JP4803676B2 - Image input device and personal authentication device - Google Patents

Description

  The present invention relates to a personal authentication device that performs personal authentication using an image obtained by imaging a subject inside a living body, and an image input device suitable as a subject image input unit of such a personal authentication device.

  Patent Documents 1, 2, and 3 disclose personal authentication devices that image a vein pattern inside a finger through a monocular optical system and perform personal authentication from the vein pattern in a state where a human finger is illuminated with infrared light or near infrared light. It is described in.

  Patent Documents 4 and 5 and Non-Patent Document 1 describe thin image input devices using a compound eye optical system. Non-Patent Document 1 also describes an application example to fingerprint input for the purpose of application to a fingerprint authentication system.

JP 2004-27281 A JP 2005-92375 A JP 7-21373 A Japanese Patent No. 3705766 JP 2001-61109 A Igenji et al., "Development of thin image input device using compound eye optical system", Journal of the Institute of Image Information and Television Engineers Vol.57, No.9, pp.1135-1141 (2003)

  When a personal authentication device that performs personal authentication using a finger vein pattern or the like is mounted on a small electronic device such as a mobile phone or a notebook computer, it is strongly required to reduce the size of the personal authentication device, in particular, to reduce its thickness.

  However, since the personal authentication device as described in Patent Documents 1, 2, and 3 is configured to image the vein pattern through the monocular optical system, the subject distance and the imaging distance are limited due to the optical imaging relationship, There is a limit to making the device thinner.

  In order to reduce the thickness of a personal authentication device, it is most important to reduce the thickness of an image input device for inputting an image of a subject inside a living body such as a finger vein. In order to reduce the thickness of the image input device, it is generally advantageous to apply a compound eye optical system as described in Patent Documents 4 and 5 and Non-Patent Document 1. However, the image input device as the subject image input means of the personal authentication device is not simply required to be thin, but must be capable of inputting the subject image as a high-quality image.

  In addition, an image input device for a personal authentication device must be able to cope with variations in subject distance. For example, when a vein inside a finger is imaged as a subject, the subject distance fluctuates with a considerable width due to individual differences such as the skin thickness of the finger. In the case of an image input device using a compound eye optical system, a subject image is captured as a compound eye image. Therefore, it is necessary to reconstruct and input a single image from the compound eye image, but the subject distance varies. Even a good quality image must be reconstructed.

  In view of the above points, the present invention provides a subject image input unit of a personal authentication device that has a configuration that can be easily reduced in thickness and that can input a subject image of good quality even when the subject distance varies. An object is to provide a suitable image input device. Another object of the present invention is to provide a personal authentication device that can be easily thinned and that is resistant to variations in subject distance.

An image input apparatus according to the invention of claim 1 is provided.
An imaging optical system for imaging a subject inside a living body as a compound eye image through a lens array in which a plurality of single lenses are arrayed; and
Parallax detection means for detecting parallax between single-eye images in a compound eye image captured by the imaging optical system;
Using the parallax detected by the parallax detection means, a single image is obtained from the plurality of single-eye images in the compound-eye image by the first reconstruction method for joining the plurality of single-eye images in the compound-eye image. The compound eye is reconstructed by using a second reconstruction method for rearranging pixel luminances of a plurality of single-eye images in the compound eye image using the disparity stitching processing means and the disparity detected by the disparity detecting means. Reconstruction processing means including pixel rearrangement processing means for reconstructing a single image from a plurality of single-eye images in the image;
As a reconstruction method in the reconstruction processing unit , the first reconstruction method is selected when the parallax detected by the parallax detection unit is larger than a predetermined threshold, and when the parallax is smaller than a predetermined threshold, and a reconstruction method selecting means for selecting a second reconstruction process,
The reconstruction processing means outputs the image reconstructed by the splicing processing means when the first reconstruction method is selected by the reconstruction method selection means, and the reconstruction method selection means outputs the image if the second reconstruction method is selected and outputting a reconstructed image by the pixel rearrangement processing means.

According to a second aspect of the present invention, in the image input apparatus according to the first aspect of the invention, the size of the image reconstructed by the reconstruction processing means by the splicing processing means or the pixel rearrangement processing means. Size normalization processing means for normalizing the image and outputting a normalized image by the size normalization processing means.

According to a third aspect of the present invention, in the image input apparatus according to the first or second aspect of the invention, the imaging optical system includes a light source that illuminates the living body.

The personal authentication device according to the invention of claim 4 is:
An image input device according to any one of claims 1 to 3 ,
Authentication processing means for performing personal authentication by a matching operation between the image output from the reconstruction processing means of the image input device or the feature thereof and the image stored as registrant information in the registration information storage means or the feature thereof. It is characterized by having.

According to a fifth aspect of the present invention, in addition to the configuration of the personal authentication device according to the fourth aspect of the invention, an image output from the reconstruction processing unit or a characteristic thereof is stored in the registration information storage unit as registrant information. It has a means for memorizing.

The image input device of the present invention has the following advantages and is suitable as a subject image input unit of a personal authentication device.
(1) Since the imaging optical system is a compound eye optical system using a lens array, the imaging optical system can be easily reduced in thickness, and thus the entire apparatus can be easily reduced in thickness.
(2) The reconstruction processing unit is configured to be able to perform reconstruction processing by two types of reconstruction methods (joining method and pixel rearrangement method), and implements the reconstruction method selected by the reconstruction method selection unit. Accordingly, it is possible to input a subject image with a stable quality reconstructed by a reconstruction method suitable for each case from a case where the subject distance is large (parallax is small) to a case where the subject distance is small (parallax is large) .
(3) By including the size normalization processing means in the reconstruction processing means , it is possible to input a subject image having a constant size regardless of which reconstruction method is selected. This is important when the subject is used for personal authentication and is susceptible to variations in image size.
(4) By including a light source for illuminating the living body in the imaging optical system, a compound eye image with good contrast can be taken even in an environment where the amount of light is insufficient, such as natural light or indoor lighting, and a subject image with good contrast can be reproduced. Can be configured and entered.

Since the personal authentication device of the present invention uses the above-described image input device as the subject image input means, the entire device can be easily reduced in thickness, and a highly accurate individual even if the subject distance varies. It has advantages such as being able to authenticate.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Here, a human finger is assumed as a living body, an image input device that inputs a vein pattern image by imaging a vein inside the finger as a subject, and a vein pattern image or an image input by the image input device. A personal authentication apparatus that uses the features for personal authentication will be described.

  FIG. 1 is an explanatory diagram of a configuration of an image input device and a personal authentication device according to an embodiment of the present invention. In FIG. 1, 1 schematically represents a finger observed from the lateral direction. Reference numeral 2 schematically shows a vein as a subject inside the finger.

  Reference numeral 10 denotes an imaging optical system for imaging the vein 2 inside the finger 1 as a compound eye image. In this imaging optical system 10, reference numeral 3 denotes a lens array for forming a subject image, and has a structure in which a plurality of aspherical single lenses are two-dimensionally arrayed in a plane orthogonal to the lens optical axis. However, the single lens of the lens array 3 may be a single-sided spherical lens or a double-sided spherical lens. Reference numeral 4 denotes a light shielding wall, which prevents crosstalk on the image plane of the light beam that has passed through each single lens of the lens array 3 and suppresses noise light such as ghost and flare.

  The lens array 3 is manufactured by a processing method such as a reflow method, an area gradation mask method, and a polishing method using a transparent resin or glass as a material, or a molding method using a mold manufactured by these processing methods. can do. The light shielding wall 4 can be produced by punching a flat plate made of resin, glass, metal or the like by etching, polishing, laser processing or the like. The light shielding wall 4 is desirably made of an opaque material or coated with a transparent material so that light transmission and reflection can be suppressed.

  In the image pickup optical system 10, reference numeral 5 denotes an image pickup element for picking up an image formed by the lens array 3, and the light receiving elements 5 a are arranged in a two-dimensional array. As this image sensor 5, a general CCD image sensor or a CMOS image sensor can be used. In addition, although the image imaged on the imaging surface of the image pick-up element 5 is a compound eye image which is a collection of images (single-eye images) formed by individual single lenses on the lens array 3, each individual eye image The light receiving elements 5a are arranged at a density capable of capturing an image as a necessary number of pixels. The image sensor 5 has a circuit that adjusts the gain of the photoelectric conversion signal by the light receiving element 5a or converts the analog signal into a digital signal, and outputs the captured image as digital image data. Reference numeral 6 denotes a transparent glass plate for protecting the imaging surface of the imaging element 5.

  Reference numeral 7 denotes a light source for irradiating the finger 1 with light in the near-infrared band having a low absorption rate by the living body. As the light source 6, a light emitting diode (LED), a laser diode (LD), or an array light source thereof can be used. Reference numeral 8 denotes an optical bandpass filter that allows only light in the vicinity of the emission wavelength of the light source 7 to pass therethrough. Near-infrared light applied to the finger 1 by the light source 7 is absorbed by reduced hemoglobin in the vein (subject) 2 inside the finger, but is hardly absorbed by tissues other than the vein, so that the vein 2 is seen through as a dark pattern. Is possible. Therefore, the lens array 3 forms a vein pattern image on the image pickup surface of the image pickup device 5 with a dark vein pattern and a bright portion.

  Although only one light source 7 is shown in FIG. 1, a plurality of light sources 7 may be provided so that a predetermined range of the finger 1 can be illuminated. In FIG. 1, the light source 7 is arranged so as to illuminate from the direction opposite to the lens array 3 with the finger 1 in between. However, since the illumination light is diffused in all directions inside the finger 1, the lens array The light source 7 may be arranged so as to illuminate from the side 3 or the side of the finger 1. In addition, since natural light such as sunlight includes near infrared rays, the light source 7 can be omitted when it is assumed that the image can be captured with natural light or indoor lighting with sufficient contrast. However, if the light source 7 is provided, a compound eye image with good contrast can be captured without being affected by the environment. In the case where it is not necessary to consider the influence of noise light other than near-infrared light, the optical bandpass filter 8 can be omitted.

  Since the imaging optical system 10 described above is a compound eye optical system that images a subject as a compound eye image through the lens array 3, the entire thickness can be easily reduced. This is advantageous when mounted on a small electronic device.

  FIG. 5 schematically shows a compound eye image formed on the imaging surface of the imaging device 5. Here, a compound eye image when a human face is the subject is shown for convenience, but it is actually a compound eye image of a vein pattern. The grid-like dark part in the illustrated compound eye image is a shadow of the light shielding wall 4. A square area divided by the shadow is a single-eye image corresponding to each single lens. In FIG. 1, 2a shows the field of view of each single lens, and 2b is an area shared by the field of view of adjacent single lenses. Therefore, adjacent single-eye images in the compound-eye image to be formed have such a shared area. An area of the single lens field of view 2a that is not shared with the field of adjacent single lenses corresponds to “parallax” between the two single lenses. The adjacent single-eye images in the compound-eye image are images that are relatively shifted in the x and y directions by the amount of parallax.

  Please refer to FIG. The image input unit 100 is means for taking in compound eye image data (hereinafter simply referred to as a compound eye image) output from the image sensor 5. The single-eye image extraction unit 101 is means for extracting a single-eye image from the captured compound eye image. The parallax detection unit 102 is means for detecting parallax between single-eye images in a compound-eye image.

  The reconstruction processing unit 103 is means for reconstructing a single image from a large number of single-eye images in a compound eye image. The reconstruction processing unit 103 includes a joining processing unit 104A for reconstruction processing by the joining method (first reconstruction method), and a reconstruction processing by the pixel rearrangement method (second reconstruction method). The pixel rearrangement processing unit 104B and the size normalization processing unit 105 that performs processing for normalizing the size of the reconstructed image by the stitching processing unit 104A or the pixel rearrangement processing unit 104B.

  The reconstruction method selection unit 106 is a unit that performs selection of a reconstruction method in the reconstruction processing 103 (that is, selection of which one of the joining processing unit 104 and the pixel rearrangement processing unit 104B is used). The reconstruction method selection unit 106 has a function of automatically selecting a reconstruction method according to the magnitude of parallax and a function of selecting one reconstruction method according to designation from an operation unit (not shown). The elements described so far are constituent elements of the image input apparatus.

  The authentication processing unit 107 stores the reconstructed image by the reconstruction processing unit 103 or its features (for example, the branch position and the number of branches of the vein pattern in the reconstructed image) and the registration information storage unit 110 as registrant information. It is a processing means for performing personal authentication to determine whether or not the owner of the finger 1 is the registrant by matching calculation with similar images or features thereof. The authentication processing unit 107 includes a feature extraction unit 108 and a matching calculation unit 109. The registration processing unit 111 is a means for storing the reconstructed image (vein pattern image) by the reconstruction processing unit 103 or the feature extracted by the feature extraction unit 108 in the registration information storage unit 110 as registrant information.

  FIG. 2 is a flowchart for explaining operations of the image input apparatus and the personal authentication apparatus according to the present embodiment. Hereinafter, the operation of each unit will be described in detail along the flow of processing shown in this flowchart.

  First, a compound eye image of a vein pattern imaged by the image sensor 5 is captured by the image input unit 100 and stored in a memory (not shown) (step S100).

  Next, the single-eye image extraction unit 101 extracts a region of each single-eye image in the compound-eye image on the memory (step S101). As schematically shown in FIG. 5, each single-eye image in the compound-eye image is divided by the shadow portion of the light-shielding wall. Since the light-shielding wall shadow portion is the darkest, the luminance of the compound-eye image is set to an appropriate threshold value. By detecting the light-shielding wall shadow portion by comparing with the above, it is possible to easily recognize the region of the single-eye image that is a square region surrounded by the light-shielding wall shadow portion. In this way, a set of single-eye images excluding the shading wall shadow portion can be accessed from the reconstruction processing unit 101 and the parallax detection unit 102.

  Next, the parallax between the single-eye images is detected by the parallax detection unit 102 (step S102). In this parallax detection process, first, a pair of single-eye images for parallax detection is selected from the same number of single-eye images as a single lens on the lens array 3. In order to detect parallax, it is necessary to use a single-eye image including a vein pattern. For example, the determination of whether or not a vein pattern is included in a single-eye image is performed by calculating a luminance difference between adjacent pixels in the horizontal and vertical (x, y) directions of the image, and the luminance difference is equal to or greater than a predetermined threshold value. It can be done by a method to check if there is.

  Next, the parallax in the x and y directions is obtained by substituting the selected luminance distribution of the pair of single-eye images into equation (1) of the block matching calculation.

In the above (1), I _B is the reference side ommatidium image, I _m is the other ommatidium images. While gradually changing Px and Py, the sum of squares E of the luminance deviations of the two-eye images is calculated for all the pixels constituting the single-eye image, and Px and Py when the values are minimum are in the x and y directions. Is the parallax between the two eye images. FIG. 3 is a bird's eye view showing changes in E when Px is taken as the x axis, Py is taken as the y axis, and E is taken as the z axis, and Px and Py are changed. As described above, the shift amount (unit: length) in the x and y directions of the other single-eye image with respect to one single-eye image is obtained as the parallax. The detected parallax is input to the reconstruction processing unit 103 and the reconstruction processing method selection unit 106. Note that the parallax can be obtained by using a method of maximizing the cross-correlation between single-eye images instead of the method of minimizing the sum of squares of luminance deviations.

  By operating an operation unit (not shown) included in the personal authentication device or an operation unit such as an electronic device on which the personal authentication device is mounted, the reconfiguration method selection unit 106 is notified of the reconfiguration method via the control signal 200. Automatic selection and selection of any reconstruction method can be specified.

  When the automatic selection of the reconstruction method is designated (step S103, Yes), the reconstruction method selection unit 106 performs a comparison determination between the parallax detected by the parallax detection unit 102 and the predetermined threshold value TH (step S104). ). When it is determined that the parallax is greater than the threshold TH (step S104, Yes), the stitching method is selected by operating the stitching processing unit 104A (step S105), and when it is determined that the parallax is equal to or smaller than the threshold TH. (Step S104, No), the pixel rearrangement method is selected to activate the pixel rearrangement processing unit 104B (Step S106).

  When the joining method is designated instead of the automatic selection of the reconstruction method (step S103, No, step S107, Yes), the reconstruction method selection unit 106 selects the joining method and selects the joining processing unit 104A. Operate (step S105). When the pixel rearrangement method is designated instead of the automatic selection of the reconstruction method (Step S103, No, Step S107, No), the reconstruction method selection unit 106 selects the pixel rearrangement method and selects the pixel rearrangement processing unit. 104B is operated (step S106).

  Here, the relationship between the subject distance / parallax and the reconstruction method will be described with reference to FIG. Due to individual differences such as finger skin thickness, the subject distance varies considerably. As shown in FIG. 4A, when the vein 2 as a subject is at a position far from the lens array 3, the shared field 2b of the field of view of the adjacent single lens becomes large. The parallax between the eye images is reduced. As shown in FIG. 4B, when the vein 2 that is the subject approaches the lens array 3, the shared area 2b of the field of view of the adjacent single lens becomes extremely small. Therefore, a single-eye image formed by both single lenses. The parallax between them is almost the maximum.

  In the present invention, a stitching method and a pixel rearrangement method can be selected as a method for reconstructing a single image from a plurality of single-eye images in a compound-eye image. The joining method is a method in which adjacent single-eye images are inverted vertically and horizontally and then simply joined so that their shared areas overlap, and the processing load is lighter and higher-speed processing is possible than the pixel rearrangement method. . However, in the stitching method, since one common area of adjacent single-eye images is invalidated, the parallax is small as shown in FIG. 4A, and therefore, invalid when the shared area between adjacent single-eye images is large. The number of regions is increased, and the number of pixels (size) of the reconstructed image is reduced accordingly, and the resolution is lowered. In contrast, the pixel rearrangement method is a method for reconstructing a single image from a plurality of single-eye images by using parallax and rearranging the pixel luminances of the single-eye images. Even if the parallax is small and the shared area between single-eye images is large as in a), a reconstructed image with high resolving power can be generated.

  On the other hand, as shown in FIG. 4B, when the shared area of the field of view of the adjacent single lens is extremely small and the parallax becomes large, the detection accuracy of the parallax deteriorates, so that the quality of the reconstructed image is deteriorated by the pixel rearrangement method. There is a fear. In this case, since the reconstructed image of a certain quality can be obtained even by simply joining the single-eye images inverted vertically and horizontally by the stitching method, the stitching method is used in terms of stability of the image quality. It is advantageous.

  If the imaging distance can be made sufficiently shorter than the subject distance, the situation as shown in FIG. 4B can be avoided. However, since the transparent glass plate 6 for protection is usually provided on the front surface of the imaging surface, it is necessary to provide a distance between the lens array 3 and the imaging surface, so there is a limit to shortening the imaging distance. is there. In addition, the variation range of the subject distance is considerably large due to individual differences such as skin thickness. Therefore, in reality, the situation shown in FIG. 4B must be considered.

In the present invention, in consideration of the above circumstances, a reconstruction method can be automatically selected according to the magnitude of parallax as described above. That is, by appropriately selecting the parallax determination threshold TH, the stitching method is selected as the reconstruction method in the parallax range where the stitching method is advantageous, and the pixel is used as the reconstruction method in the parallax range where the pixel rearrangement method is advantageous. Select a relocation method. In this way, it is possible to reconstruct a single image of a required quality from the case where the parallax is small as shown in FIG. 4 to the case where the parallax is large .

  In addition, in the case of a child, the subject distance is generally smaller than that of an adult. Even in adults, women generally tend to have a shorter subject distance than men. Accordingly, there may be a case where it is desired to specify a reconstruction method based on the attribute of the person to be authenticated. In consideration of this point, in the present embodiment, the joining method or the pixel rearrangement method can be arbitrarily designated by the control signal 200 in accordance with an operation of an operation unit (not shown) or the like.

  Next, in the reconstruction processing unit 103, a process of reconstructing a single image from a plurality of single-eye images in a compound eye image by the reconstruction method selected by the reconstruction method selection unit 106 is combined processing unit 104A or pixel The relocation processing unit 104B executes (Step S108 or S109). Since the size of the reconstructed image changes depending on the reconstruction method and parallax, the size normalization processing unit 105 of the reconstruction processing unit 103 performs a process of normalizing the reconstructed image to a predetermined size (step S110). ). However, this size normalization process can be omitted when there is no problem in image size variation on the authentication processing unit 107 side.

  The splicing processing unit 104 performs reconstruction processing by a splicing method. As schematically shown in the lower part of FIG. 5, this processing is performed so that the adjacent single-eye image (1) and single-eye image (2) are reversed from left to right and upside down, and then the shared areas overlap. It is a process to connect to. An example of reconstruction processing by the joining method when the parallaxes in the x and y directions between the single-eye images can be regarded as being equal will be described with reference to FIG. Here, the single-eye images in the uppermost row (first row) of the compound-eye image shown in FIG. 5 are (0, 0), (0, 1), (0, 2),. . The single-eye images in the second row below are represented by (1, 0), (1, 1), (1, 2),. . . It shall be expressed with the coordinates of.

  As shown in FIG. 6A, the single-line images in the first row are connected so as to overlap the shared areas after being inverted vertically and horizontally. In other words, the single-eye image (0, 0) after flipping up / down / left / right is written in the memory, and then the single-eye image (0, 1) after flipping up / down / left / right is shifted to a position shifted by the amount of parallax in the x direction. Write. In the first row, only the parallax in the x direction needs to be considered. The hatched area is a shared area of the single-eye image (0, 0) and the single-eye image (0, 1), and this area is overwritten by the single-eye image (0, 1). Similar writing is repeated to obtain an image obtained by joining the single-eye images in the first row.

  Next, as shown in FIG. 6B, the single-eye images (1, 0), (1, 1), (1, 2),. . . Are sequentially written at positions shifted by the amount of parallax in the x and y directions. The hatched area is a shared area of the single-eye image (1, 0) and the single-line image (0, 0) in the first row. Thus, an image obtained by joining the single-eye images up to the second row is obtained. By performing the same joining process for the single-eye images on and after the third row, it is possible to obtain a reconstructed image obtained by joining all the single-eye images. For each row, it is also possible to take a processing method such as performing a joining process as shown in FIG. 6A to generate a row image and joining adjacent row images.

  The pixel rearrangement processing unit 104 performs reconstruction processing by a pixel rearrangement method. As schematically illustrated in FIG. 7, this process is determined according to the position and parallax of the single-eye image in the reconstructed image space 8 in the memory by extracting the pixel luminance from each single-eye image 9 a in the compound-eye image 9. This is a process of reconstructing a single image in the reconstructed image space 8 by repeating the operation of rearranging the extracted pixel brightness at the position for all the pixels of each single-eye image. In addition, when a pixel having lost luminance is generated in the reconstructed image, interpolation is performed with reference to the luminance of the adjacent pixel. When the parallax is smaller than the pixel size, the reconstructed image is enlarged so that the size of the parallax is the pixel size or an integer multiple thereof, that is, the number of constituent pixels of the reconstructed image is increased, and the same pixel luminance rearrangement is performed. Can be done.

  As can be understood from the above description, the single-eye image extraction unit 101 is positioned as a preprocessing unit for the parallax detection unit 102 and the reconstruction processing unit 103. Therefore, if the parallax detection unit 102 and / or the reconstruction processing unit 103 has the function of the single-eye image extraction unit 101, it is not necessary to provide the independent single-eye image extraction unit 101. The operation as the image input device has been described above.

  When a reconstructed image is output from the reconstruction processing unit 103, authentication processing by the authentication processing unit 107 (step S112) or registration processing by the registration processing unit 111 (step S113) is executed. Which process is to be performed is determined based on the control signals 201 and 202 corresponding to the operation of an operation unit (not shown) (step S111).

  The authentication process (step S112) will be described. The authentication processing unit 107 can perform authentication processing by image matching and authentication processing by feature matching, and which authentication processing is selected is designated by the control signal 201.

  When authentication processing by feature matching is designated, features of vein patterns in the image output from the reconstruction processing unit 103, such as branch point positions and the number of branch points, are extracted by the feature extraction unit 108. The matching calculation unit 109 performs a matching calculation between the extracted feature and the similar feature of the registrant read from the registration information storage unit 110, and when the similarity equal to or higher than a predetermined level is obtained, the owner of the finger 1 is determined. Authenticate with the registrant.

  When authentication processing by image matching is designated, feature extraction by the feature extraction unit 108 is not performed. The matching calculation unit 109 performs a matching calculation between the image output from the reconstruction processing unit 103 and the similar image of the registrant read from the registration information storage unit 110, and when a similarity degree equal to or higher than a predetermined level is obtained. Authenticate with the registrant.

  The registration process (step S113) will be described. The registration processing unit 107 can perform processing for registering an image as registrant information and processing for registering a feature. Which authentication processing is selected is controlled according to the operation of an operation unit (not shown). Designated by signal 202. When processing for registering a feature is designated, the feature extraction unit 108 is activated by the control signal 201.

  When processing for registering an image is designated, the registration processing unit 111 writes the image output from the reconstruction processing unit 103 into the registration information storage unit 110. When the process for registering the feature is designated, the feature of the vein pattern in the image is extracted by the feature extraction 108, and this feature is written in the registration information storage unit 110 as registrant information by the registration processing unit 111.

  Due to differences in reconstruction methods, there may be some differences in the reconstructed image or the features extracted from it. However, the same registrant has a very high probability that the same reconstruction method is selected both at the time of registration and at the time of authentication, and thus is less susceptible to the difference in such a reconstruction method. Note that the registration processing unit 111 can be provided with a feature extraction unit similar to the feature extraction unit 107. However, it is disadvantageous in terms of cost to provide similar means in the authentication processing unit 107 and the registration processing unit 111 in an overlapping manner.

  According to another embodiment of the present invention, in the registration process, instead of the reconstructed image by the reconstruction processing unit 103, a compound eye image (a light shielding wall of a light shielding wall) that is a set of single-eye images extracted by the single-eye image extraction unit 101 is used. It is also possible to store the compound eye image from which the shadow is removed) in the registration information storage unit 111. In this case, the parallax detected by the parallax detection unit 102 is preferably stored in the registration information storage unit 110 in combination with a compound eye image. A compound eye image, which is a collection of a large number of monocular images with low resolving power and inversion of the top, bottom, left, and right of the image, is less recognizable than the reconstructed image and has high confidentiality. Therefore, registering a compound eye image is more advantageous in terms of protecting personal information than registering a reconstructed image.

  The personal authentication process using the compound eye image registered in the registration information storage unit 110 is as follows. The registrant's compound eye image and parallax are read from the registration information storage unit 110, the compound eye image and parallax are input to the reconstruction processing unit 103, and the parallax is input to the reconstruction method selection unit 106. The reconstruction method selection unit 106 selects a reconstruction method based on the parallax, the reconstruction processing by the selected reconstruction method is executed by the reconstruction processing unit 103, and the reconstructed image is input to the authentication processing unit 109. The The reconstructed image generated based on the compound eye image captured from the image input unit 100 is also input to the authentication processing unit 107. In the authentication processing unit 109, the image matching operation for the two reconstructed images is executed by the matching operation unit 109, or the feature matching operation for the features extracted from the reconstructed images by the feature extracting unit 108 is matched. This is executed by the calculation unit 109.

  When the reconstruction processing unit 103 has the function of the single-eye image extraction unit 101, the compound eye image itself captured by the image input unit 100 can be stored in the registration information storage unit 110 as registrant information. It is. In this case, the reconstruction processing unit 103 extracts a single-eye image excluding the shadow portion of the light shielding wall from the compound eye image as the registrant information, and performs the reconstruction processing to the single image.

1 is a configuration explanatory diagram of an image input device and a personal authentication device according to an embodiment of the present invention. FIG. It is a flowchart for operation | movement description of a personal authentication apparatus. The change in E accompanying the change in parallax when the parallax in the x and y directions between the single images is taken on the x and y axes and the square sum E of the pixel luminance deviation between the single images is taken on the z axis. FIG. It is the schematic diagram which showed the relationship between a to-be-photographed object distance and the visual field shared between the visual field of each single lens, and an adjacent single lens. It is a schematic diagram for demonstrating the joining process of a compound eye image and a single-eye image. It is a schematic diagram for demonstrating an example of the reconstruction process by the joining method in case the parallax between single-eye images is equal. It is a schematic diagram for demonstrating the reconstruction process by a pixel rearrangement method.

Explanation of symbols

1 finger (living body)
2 Veins (subject)
2a Field of view of single lens 2b Common area 3 of field of adjacent single lens 3 Lens array 4 Light blocking wall 5 Image sensor 6 Transparent glass plate 7 Light source 8 Optical bandpass filter 10 Imaging optical system 100 Image input unit 101 Single eye image extraction unit 102 Parallax Detection Unit 103 Reconstruction Processing Unit 104A Joint Processing Unit 104B Pixel Relocation Processing Unit 105 Size Normalization Processing Unit 106 Reconstruction Method Selection Unit 107 Authentication Processing Unit 108 Feature Extraction Unit 109 Matching Operation Unit 110 Registration Information Storage Unit 111 Registration processing department

Claims (5)

An imaging optical system for imaging a subject inside a living body as a compound eye image through a lens array in which a plurality of single lenses are arrayed; and
Parallax detection means for detecting parallax between single-eye images in a compound eye image captured by the imaging optical system;
Using the parallax detected by the parallax detection means, a single image is obtained from the plurality of single-eye images in the compound-eye image by the first reconstruction method for joining the plurality of single-eye images in the compound-eye image. The compound eye is reconstructed by using a second reconstruction method for rearranging pixel luminances of a plurality of single-eye images in the compound eye image using the disparity stitching processing means and the disparity detected by the disparity detecting means. Reconstruction processing means including pixel rearrangement processing means for reconstructing a single image from a plurality of single-eye images in the image;
As a reconstruction method in the reconstruction processing unit , the first reconstruction method is selected when the parallax detected by the parallax detection unit is larger than a predetermined threshold, and when the parallax is smaller than a predetermined threshold, and a reconstruction method selecting means for selecting a second reconstruction process,
The reconstruction processing means outputs the image reconstructed by the splicing processing means when the first reconstruction method is selected by the reconstruction method selection means, and the reconstruction method selection means outputs the image An image input apparatus that outputs an image reconstructed by the pixel rearrangement processing means when the second reconstruction method is selected. The reconstruction processing means includes a size normalization processing means for normalizing the size of the image reconstructed by the splicing processing means or the pixel rearrangement processing means, and after the normalization by the size normalization processing means The image input apparatus according to claim 1, wherein an image is output. The image input apparatus according to claim 1, wherein the imaging optical system includes a light source that illuminates the living body. An image input device according to any one of claims 1 to 3,
Authentication processing means for performing personal authentication by a matching operation between the image output from the reconstruction processing means of the image input device or the feature thereof and the image stored as registrant information in the registration information storage means or the feature thereof. A personal authentication device comprising: 5. The personal authentication apparatus according to claim 4, further comprising means for storing the image output from the reconstruction processing means or a feature thereof as registrant information in the registration information storage means.

Images