controls the brightness of the LED 13 so that this pupil diameter error signal disappears. That is, when the pupil diameter is larger than the inner peripheral edge of the polar coordinate sensor unit 10a, the brightness of the LED 13 is increased, and when the pupil diameter is smaller than the inner peripheral edge of the polar coordinate sensor unit 10a, the brightness of the LED 13 is decreased. Do. Specifically, for example, a servo operation is performed to control the brightness of the LED 13 so that the difference between the output of the edge detection pixel on the inner peripheral side and the output of the effective pixel outside the pixel is maximized.

  As a result, the pupil diameter always matches the inner periphery of the polar sensor unit 10a, and the outer periphery of the polar sensor unit 10a matches the outer diameter of the iris by a steering error signal, a zoom error signal, etc. The iris part is always imaged in accordance with the pixel arrangement pattern on the polar coordinate sensor part 10a.

  Therefore, the radial width of each pixel in the polar coordinate sensor unit 10a may be divided into a necessary resolution, for example, eight equal parts, and the iris code is directly used by using the imaging signal from the sensor 10 having such a polar coordinate sensor unit 10a. Can be generated.

  The imaging signal from the sensor 10 is sent to the high pass filter 14. The high-pass filter 14 extracts the high-frequency component level of the imaging signal and outputs it as a focus control signal. The steering error signal, zoom error signal, and focus control signal as described above are sent to the iris acquisition controller 6 described above. As described above, the iris acquisition controller 6 controls the steering motor 7b and the zoom / focus motor 7a in the zoom camera so that the iris to be imaged coincides with the center of the polar coordinate sensor unit 10a, and the outer diameter of the iris is set to polar coordinates. It is made to correspond to the outer periphery of the sensor part 10a, and a focusing process is performed.

  The timing generator 18 drives the driver 19 of the sensor 10 to perform scanning for outputting a pixel signal in the θ direction (rotation direction) for each ring. When the fine adjustment by the feedback process is performed and proper imaging is possible, the read output signal from each pixel is output to the weighted moving average circuit 15 via the read potential holding unit as will be described in detail later. Sent to. The weighted moving average circuit 15 extracts a change point of the transmitted pixel output signal, that is, a change in the θ direction of the iris pattern. Further, the binarization circuit 16 generates an iris code obtained by performing binarization processing on the extracted signal by an inherent comparator. The iris code is sent to the iris acquisition controller 6 via the latch circuit 17. As described above, the iris acquisition controller 6 compares the iris code of each ring band with the iris code registered in advance in the host computer to perform personal authentication. Note that the iris code is preferably encrypted or encoded for security and transmitted between the circuit devices.

  According to the above-described method, the sensor 10 having a large number of pixels in the radial direction does not need to be divided into each ring band, and the iris center is determined, so that an iris image is registered in advance. By doing so, the authentication process can be performed by very simple pattern matching. That is, since the inner diameter and outer diameter of the iris are known, pattern matching can be performed only by rotating at the center of the iris, and the comparison speed can be significantly improved.

  FIG. 3 shows a detailed configuration example of a part from the output of the iris information sensor 10 of FIG. 2 to the weighted moving average circuit 15 and the binarization circuit 16. 3 includes a read potential holding unit 31 that receives read signals Soutn + 1, Soutn, Soutn-1,... From each pixel of the polar sensor unit 10a (not shown), a circumferential shifter 33, and a switch circuit. 35, a weighted average circuit 37, a comparator 39, and the like.

  The polar coordinate sensor unit 10a is configured by arranging photoelectric conversion pixels such as photodiodes (PD) in a polar coordinate form. As the photoelectric conversion pixels, for example, pixels that are arranged in the circumferential direction along eight radial positions (ring bands) around the central area excluding the central area having a predetermined radius from the polar coordinate pole are used. The number of pixels in the circumferential direction in each ring band is equal, for example, 256.

  The readout signal Soutn from each pixel input to the readout potential holding unit 31 in FIG. 3 corresponds to the readout signal from one pixel in the circumferential direction of the polar coordinate sensor unit 10a.

  The read potential holding unit 31 is disposed, for example, on the outer periphery of the polar coordinate sensor unit 10a described above. In addition, the circumferential shifter 33 is disposed, for example, on the outer periphery of the read potential holding unit 31.

  Further, a predetermined number of read lines SiGout-4,..., SiGout,..., SiGout + 4 are arranged on the outer circumference of the circumferential shifter 33 on concentric circles. The number of read lines corresponds to the number of inputs of the weighted average circuit 37 and is nine in FIG. Therefore, an example in which the number of inputs of the weighted average circuit 37 is nine, that is, nine taps is shown. Note that the number of read lines and the number of inputs to the weighted average circuit 37 are not limited to nine, and may be larger or smaller. Further, these numbers may be odd numbers or even numbers as in the present embodiment.

  In response to the signal from each one pixel in the circumferential direction, the read potential holding unit 31 includes a load transistor, a slice transistor (STr), a hold transistor (HTr), and two capacitors Cl, These are connected as shown in the figure.

  A readout signal (..., Soutn-1, Soutn, Soutn + 1,...) From the pixels for one round is sent to the readout potential holding unit 31 from each polar position, that is, the ring band, from the polar coordinate sensor unit 10a. And memorized at the same time. The load transistor constitutes a load circuit for each pixel.

  Next, the operation of the read potential holding unit will be schematically described. In order to simplify the description, it is assumed that the power supply voltage of the slice transistor STr is 2 volts, the power supply voltage of the hold transistor HTr is 2.5 volts, and the capacitances of the capacitors Cl and Ch have the same value.

  (1) First, with a potential of 2.5 volts applied to the line Xfer, a positive voltage, for example, 3.3 volts, is applied to the slice line (Slice) and hold line (Hold). Turn on the transistor. This charges the capacitors Cl and Ch (-0.5 volts and +2.5 volts, respectively).

  (2) Next, a pulse in the negative voltage direction, for example, a pulse from 2.5 volts to 0 volts, is applied to the Xfer line, and the photopotential (Vs) accumulated in the pixel is sampled. As a result, the charge of the capacitor Ch moves to the capacitor Cl by an amount corresponding to the potential Vs.

  (3) The hold transistor HTr is turned on and only the capacitor Ch is charged again. At the same time, the pixel is reset (RS) to erase the photocharge.

  (4) Apply a negative pulse to the Xfer line again to sample the initial potential (Vn) of the pixel. As a result, as described in the above (2), the charge of the capacitor Ch moves to the capacitor Cl. However, since the charge of Vs has already accumulated in the capacitor Cl, the charge proportional to Vn−Vs is increased. It moves from the capacitor Ch to the capacitor Cl.

(5) From the above operation, the noise-cancelled potential given by the following equation is output to the output (Voutn) of the read potential holding unit 31 with the power supply voltage of the hold transistor (HTr) being 2.5 volts as an offset. Is done.

[Equation 1]
Vout = (offset potential) − (Vn−Vs) · Cl / Ch

That is, the read potential holding unit 31 provides a pixel output in which noise of each pixel, that is, the initial potential, is canceled by using two capacitors Cl and Ch.

  In this way, the read potential holding unit 31 holds the read potential of the pixels for one round from one ring band of the polar coordinate sensor unit 10a and provides it to the switch circuit 35 described below.

  The switch circuit 35 is selectively turned on by a selection signal..., Seln + 6,..., Seln,. The output signals Voutn and the like of a plurality of pixels are sequentially input in parallel to the readout line SiGout and the like.

  The circumferential shifter 33 is formed by, for example, sequentially connecting a plurality of sets of inverters and transmission gate circuits in series in the circumferential direction. For example, “1” is generated by the slave clock signal (SCL) and the master clock signal (MCL). The signals corresponding to are sequentially shifted along the circumferential direction. Accordingly, the selection signals..., Seln + 6,..., Seln,. That is, only one of these selection signals is sequentially activated.

  For example, when the selection signal Seln becomes active (that is, “1”), for example, the output Voutn of the read potential holding unit 31 is transferred to the output b0 of the switch circuit 35 and the output Voutn−1 of the read potential holding unit 31 is switched to the switch circuit. 31, and the output Voutn + 1 of the read potential holding unit 31 is provided to the output b + 1 of the switch circuit 35. In this way, the output of nine pixels from the read potential holding unit 31 is transferred to nine read lines at a time, and nine outputs b + 4, b + 3,..., B0 from each read line. ,..., B-4, and transferred to the weighted average circuit 37.

  When the data “1” is shifted by 1 bit by the circumferential shifter 33, read signals from 9 pixels shifted by 1 bit from the previous 9 pixels are transferred to the 9 outputs of the switch circuit 35. Is done. In this way, readout signals from nine pixels are sequentially transferred per one round of pixels. When the reading of pixels for one round is completed, the next ring band is read.

  The weighted average circuit 37 receives the output signals from, for example, the nine readout lines described above, and weights them with the weighting resistors R4, R3,..., R0,. Post-add to produce a weighted average output. This weighted average output is provided to one input of the comparator 39. The other input of the comparator 39 receives a signal from the pixel of interest, for example, the output b0 of the switch circuit 35, through a buffer amplifier and a weighting resistor R. Note that the pixel of interest normally uses the signal of the central pixel in the nine readout lines, but other ones of these nine signals may be used. Further, signals other than these nine signals may be used.

  That is, in the configuration of FIG. 3, since the read potential for the entire circumference is stored in the read potential holding unit 31, the front and rear data already exist at both ends of each read line, for example, SiGout. Therefore, by simply adding these data through the weighted resistors R4, R3,..., R0,. It is done. Therefore, the feature extraction and the binarization of the iris image are performed by supplying this weighted average value to one input of the comparator 39 and inputting the center data to the other input. As a result, a feature code, that is, an iris code can be output from the comparator 39. As a result, the same processing as the FIR filter can be performed without using a complicated delay element.

In such a configuration, the relationship between the number of taps (k) and the extraction frequency (ft: spatial angular frequency) is, for example, a simple average filter when all weights are equal, and is expressed by the following equation: .

[Equation 2]
ft = m · fs / k

However, fs is a sampling frequency, m is an integer, and ft <fs / 2.

The sampling frequency fs is as follows if the number of pixels in the circumferential direction is n.

[Equation 3]
fs = 2πn [rad / sec]

Therefore, when the value of fs is substituted into Equation 2, the extraction frequency ft is expressed by the following equation.

[Equation 4]
ft = m · 2πn / k

That is, the extraction frequency ft is a harmonic component that is an integral multiple of 1 / t the number of taps (k) of the number of pixels (n) in the circumferential direction.

  FIG. 4 shows an example of characteristics of an equal weighted average filter, that is, a simple average filter. In the figure, the horizontal axis represents frequency (f) and the vertical axis represents gain (Gain). Therefore, when the output of such an equal weighted average filter and the center level are compared, that is, the difference between the two is obtained, a high-pass filter that particularly extracts the frequency component as shown in the feature extraction filter characteristic example of FIG. Characteristics are obtained.

  When the weighted moving average comparison method is used for the feature extraction of the iris information as described above, the same processing as that of the FIR filter is performed, and the iris code can be obtained by extracting and binarizing the feature of the iris image.

  In this case, the frequency band to be extracted can be changed to various frequency characteristics by changing the number of taps and adjusting the weight ratio. Therefore, as another embodiment of the present invention, a configuration in which a plurality of combinations of the weighted average circuit 37 and the comparator 39 as shown in FIG. 3 is provided is possible.

  For example, in the configuration shown in FIG. 3, nine readout lines SiGout-4,..., SiGout,..., SiGout + 4 are provided, and the weighted average circuit 37 has nine inputs for these readout lines. That is, it has 9 taps.

  Therefore, as an example, the number of the readout lines is set to 9, and in addition to the 9-tap weighted averaging circuit 37, 7-tap signals for inputting signals from, for example, 7 of the 9 readout lines are input. A weighted average circuit and, for example, a 3-tap weighted average circuit for inputting signals from three of the nine readout lines are provided. As is apparent from FIG. 3, the weighted moving average circuit has a simple configuration in which resistors are connected and put into a comparator. Therefore, even if a plurality of weighted average circuits are provided in this way, the circuit configuration is not much. Not complicated.

  Thus, feature information in a plurality of frequency bands can be coded and output simultaneously. If the same number of iris codes to be compared are stored in advance and compared, the feature code amount increases, but the iris identification accuracy can be greatly improved. For example, a weighted moving average circuit with a large number of taps does not have a sudden change in feature extraction data even if the iris image is slightly out of focus, whereas a weighted moving average circuit with a small number of taps is sensitive to the focus of the iris image. Change. Therefore, iris identification can be performed with high accuracy under various conditions by combining the weighted moving average circuits having various tap numbers.

  The present invention can be applied to an apparatus such as an ATM that has high reliability and high convenience and needs to perform personal identification with high accuracy, such as an iris information acquisition apparatus and an iris identification apparatus in such an apparatus. Identification accuracy can be improved without complicating the apparatus configuration.

It is a block diagram which shows the structure of the outline of the iris information acquisition apparatus using the iris information sensor concerning one Embodiment of this invention. It is a block diagram which shows the structure of the outline of the iris acquisition part in the iris information acquisition apparatus of FIG. It is a partial circuit diagram which shows the structure of the outline of the data reading part of the iris information sensor concerning one Embodiment of this invention. It is a graph for demonstrating the characteristic of the movement weighted average circuit of the iris information sensor of FIG. It is a block diagram which shows the structure of the feature extraction circuit using the conventional FIR filter.

Explanation of symbols

1a, 1b Wide-angle camera 6 Iris acquisition controller 7 Zoom camera 7a Zoom / focus motor 7b Steering motor 8 Zoom lens optical system 10 Iris information sensor 10a Polar coordinate sensor unit 10b Central area 10c Position error detection PD sensor 11 Iris inner diameter detection circuit 12 Iris Outer diameter detection circuit 13 Light emitting diode 14 High-pass filter 15 Weighted moving average circuit 16 Binarization circuit 17 Latch circuit 18 Timing generator 19 Driver 31 Read potential holding unit 33 Circumferential shifter 35 Switch circuit 37 Weighted average circuit 39 Comparator

Claims (8)

A photoelectric conversion pixel group arranged in a polar coordinate, and the photoelectric conversion of image light from an iris region of an eye so that a center of an iris image formed on the pixel group and a pole of the polar coordinate coincide with each other An imaging sensor unit that forms an image on a pixel group and reads an iris image signal; and
A readout potential holding unit that is arranged around the imaging sensor unit and temporarily stores image signals of pixels for one round sequentially for each radial position of the photoelectric conversion pixel group;
A weighted moving average circuit that sequentially weights and outputs signals corresponding to image signals from a predetermined number of pixels continuous in the circumferential direction among the image signals stored in the read potential holding unit;
An output of the weighted moving average circuit, and a signal corresponding to an image signal from a predetermined one of a predetermined number of pixels continuous in the circumferential direction of the image signal stored in the read potential holding unit; A comparator for comparing
An iris information sensor comprising:   The weighted moving average circuit receives one of signals corresponding to image signals from a predetermined number of pixels each having one end continuous in the circumferential direction among the image signals stored in the read potential holding unit, The iris information sensor according to claim 1, further comprising a plurality of resistors that are connected together to form an output terminal that supplies a weighted average output. further,
A signal corresponding to an image signal from a predetermined number of pixels continuous in the circumferential direction is supplied from among the image signals for one round of pixels arranged around the imaging sensor unit and stored in the readout potential holding unit. A plurality of read lines, and
A switch circuit connected between an output of an image signal corresponding to each pixel of the readout potential holding unit and the readout line;
The switch circuit is operated to sequentially select a predetermined number of pixels continuously in the circumferential direction from the image signals of the pixels for one rotation stored in the read potential holding unit, and from the selected predetermined number of pixels. A circumferential shifter for supplying a signal corresponding to an image signal to a corresponding readout line;
3. A signal corresponding to an image signal from a predetermined number of pixels continuous in the circumferential direction is supplied from the readout line to a plurality of inputs of the weighted moving average circuit, respectively. The described iris information sensor.   2. The iris information sensor according to claim 1, wherein the predetermined one pixel is a pixel at a center position among a predetermined number of pixels continuous in the circumferential direction.   2. The iris information sensor according to claim 1, wherein the read potential holding unit includes a capacitor of a number that accumulates signal charges of pixels of at least one round.   The read potential holding unit corresponds to each of at least one round of pixels, a first capacitor that accumulates signal charges immediately after pixel reset, and signal charges obtained by imaging an iris image after pixel reset The iris information sensor according to claim 5, further comprising a second capacitor for storage.   A plurality of weighted moving average circuits and a plurality of comparators are provided, the plurality of weighted moving average circuits having different numbers of inputs and corresponding to image signals from a predetermined number of pixels continuous in the circumferential direction. The iris information sensor according to claim 1, wherein features in a plurality of frequency bands are simultaneously encoded and output by performing weighted averaging by receiving a different number of signals among the signals. Using the iris information sensor according to claim 7 to simultaneously encode and output features in a plurality of frequency bands, and compare the output feature codes in the plurality of frequency bands with a plurality of pre-registered feature codes. An iris identification device characterized by performing iris identification by

Images