TWI547882B - Biometric recognition system, recognition method, storage medium and biometric recognition processing chip - Google Patents

Description

Biometric identification system, identification method, storage medium and biometric identification processing chip

The invention relates to an identification system, an identification method, a storage medium and an identification processing chip, in particular to a biometric identification system, an identification method, a storage medium and a biometric identification processing chip.

In the prior art, when accessing confidential documents, personal data, or access to a controlled place, the identity must be verified by a key, password, or access card, but the key or access card may be lost or copied. The risk, and the password may be forgotten or known by others, causing security concerns. Therefore, biometric systems have always been a topic of focus.

The biometric system uses the unique physiological characteristics of each person to identify the user's identity. Using biometric technology, the human body is a password, there is no need to memorize a long list of numbers that are difficult to remember, and it is not afraid of being lost and difficult to copy, not to worry about being stolen. Therefore, the use of unique biological characteristics of human physiology to identify users has become an important trend in safety technology.

Taking iris texture as an example, iris is the most unique biological feature, and has the characteristics of high accuracy, high stability and non-contact. Therefore, it has become one of the main research directions of biological characteristics.

In general, the iris recognition system first stores a clearer iris texture image (hereinafter referred to as iris image) of at least one registrant, and the clearer iris image is a camera (referred to herein as camera A). Capture. However, most of the iris recognition systems use another camera (at This is called camera B, for example, set at the entrance to capture the iris image of a person to be tested, and compares the captured image of the iris texture film with the iris image captured by camera A to determine Whether the iris image of the person to be tested has been registered in the iris recognition system. In other words, whether the person to be tested and the registrant are the same person.

In the close-range iris recognition system, the above comparison is less likely to be a problem. However, in the medium and long distance iris recognition system, the degree of recognition is degraded. The reason is: when using the camera A to capture the iris image of the registrant, it is captured at a close distance, so the captured iris image is a highly resolved image; however, the camera used in the iris recognition system B is the image of the iris texture of the person at a relatively long distance, so the resolution of the obtained iris image is low. However, the conventional iris recognition system compares the iris image with lower resolution and the iris image with higher resolution, so that the recognition of the iris recognition system is not high. This problem is called the cross-sensor matching problem. In fact, not only the iris recognition system, but also other biometric identification systems, such as face recognition, there are also problems with cross sensor matching.

In view of the above problems, an object of the present invention is to provide a biometric identification system, an identification method, a storage medium, and a biometric identification processing chip which can improve the cross sensor matching problem and improve the biometric identification rate.

To achieve the above object, a biometric identification system according to the present invention cooperates with a first memory unit that stores a database, and the biometric identification system includes a first image capturing unit and a processing unit. The first image capturing unit captures a first image of the first biometric feature of one of the subjects to be tested. The processing unit divides the first image into a plurality of first image segments, and the processing unit compares each of the first image segments with a plurality of second image segments of at least one registrant in the database, and when the processing unit confirms the When one of the image segments is similar to one of the second image segments, a third image segment associated with the second image segment similar in alignment is selected in the database, and the processing unit is compared. After the at least two second video segments, the selected third video segments are recombined to identify whether the test subject and the registrant are the same person, wherein the resolution of the third video segment is greater than the first video segment and the first image segment Solution of two video segments Degree of analysis.

In one embodiment, the first memory unit is a cloud memory or an internal memory of the processing unit.

In an embodiment, the biometric identification system further includes an image transfer unit electrically connected to the first image capture unit, and the image transfer unit transmits the first image to the processing unit.

In an embodiment, the biometric system further includes a second image capturing unit that captures a second image of the first biometric of the registrant, and the processing unit segments the second image into the second images. Fragment.

In an embodiment, the processing unit further divides a third image from the first biometric or a second biometric of the registrant into the third image segments, and associates each of the second image segments respectively. Up to one of the third image segments and generating at least one associated material.

In an embodiment, the biometric system further includes a second memory unit that stores the associated data.

In an embodiment, the first biometric or second biometric is an iris texture, a face, a fingerprint, a palm shape, a palm vein, or a finger vein.

In an embodiment, the processing unit reorganizes the selected third image segments to form a fourth image, and compares the fourth image with the third image to determine whether the tester and the registrant are For the same person.

In order to achieve the above object, a biometric identification method according to the present invention includes the steps of: capturing, by a first image capturing unit, a first image of a first biometric of one of the subject; dividing the first image into Comparing the first image segments with the plurality of second image segments in a database, wherein each of the second image segments and at least one of the plurality of third image segments in the database One of the second image segments and the third image segments are from at least one registrant, and the resolution of the third image segment is greater than the resolution of the first image segment and the second image segment; When one of the first image segments is similar to one of the second image segments, the third image segment associated with the similar second image segment is selected; and the comparison is at least two After the second video clip, it will be selected The third video segments are reorganized to determine whether the subject and the registrant are the same person.

In an embodiment, the step of establishing a database includes the following steps: capturing, by the first image capturing unit, a second image of the first biometric feature of the registrant; and dividing the second image into the second image segments; Obtaining, by a second image capturing unit, a first biometric of the registrant or a third image of a second biometric; dividing the third image into the third image segments; and respectively respectively The second image segments corresponding to the three images are associated with the third image segments and stored in the database.

In an embodiment, the identification method further includes the following steps: recombining the selected third image segments into a fourth image, wherein the third image segments are segmented by a third image; The image is compared with the third image; and when the fourth image is similar to the third image, the subject is determined to be the same person as the registrant.

In one embodiment, the database is a cloud database or an internal database of biometric processing chips.

To achieve the above objective, a storage medium according to the present invention stores a database including a plurality of first resolution image segments and a plurality of second resolution image segments. The first resolution image segments are extracted from one of the at least one registrant first biometric. The second resolution image segments are extracted from the first biometric feature or a second biometric feature of the registrant, and the first resolution image segments are respectively associated with the second resolution image segments. The resolution of the second resolution video segment is greater than the resolution of the first resolution video segment.

In an embodiment, the database further includes a plurality of image segments to be tested, wherein the image segments to be tested are captured and segmented from a first image of the first biological feature of the subject.

In an embodiment, the database further includes a recombined image, wherein each of the image segments to be tested is respectively compared with the first resolution image segments, and one of the image segments to be tested is confirmed When one of the first resolution image segments is similar, the second resolution image segment associated with the first resolution image segment similar in alignment is selected in the database, and the selected second resolutions are selected. The image segments are recombined to form a recombinant image.

In one embodiment, the storage medium is an internal memory of a cloud memory or a biometric processing chip.

In order to achieve the above object, a biometric recognition processing chip according to the present invention In cooperation with a memory unit, the memory unit stores a database, and the biometric processing chip includes a matching unit and a reconstruction unit. The comparing unit compares the plurality of first image segments with the plurality of second image segments of the at least one registrant in the database, and confirms one of the first image segments and the second images by the comparing unit When one of the segments is similar, the comparison unit selects a third image segment associated with the second image segment that is similar in alignment. The recombining unit is electrically connected to the comparing unit, and after the comparing unit compares the at least two second image segments, the recombining unit recombines the selected third image segments to identify whether the test subject and the registrant are The same person, wherein the resolution of the third video segment is greater than the resolution of the first video segment and the second video segment.

In an embodiment, the biometrics processing chip further includes a dividing unit electrically connected to the comparing unit, and dividing the first image of one of the examinees having a first biometric into the first images. Image clip.

In an embodiment, the segmentation unit further divides a second image having the first biometric of the registrant into the second image segments, and further the first biometric feature or a second biometric feature from the registrant. The third image is segmented into the third image segments, and each of the second image segments is respectively connected to one of the third image segments, and at least one associated data is generated.

In an embodiment, the recombining unit recombines the selected third image segments to form a fourth image, and the comparison unit compares the fourth image with the third image to determine whether the candidate is The registrant is the same person.

In one embodiment, the memory unit is an internal memory of a cloud memory or biometric processing chip.

In one embodiment, the biometrics processing chip further includes an image transceiving unit electrically connected to the dividing unit, the comparing unit and the recombining unit, and the biometric processing chip is connected to the database through the image transceiving unit.

According to the above, in the biometric identification system, the identification method, the storage medium, and the biometric identification processing chip of the present invention, the second image segments of the at least one registrant of the database are respectively associated with the third The image segments are associated with each other, and the resolution of the third image segment is greater than the resolution of the second image segment. In addition, through the processing unit, each of the test subject The first image segment is respectively compared with the plurality of second image segments of the registrant in the database, and when the processing unit confirms that one of the first image segments is similar to one of the second image segments, Then, a third image segment associated with the second video segment similar in comparison is selected in the database. After the processing unit compares the at least two second video segments, the selected third video segments are recombined to identify whether the test subject and the registrant are the same person. Thereby, it has been proved by practice that the present invention can effectively improve the conventional cross sensor matching problem, thereby improving the recognition rate of biometrics.

1, 1a, 1b‧‧‧Biometric Identification System

11‧‧‧First image capture unit

12‧‧‧Processing unit

121‧‧‧Dividing unit

122‧‧‧ comparison unit

123‧‧‧Reorganization unit

124, 14‧‧‧ first memory unit

125‧‧‧Image Transceiver Unit

13‧‧‧Image Transfer Unit

S01~S05, P01~P05‧‧‧ steps

P1-1~P1-N, P2-1~P2-N, P3-1~P3-N‧‧‧ video clips

1A is a functional block diagram of a biometric identification system in accordance with a preferred embodiment of the present invention.

FIG. 1B is a flow chart of a biometric identification method according to a preferred embodiment of the present invention.

FIG. 1C is a schematic diagram of an image of the biometric identification system of FIG. 1 and the identification method of FIG. 1B.

FIG. 1D is a flow chart of establishing a database according to the present invention.

FIG. 1E is a schematic diagram showing the relationship between different image segment sizes and a Hamming distance in an embodiment.

2A and 2B are respectively functional block diagrams of a biometric identification system according to various embodiments of the present invention.

FIG. 3 is a schematic diagram showing the comparison of the recognition rates of the biometric identification method and the other three identification methods of the present invention.

Hereinafter, a biometric identification system, an identification method, a storage medium, and a biometric recognition processing wafer according to a preferred embodiment of the present invention will be described with reference to the accompanying drawings, wherein like elements will be described with the same reference numerals.

1A to FIG. 1C, FIG. 1A is a functional block diagram of a biometric identification system 1 according to a preferred embodiment of the present invention, and FIG. 1B is a flowchart of a biometric identification method according to a preferred embodiment of the present invention. FIG. 1C is a schematic diagram of an image applied to the biometric identification system 1 and the identification method.

The biometric identification system 1 and the biometric identification method can be applied to, for example, It is not limited to identifying the biological characteristics of the human body. In the present embodiment, the iris texture of the human eye is taken as an example. However, in different embodiments, the biometric system 1 and the identification method can also be applied to identify biometric features such as a human face, a fingerprint, a palm shape, a palm vein, or a finger vein, or can also be applied to identify non-humans. The biological characteristics are not limited by the present invention.

The biometric identification system 1 cooperates with a first memory unit 124 storing a database, and includes a first image capturing unit 11 and a processing unit 12. The first image capturing unit 11 is, for example but not limited to, a camera or a camera having a charge-coupled device (CCD) or a CMOS. In an embodiment, the first image capturing unit 11 can be an IMR (Iris-On-the-Move System) image sensor of the Sarnoff Corporation. In addition, the processing unit 12 of the embodiment can be located, for example. In the local server (the local server and the first image capturing unit 11 do not have to be in the same room), or for example, can be located in a computer, in this embodiment, the processing unit 12 is The local server is taken as an example, and is composed of at least one dividing unit 121, a matching unit 122, and a recombining unit 123. The recombination unit 123 is electrically connected to the comparison unit 122, and the division unit 121 is electrically connected to the comparison unit 122. The dividing unit 121, the comparing unit 122, and the recombining unit 123 can be implemented by a software program and executed by a processor (for example, an MCU); or, the hardware or firmware can be applied to implement the above dividing unit. 121. The functions of the matching unit 122 and the recombining unit 123. In addition, the first memory unit 124 of the embodiment is the internal memory of the processing unit 12 (located in the local server), but in another embodiment, the first memory unit 124 can also be located in a cloud server (cloud) The server is internal to the cloud memory; or, in another embodiment, the segmentation unit 121, the comparison unit 122, the reassembly unit 123, and the first memory unit 124 may all be located in the cloud server; or In the embodiment, the dividing unit 121 is located in the local server, and the comparing unit 122, the recombining unit 123 and the first memory unit 124 are located in the cloud server, or the dividing unit 121 is integrated in the first image capturing unit 11 The present invention is not limited (more details will be described below).

Taking the airport customs system as an example, it has a plurality of customs windows, each customs window may have a first image capturing unit 11, and the processing unit 12 (the dividing unit 121, the comparing unit 122 and the reorganization unit 123) may be located in the customs computer. The local memory unit of the room, and the first memory unit 124 can store biometric information of the people of the country, and can be located at the far end for the remote memory. Server, and the local server is connected to the remote server. The image captured by the first image capturing unit 11 can be transmitted to the local server to be divided, compared, and reorganized by the processing unit through the connection between the local server and the remote server. Identify the identity of the entry and exit personnel.

In addition, in some embodiments, the matching unit 122 and the recombining unit 123 can be integrated to form a biometric identification processing chip; in other embodiments, the dividing unit 121, the comparing unit 122, and the recombining unit 123 can also be integrated. The biometric identification processing chip is fabricated, or in still other embodiments, the segmentation unit 121, the comparison unit 122, the recombination unit 123, and the first memory unit 124 can be integrated into the biometric identification processing chip. In this embodiment, the dividing unit 121, the comparing unit 122, the recombining unit 123, and the first memory unit 124 are integrated into one identification wafer. In addition, the biometric processing chip can be located within the local server or within the cloud server. Here, the biometric processing chip (including the dividing unit 121, the matching unit 122, the recombining unit 123, and the first memory unit 124) is located in the local server, or in other embodiments, the biometrics may also be The identification processing chip is integrated in the first image capturing unit 11, and the present invention is not limited. In addition, the above database may be a cloud database (located in the cloud server) or an internal database of the biometric processing chip (located on the local server). Here, the internal database of the biometric identification processing chip is taken as an example.

Before describing the biometric system and method in detail, how to create a database in the first memory unit 124 will be described. Here, the database may be referred to as a heterogeneous iris dictionary. Before establishing the database, the data of the iris texture of at least one person (called Registrants) needs to be established in advance, and the processing unit 12 can perform the comparison when receiving the first image of the person to be tested, to confirm whether the candidate is For previously registered personnel. For example, a bank needs to establish an iris database for a specific person in advance. When someone wants to enter a bank vault, it can be compared to a person whose iris information is a company-specific authority. If so, it is allowed to proceed; if not, then refuse Enter and issue a warning signal. In addition, the company's first-level supervisory authority is different from the general internal staff's authority. Generally, internal personnel can only access the general information of the company's server, but when accessing higher-order confidential information, or changing internal settings, it is necessary to pass The iris is identified and identified as a high-level supervisor of a certain level.

The step of establishing a database may include steps P01 to P05 as shown in FIG. 1D: the first image capturing unit 11 captures a second image of the first biometric of the registrant (step Step P01, Figure 1C). Here, the first image capturing unit 11 can capture the second image of the first biometric feature (human eye iris texture) of the same registrant that obtains the first image (the second image may be referred to as a first resolution image) ). Therefore, the resolution of the second image is the same as that of the first image, and is a lower resolution image, such as an image obtained at a longer distance.

Then, the second image is segmented into the second image segments (the second image segment may be referred to as a first resolution video segment) (step P02). In this case, the second image may be divided into the second image segments by the segmentation unit 121, or the image segmentation may be performed by different segmentation units (for example, located in the cloud server).

Thereafter, a second image capturing unit (for example, image sensor PIER of securiMetrics Inc., not shown) captures a first biometric of the registrant or a third image of a second biometric feature (the third image may be It is called a second resolution video clip, and its resolution is high (step P03), and the third video is divided into the third video clips (step P04). Since the second image capturing unit captures the iris image of the registrant at a relatively close distance, the third image of the registrant (built-in data) is a clearer (higher resolution) image, and thus the third image The resolution of the (slice) is higher than the resolution of the first image (slice) and the second image (slice). In addition, when the second biometric feature of the registrant is captured, the second biometric feature may be different from the first biometric feature. For example, when the first biometric feature is an iris, the second biometric feature may be a face shape, such that the third image is the The facial image of the registrant is not limited. Of course, the second biometric and the first biometric may also be the same biometric, such as an iris texture.

When the third image is sent to the biometric system 1 of the present embodiment, the third image is segmented into the third image segments by the dividing unit 121 as in the embodiment (the number of image segments needs to be the same as the first image segment or The number of second video clips is the same). In a preferred embodiment, before the second image and the third image are segmented, a global alignment operation is performed to align the boundaries of the second image and the third image and then divide the image into a second image. The clip and the third video clip. Furthermore, a partial overlap between the two adjacent first image segments is required, thereby eliminating the boundary line of the image segment during the subsequent image recombination process. In different embodiments, the second image and the third image may be subjected to image segmentation by different segmentation units (eg, located in a cloud server). In addition, the third image capturing unit captures the third image of the registrant's iris. Wherein, although the second image capturing unit is used to capture the first The third image, but in an embodiment, the second image capturing unit can also be the same image capturing unit as the first image capturing unit. In other words, the same image capturing unit can capture the first image. The second image and the third image are not limited by the present invention.

Then, the second image segments corresponding to the second image and the third image segments are associated with the third image segments, and stored in the database (step P05). Here, "generating association" means that a relatively clear third video segment and a relatively blurred second video segment are in a one-to-one connection relationship. For example, the second image and the third image are respectively divided into N video segments, that is, N second video segments and N third video segments respectively, and corresponding to different regions, that is, the second image. The segment 1 is the image of the region 1 of the second image, and the second image segment 2 is the image of the region 2, . By analogy, the second video segment N is an image of the region N. In addition, the third video segment 1 is the image of the region 1 of the third image, the third video segment 2 is the image of the region 2, and so on, and the third image segment N is the image of the region N. The processing unit 12 can generate a one-to-one association between the second image segments 1 to N corresponding to the second image and the third image segments 1 to N, respectively, because the corresponding regions are the same. So that the second video segment 1 corresponds to the third video segment 1 and the second video segment 2 corresponds to the third video segment 2, . By analogy, the second image segment N corresponds to the third image segment N, wherein the second image segment i and the third image segment i may correspond to images of different biometric features. Thereby, each of the second image segments can be associated with one of the third image segments respectively, and N related data are generated. In an embodiment, the associated data may be stored in the database of the first memory unit 124, or may be stored in a second memory unit, and the first memory unit 124 and the second memory unit may be located in the same directory. Within the server, one is located in the local server, and the other is located in the cloud server, the invention is not limited. In addition, in practice, the database generally has a second image and a third image of a plurality of registrants, and each registrant has an associated second image segment and a third image segment. In addition, the above related data may be stored in the form of a look-up table (LUT).

Next, referring to FIG. 1B again, the biometric identification method of the present invention includes the following steps S01 to S05.

First, step S01 is performed: the first image capturing unit 11 captures a first image of the first biometric feature of one of the test subjects (as shown in FIG. 1C, the first image may also be referred to as a test object. Image, or image to be recognized). The first biometric feature of this embodiment is the iris texture of the human eye. In addition, as shown in FIG. 1A , the first image captured by the first image capturing unit 11 can be transmitted to the processing unit 12 through one of the image transmitting units 13 electrically connected to the first image capturing unit 11 . The image transmission unit 13 can be a wired or wireless transmission module to transmit the first image to the processing unit 12 by wire or wirelessly. In an embodiment, the image transfer unit 13 can be integrated with the first image capture unit 11 into a wafer. Of course, in addition to the image transfer unit 13, in other embodiments, the biometric system 1 can also include an image receiving unit (not shown) for receiving data transmitted by the processing unit 12 or other devices. In addition, in an embodiment, the size of the first image may be, for example, 640×480 pixels.

After the processing unit 12 receives the first image, step S02 is performed to divide the first image into a plurality of first video clips (the first video segment may also be referred to as a video segment to be tested or a video segment to be recognized). . Here, the first image is divided by the dividing unit 121 into a plurality of first video segments of the same size. In an embodiment, the image can be segmented into, for example, 111 video segments. Since the first image capturing unit 11 is to capture the iris image of the person to be tested, the eye of the subject may not be close to the first image capturing unit 11, and therefore, the first of the captured persons to be tested The image (fragment) may not be quite clear, that is, its resolution may not be quite high. In addition, it is possible that part of the iris texture may be blocked by the eyelids or eyelashes, or the image may be partially unrecognizable due to reflection factors. Therefore, when capturing the image segment, the first image capturing unit 11 or processing The unit 12 or the biometric identification processing chip first undergoes an Iris mask estimation process to remove the image fragments that are too confusing, unrecognizable or reflective, leaving only the image segments that can be subsequently identified. The second image and the third image may also go through this processing step. Furthermore, a partial overlap between the two adjacent first image segments is required, so that the boundary of the image segment can be eliminated during the subsequent image recombination process. It should be noted that, in this embodiment, the size of the first image segment that is divided should be the same as the size of the second image segment and the third image segment. In addition, after the first image of the person to be tested is captured and image segmentation is performed, the general process may further include normalization and feature extraction. Among them, the normalization and the extraction of the eigenvalues are not the focus of the present invention, and will not be described here.

In addition, please refer to FIG. 1E, which is an embodiment, different video segments. Schematic diagram of the relationship between size and Hamming distance. In order to analyze the effect of the size of the image segment on the accuracy of the recognition system, an experiment to optimize the image size was performed. In this embodiment, the size (resolution) of the iris image segment tested is from 3×3 pixels to 30×30 pixels, and as shown in FIG. 1E, the preferred image segment size is greater than 6×6. The best image segment is 17×17 pixels. At this time, the Hamming distance is the lowest, which means the recognition error rate is the lowest.

Referring to FIG. 1B again, step S03 is performed: comparing each of the first video segments with a plurality of second video segments in the database, wherein each of the second video segments and the plurality of data frames respectively At least one of the third image segments is associated with the at least one registrant, and the third image segment has a greater resolution than the first image segment and the second image The resolution of the fragment. In this case, the first video segments of the test subject are compared with the second video segments of all the registrants in the database through the comparison unit 122, and the comparison unit 122 is based on a comparison. The principle is compared. The comparison principle can be, for example, an algorithm and can be, for example, an Orthogonal Matching Pursuit (OMP) algorithm.

Then, in step S04, when one of the first video segments is similar to one of the second video segments (or the alignment is successful), the registrant similar in comparison is selected. The third video segment associated with the second video segment. Here, when the comparing unit 122 confirms that each of the first video segments of the test subject is similar to one of the second video segments of a certain registrant (for example, when the threshold value exceeds a threshold, it is determined as The alignment unit 122 will select a third video segment in the database that is associated with the second video segment of the registrant that is similar in comparison.

Finally, step S05 is performed: after comparing the at least two second video segments, the selected third video segments are recombined to determine whether the test subject and the registrant are the same person. In this embodiment, the comparing unit 122 compares each first video segment with all the second video segments, and confirms that the first video segment is similar to a certain second video segment, as shown in FIG. 1C. For example, when it is confirmed that the first video segment P1-1 is similar to the second video segment P2-1, the comparison unit 122 may select the third video segment P3-1 associated with the second video segment P2-1. And so on. Then, the selected third image segments are recombined to form a fourth image (the fourth image may be referred to as a recombined image). Here, the reorganization unit 123 will The selected third image segments that are similar in alignment are recombined to form a fourth image. The first image and the fourth image may be stored in a temporary storage area (for example, a RAM) of the processing unit 12. Then, the fourth image and the third image are further compared by the comparison unit 122. When the comparison result of the comparison unit 122 is judged to be similar, it means that the test subject is a certain registrant in the database. Therefore, the processing unit 12 can issue a comparison result indicating that the judgment is similar, and can display the fourth image on a display unit (for example, a display screen, not shown). It is worth mentioning that, if it is applied to turn on an electronic lock device, the electronic lock device can receive the comparison similar signal and output a control signal to control the electronic lock to be turned on; or, the comparison similar signal can be entered. The first-stage access control, for example, for job title comparison, can only enter the vault when it is confirmed as an accountant or a high-ranking person. When the comparison result of the comparison unit 122 is judged to be dissimilar, it indicates that the test subject is not one of the registrants, so the processing unit 12 may issue another comparison to indicate that the judgment is dissimilar. The output comparison result signal is notified to the manager so that the manager can generate a corresponding action.

In addition, the storage medium disclosed in the present invention is a non-transitory computer readable storage medium, which may include, for example, at least one memory, a memory card, a CD, a video tape, and a Computer tape, or any combination thereof. The memory may include a read only memory (ROM), a random access memory (RAM), a flash memory, or a Field-Programmable Gate Array (FPGA), or other Form of memory, or a combination thereof. In addition, the storage medium can be a cloud memory or a biometric identification processing chip internal memory. As shown in FIG. 1A, the storage medium of the present embodiment is an internal memory of the biometric identification processing chip, and is the first memory unit 124; or, in different embodiments, the storage medium may be the first memory. The unit 124 adds the second memory unit that stores the associated data. Alternatively, the storage medium may be a second memory unit that stores only the associated data, and is not limited.

The storage medium stores the above-mentioned database, and the database includes a plurality of first resolution image segments and a plurality of second resolution image segments. The first resolution image segment is the second image segment and is segmented by the second image segment. The second resolution image segment is the third image segment and is segmented by the third image. And the first resolution image segments are extracted from the first biometric feature of at least one registrant, and the second resolution image segments are captured. The segment is taken from the first biometric or second biometric of the registrant. The first resolution video segments are associated with the second resolution video segments, and the resolution of the second resolution video segment is greater than the resolution of the first resolution video segment. In addition, the details of the database can be referred to the above description, and will not be described again.

In some embodiments, the database may further include a plurality of image segments to be tested, wherein the image segments to be tested are captured and segmented from the image of the first biological feature of the subject to be tested. The image segment to be tested may be the first image segment described above, and is segmented by the first image. In addition, in some embodiments, the database may further include a recombined image, wherein each of the image segments to be tested is respectively compared with the first resolution image segments, and the image segments to be tested are confirmed. And one of the first resolution image segments is similar to one of the first resolution image segments, and the second resolution image segment associated with the first resolution image segment similar in alignment is selected in the database, and the selected image segment is selected. The second resolution image segments are reconstructed to form a reconstructed image. Here, the reconstructed image is the fourth image described above.

In addition, please refer to FIG. 2A and FIG. 2B respectively, which are functional block diagrams of the biometric identification systems 1a and 1b according to different embodiments of the present invention, respectively.

As shown in FIG. 2A, the main difference between the biometric identification system 1a and the biometric identification system 1 is that the processing unit 12 of the present embodiment is located in the cloud server, and therefore, the first image capturing unit 11 captures The image can be transmitted to the processing unit 12 located at the cloud server through the wirelessly transmitted image transfer unit 13. In addition, the first memory unit 14 is a cloud memory. In addition, the dividing unit 121, the comparing unit 122, and the recombining unit 123 are fabricated into a biometric identification processing chip and are located in the cloud server.

In addition, as shown in FIG. 2B, the main difference between the biometric identification system 1b and the biometric identification system 1 is that the processing unit 12 of the embodiment is located in the local server, but the first memory unit 14 is located in the cloud servo. Inside, it is cloud memory. In addition, the biometric identification processing chip 1b further includes an image transceiving unit 125, and the image transceiving unit 125 is electrically connected to the dividing unit 121, the comparing unit 122, and the recombining unit 123, respectively, and the biometrics processing chip (including the dividing unit 121, The comparison unit 122, the reassembly unit 123, and the image transceiving unit 125) are connected to the database of the first memory unit 14 of the cloud server through the image transceiving unit 125 to transmit and receive images (fragments) by wireless transmission.

In addition, the biometric identification system 1a, 1b has all the technical features of the above-described biometric identification system 1 and its variants, and can be referred to the above detailed description, and will not be further described herein.

In addition, please refer to FIG. 3 , which is a schematic diagram for comparing the recognition rates of the biometric identification method and the other three identification methods. Among them, the image of curve 1 is taken as the above PIER and IOM image sensor, and is a general iris recognition method, and has not been subjected to other curves improved by the recognition rate algorithm (refer to the literature: J. Daugman "How iris recognition works," IEEE Transactions on Circuits and Systems for Video Technology, vol. 14, no. 1, pp. 21-30, Jan. 2004.); the image of curve 2 is the same as PIER and IOM images. Sensors, and results obtained using the eigeniris method (References: Li, M. Savvides, and V. Bhagavatula, 2006, "Illumination Tolerant Face Recogniton Using a Novel Face From Sketch Synthesis Approach and Advanced Correlation Filters", 2006 IEEE International Conference on Acoustics, Speech and Signal Processing, 2006. ICASSP 2006 Proceedings, vol. 2, no., pp. II, II, 14-19.); image capture of curve 3 is the same as PIER and IOM image sensor And the results obtained using the kernel learning method (References: Pillai, M.Puertas, and R. Chellappa, 2014, "Cross-sensor Iris Recognition through Kernel Learning," IEEE Transactions on Patte Rn Analysis and Machine Intelligence, vol. 36, no. 1, pp. 73-85); and curve 4 is the result obtained by applying the identification method of the present invention.

In FIG. 3, the abscissa is the percentage of False Acceptance Rate (FAR), and the ordinate is the Verification rate. It can be found from Fig. 3 that when FAR = 10 ^-2 %, the recognition rate (curve 4) of the present invention can reach 95.45%, which is better than 81.57% of curve 2 and 64.48% of curve 1. In addition, the equal error rate (EER), that is, the FAR and the False Rejection Rate (FRR) identification error rate are the same, the present invention is 0.87567%, compared with 4.7726 of the curve 1. The % is much lower, which proves that the biometric identification system and the identification method of the invention can effectively improve the biometric identification rate.

In summary, in the biometric identification system, the identification method, the storage medium, and the biometric identification processing chip of the present invention, the second image segments of the at least one registrant of the database are respectively associated with the third Image segments are associated with each other and the third image segment is parsed The degree is higher than the resolution of the second image segment. In addition, each of the first video segments is compared with the plurality of second video segments of the registrant in the database through the processing unit, and the processing unit confirms one of the first video segments and the second When one of the image segments is similar, a third image segment associated with the second image segment similar in alignment is selected in the database. In addition, after the processing unit compares the at least two second video segments, the selected third video segments are recombined to identify whether the test subject and the registrant are the same person, wherein the third video segment is parsed. The degree is greater than the resolution of the first image segment and the second image segment. Thereby, the actual identification proves that the invention can effectively improve the conventional cross sensor matching problem, thereby improving the recognition rate of the biometrics.

The above is intended to be illustrative only and not limiting. Any equivalent modifications or alterations to the spirit and scope of the invention are intended to be included in the scope of the appended claims.

1‧‧‧Biometric System

11‧‧‧First image capture unit

12‧‧‧Processing unit

121‧‧‧Dividing unit

122‧‧‧ comparison unit

123‧‧‧Reorganization unit

124‧‧‧First memory unit

13‧‧‧Image Transfer Unit

Claims (25)

A biometric identification system cooperates with a first memory unit that stores a database, the biometric identification system includes: a first image capturing unit that captures a first biometric feature of one of the subjects to be tested And a processing unit, the first image is divided into a plurality of first image segments, and the processing unit compares each of the first image segments with a plurality of second image segments of at least one registrant in the database, And when the processing unit confirms that one of the first image segments is similar to one of the second image segments, selecting one of the databases associated with the second image segment that is similar in alignment a third image segment, and after the processing unit compares at least two of the second image segments, the selected third image segments are recombined to identify whether the test subject and the registrant are the same person, wherein The resolution of the third video segment is greater than the resolution of the first video segment and the second video segment. The biometric identification system of claim 1, wherein the first memory unit is a cloud memory or an internal memory of the processing unit. The biometric identification system of claim 1, further comprising: an image transfer unit electrically connected to the first image capture unit, and the image transfer unit transmits the first image to the processing unit . The biometric identification system of claim 1, further comprising: a second image capturing unit that captures a second image of the first biometric of the registrant, and the processing unit The two images are segmented into the second image segments. The biometric identification system of claim 4, wherein the processing unit further divides a third image from the first biometric or a second biometric of the registrant into the third images. And segmenting each of the second image segments to one of the third image segments and generating at least one associated material. The biometric identification system of claim 5, further comprising: a second memory unit for storing the associated data. The biometric identification system of claim 5, wherein the first biometric feature Or the second biometric is iris texture, face shape, fingerprint, palm shape, palm vein, or finger vein. The biometric identification system of claim 5, wherein the processing unit recombines the selected third image segments to form a fourth image, and the fourth image and the third image The comparison is made to determine whether the subject is the same person as the registrant. A biometric identification method includes the following steps: capturing, by a first image capturing unit, a first image of a first biometric feature of a subject; dividing the first image into a plurality of first video segments; Each of the first image segments is compared with a plurality of second image segments in a database, wherein each of the second image segments is associated with at least one of the plurality of third image segments in the database, and The second image segment and the third image segment are from at least one registrant, and the resolution of the third image segment is greater than the resolution of the first image segment and the second image segment; When one of the image segments is similar to one of the second image segments, the third image segment associated with the second image segment is selected; and the comparison is performed at least two After the two video segments, the selected third video segments are recombined to determine whether the test subject and the registrant are the same person. The identification method of claim 9, wherein the step of establishing the database comprises the steps of: capturing, by the first image capturing unit, a second image of the first biometric of the registrant; Dividing the second image into the second image segments; capturing, by the second image capturing unit, the first biometric feature of the registrant or a third image of the second biometric feature; And storing the second image segment and the second image segment corresponding to the third image and the third image segment respectively, and storing the data in the database. The identification method of claim 10, wherein the first biometric or the second biometric is an iris texture, a face shape, a fingerprint, a palm shape, a palm vein, or a finger vein. The identification method of claim 9, further comprising the steps of: recombining the selected third image segments into a fourth image, wherein the third image segments are segmented by a third image. Comparing the fourth image with the third image; and when the fourth image is similar to the third image, determining that the subject is the same person as the registrant. The identification method of claim 9, wherein the database is a cloud database or an internal database of a biometric processing chip. A storage medium storing a database, the database comprising: a plurality of first resolution image segments, the first biometric feature captured from at least one of the registrants; and a plurality of second resolution video segments, the system capturing The first biometric feature or a second biometric feature of the registrant; wherein the first resolution video segments are respectively associated with the second resolution video segments, and the second resolution video segment is The resolution is greater than the resolution of the first resolution video segment. The storage medium of claim 14, wherein the database further comprises: a plurality of image segments to be tested, wherein the image to be measured is captured and segmented from a first biometric of the subject. The storage medium of claim 15 , wherein the database further comprises: a recombination image, wherein each of the image segments to be tested are respectively compared with the first resolution image segments, and the When one of the image segments to be tested is similar to one of the first resolution image segments, the second resolution image associated with the first resolution image segment similar to the comparison is selected in the database And segmenting the selected second resolution image segments to form the reconstructed image. The storage medium of claim 14, wherein the first biometric feature or the second feature is an iris texture, a face shape, a fingerprint, a palm shape, a palm vein, or a finger vein. The storage medium of claim 14, which is a cloud memory or an internal memory of a biometric processing chip. A biometric identification processing chip is coupled to a memory unit, wherein the memory unit stores a database, the biometric identification processing chip includes: a comparison unit, respectively registering the plurality of first image segments with at least one of the databases The plurality of second image segments are aligned, and when the comparison unit confirms that one of the first image segments is similar to one of the second image segments, the comparison unit selects the data a third image segment associated with the second image segment similar in comparison; and a recombining unit electrically connected to the comparing unit, and comparing the at least two second images to the comparing unit After the segment, the recombining unit reassembles the selected third image segments to identify whether the test subject and the registrant are the same person, wherein the third video segment has a resolution greater than the first video segment and The second image segment resolution. The biometric identification processing chip according to claim 19, further comprising: a dividing unit electrically connected to the comparing unit, and the first image of one of the examinees having a first biometric feature Divided into the first video segments. The biometric identification processing chip according to claim 19, wherein the dividing unit further divides a second image having the first biometric of the registrant into the second image segments, and more A third image of the first biometric or a second biometric of the registrant is segmented into the third image segments, and each of the second image segments is respectively connected to the third image segments. First, and generate at least one associated material. The biometric identification processing chip according to claim 21, wherein the recombining unit recombines the selected third image segments to form a fourth image, and the comparison unit further compares the fourth image with The third image is compared to determine whether the subject is the same person as the registrant. The biometric identification processing chip according to claim 21, wherein the first biometric feature or the second biometric feature is an iris texture, a face shape, a fingerprint, a palm shape, a palm vein, or a finger vein. The biometric identification processing chip according to claim 19, wherein the memory unit is a cloud memory or an internal memory of the biometric processing chip. The biometric identification processing chip of claim 20, further comprising: an image transceiving unit electrically connected to the dividing unit, the comparing unit and the recombining unit, wherein the biometric identification processing chip passes through the The image transceiving unit is connected to the database.