TW200807306A - System for and method of assigning confidence values to fingerprint minutiae points - Google Patents

Abstract

A method in accordance with the present invention is used to organize minutiae and other identifying characteristics of a patterned object to verify a user. In one embodiment, the method includes matching minutiae points of the patterned object in one more sets with minutiae points of the patterned object in a candidate set. The method also includes adjusting a confidence value of each minutia point in the one or more sets based on results of the matching. The method also includes organizing the one or more sets based on the confidence values, such as by ordering the minutiae in the one or more sets based on the confidence values, or by deleting from the one or more sets any identifying characteristics with a confidence value below a threshold.

Description

200807306 IX. Description of invention: [Related application] The application of this application is hereby incorporated by reference. The number of the patent in the patent of the year of the year is __.. 9 (8), the title "distribution of confidence value to the fingerprint Method, _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ System and method of biological characteristics [First as technology] Since, fingerprint processing is a mature technology that is gradually more extensive and faster. Itr (four) with calculation method can be used to determine the details of the two fingerprint templates 5 疋 = match. Usually, These matching calculus relies on individual details of confidence. Hunting removes low confidence details to improve performance or reduce the complexity of matching processing. Previous technical systems assign these details by their local and local touches. Confidence value. These image characteristics are not closely related to the stomach, so the county can make the future test (4) how to match - the reliable confidence value of the detail point. This may be estimated to delete or cut The number of details in the template is the most important. [There is no transfer of confidence value or deletion of individual templates. Ben Lemming assigns confidence values by using matching template results to each other. 5 200807306 Patterns in the template, each template It is created by scanning different images of the same finger. Because the invention relies on matching to estimate the probability of future matching, the goods are not hanged, and are the exact points for accurate matching in future verification or recognition attempts. Possible superior predictor. Another advantage of this method is that the confidence value can be re-evaluated to be more reliable, and (iv) more specific, with additional detail mode

Board data and does not access the original fingerprint image. It is not possible to use a prior art system. 'A invention of the first special wealth, the method of the system is used to verify the visual identity of the patterned object. The square note includes the patterned object recognition characteristics in the one or more groups, four =, the patterned object recognition characteristics in the candidate group. The method also includes the second: the matching result is based on adjusting each of the identification features and the centroid values in the one or more groups, and organizing the one or more groups based on the confidence value. : Sorting the identification of the one or more groups based on the heart value H or more _ in addition to having a lower than - threshold - letter or two silk group (four) i money record. - Implementation ^ Two ML systems are silent. Alternatively, the secret is dynamic. The defamation ^ section template. In another embodiment, 匕3 adds the leeway group to the one or more groups. In the second feature of the present invention, an object verification method for inserting a seed, a seed, a pattern, and a second pattern is used to generate an object image generation-candidate identification characteristic template; the lion is to be difficult to find a material. The pattern should be. The value is based on the execution of the match to verify the patterned object. As used herein, "match C matches · try and produce "success," or "fail, match."

...: In the embodiment, the identification characteristic is fingerprint detail. In this embodiment, the method also includes using a finger-fingerprint to read the patterned object image. The recognition characteristic includes fingerprint details. In this other embodiment, the method also includes the visual mode scanner reading the patterned object image. The one or more other templates are generated from one or more images of the earned object. The candidate for the job (4) is matched to the stored identifiability based on the confidence value. Preferably, the financial system is in the order of highest to lowest confidence value. The method also includes calibrating the bribed lion with the one or more other templates to perform a more precise match. In the third feature of the present invention, the template generating method corresponding to the image data includes one or more pairwise matchings for performing two or more templates each containing the data of the corresponding identifying characteristic, and matching between the identifying characteristics Based on this, a confidence value is assigned to each of the identification characteristics in each of the two or more templates. In one embodiment, the confidence value is assigned to an identification characteristic by an added value update confidence value associated with the identification characteristic. The added value depends on the type of match between the matching characteristics that are matched during the pairwise match. The match is "failed match", "successful loose" match, "successful rigor, match, package 7 200807306 success match within the box, or these matches. 々 Preferably, each read match contains - the first The order, which includes a low-level and a second-order, which includes a strict or high tolerance match: the heart-specific wealth, the age-based confidence value to the graphically-identified object recognition feature, including a plate construction And the distribution of crying. Miscellaneous board builder storage configuration and construction knowing ^ each template contains the mapping = letter 'read to - related literacy - record. The dispatcher is configured to, the bottom "value to - the first Each of the identities in the template matches each of the identities 'calls' - one or more other template towels. The dispatcher is also configured to organize, based on the confidence value, all of the identification characteristics of the '5 模板 template builder' to discard less than one door. As used herein, a configuration device is used to throw software, hardware, firmware, any other device that configures a component, and any combination of these.

The device also includes a biosensor coupled to the formwork builder. The biosensor is configured to read a patterned object corresponding to the identification characteristic and to retrieve the data. In one embodiment, the biosensor is a finger image sensor. In another embodiment, the biosensor is a visual mode sensor. Preferably, the apparatus also includes a template repository configured to store one or more templates containing data mapping confidence values to identification characteristics. The apparatus also includes a matcher that matches the patterned object recognition characteristics and the template database identification characteristics read by the biosensor in a predetermined manner, thereby verifying the patterned object. The predetermined method includes comparing the identification characteristics in the order of the 200807306 η. % confidence value. Preferably, the matcher includes one of the first order loose matchers and one of the first order strict matchers. If the first order determines a successful match and reaches a second value greater than the first value, and if the second order determines a successful match, then the matcher is configured to increase a confidence value by a first value. The apparatus also includes a master device configured to perform a function based on a successful match between a candidate template and a template in the template database. The 10 纟 device is a mobile phone, a personal computer, a digital camera, a digital video player, a digital voice/video player, or a health/monitor device. In a fifth feature of the present invention, a patterned object detail matching device includes a finger sensor that can read the image of the patterned object; a template builder configured to construct a detail template from the image; a storage, comprising one or more templates of one of the templates, wherein each of the ones or the other may wish to pass the confidence value from the imaged object to the image, to the detail, or to find a template towel The detail is based on the confidence value - the organization; and - the matcher is configured (4) to match any template pairing to verify the patterned object. The apparatus also includes a dispatcher configured to assign a confidence value to the detail based on the matching result, and to organize the template based on the confidence value. In the sixth feature of the present invention, the computer-readable system can store a data structure containing a plurality of records. Each record contains a - column containing information indicating a detail point and a second block containing information indicating the confidence value for one of the minutiae points. 9 200807306 [Embodiment] The present invention is used to confirm a patterned object such as a finger by matching the identification characteristics of the patterned object with the stored identification characteristics of the patterned object. Matches based on the confidence value associated with the identification characteristics. For example, these identification features with the highest confidence values are first matched to the recognition characteristics of the image being read. In other embodiments, data corresponding to the identification characteristics of low confidence values are deleted, thereby saving money, and for example, portable devices such as mobile phones, digital cameras, and digital voice players are always considered. ^ Figure 1 shows a system 100 for verifying the image of the finger 11 after the finger 110 is placed on the finger sensor 1〇5 or across the finger sensor 105. System 100 includes a verification template 140 that is coupled to finger sensor 105. The detail points from the finger 110 are captured and matched to the minutiae generated by other scans of the finger 11 as captured below during the registration or initialization process. In accordance with the present invention, the finger UG (and - user) is quickly verified, more efficiently matched, and uses less memory. Figure 2 shows that the viewfinder image read by the screen mode 155 is verified to authorize a user to use the system 160 of the system 160. The screen mode scanner 155 is coupled to the verification template 157. The details of the network model are extracted and matched to the genus of the pre-existing bribery. It will be appreciated that the vehicle and the eye-and-eye image can be verified in accordance with the present invention, and that identification characteristics other than the details of the screen mode can also be used. The figure is a system 2 〇〇 high level pattern in accordance with the present invention. System 200 includes a biosensor 2〇5 that is lightly coupled to template generator 210 coupled to verification template 220 in sequence 200807306. In operation, biosensor 205 can read an image of one of the fingerprints to capture an image of the patterned object. The template generator can generate a template T1 containing the identifying characteristics of the captured image. The verification template 22〇 can match the recognition characteristics of the other templates generated by the 7F of the τι information table to other images of the object to be verified to verify the patterned object. More

Preferably, the identification characteristics produced by the other images of the patterned object are stored in the Tmult towel. Each of the templates Tmult contains a characteristic (four) indicating the other image, and a record associating the confidence value with the corresponding identification characteristic. This confidence value is updated to reflect more or less confidence in the _ identification feature. For example, because the recognition feature with a smaller confidence value finds more of the patterned object, the recognition feature has a high degree of centroid (a high confidence value). Because - literacy can be found in a number of such scanned images, or can be associated with temporary scratches on the sensor on one of the patterned objects [identification characteristics can have - low confidence values . According to the invention, when the image is scanned, the scanned identification characteristic is assigned to a characteristic with high confidence. Xuan Shi = :, not even with low confidence (10) _ miscellaneous, not to mention matching, hunting this money and the processing time required to perform the match. If the pre-employment is successfully matched with the age-recognized feature, the imaged object and the patterned object are verified. This threshold can be set using different criteria. Example ΐ When you need to check the money for a few low-level tasks on the computer age = 200807306 You must successfully match a few identification features. When the user must be authenticated to perform a high level task, more recognition characteristics must be successfully matched. Fig. 4, which is used to explain the following example, shows the fingerprint image 3GG read by the finger sensor. The fingerprint image 3GG contains identification characteristics that are used to identify the finger. Here, the identification feature includes minutiae points 301A-C and 301E_I, which include ridge endpoints and bifurcation points.

The display J is not used to capture fingerprint details and generate a detail template as a component 310 for identifying characteristics. Component 310 includes a head sensor 325, such as a finger swipe sensor or a finger displacement sensor, coupled to a template generator 325. The plate generator 325 contains a detail 315 that is lightly coupled to the formwork builder 32 for construction, panel T1. Preferably, the template T1 contains a record that links the confidence value to the data representing the minutiae point. When the identification characteristics in the templates are matched to the identification characteristics in the other templates, the confidence value is initialized and later adjusted (increase, decrease or not change). In an alternative embodiment, the data structure that is different from the template = but is associated with the confidence value. - Applying a towel, the details are light and light. Depending on the two or more templates taken from two or more different scans of the same fingerprint, the heart value is divided into the respective points of each of the difficult towels. Davide Maltoni, Dario Maio, Anil K. Jain ^ Sali, Prabhakar ^ ^ (2 (T+Springef first edition)) are the details of the match. The details of these deceleration boards Tj i are In order to interact with the model's (four) entity points. In fact, the matching state can be Τ-the second group of detail points in Figure 1, so that the pin frequency "^二二^- the next group of details 〒 any given minutiae Hey, it is known whether or not 200807306 a) friend points with the board Tj 2 + position M, or the point has no corresponding point. In addition, 'you can call the department has a job point M, the point m is located in T - Figure 1 The matching detail points are enclosed in the box or externally connected to it. A similar analysis is performed on all the fine points in Fig. 2. The US Patent No. 6'546, 122 'titled by Rus is the fingerprint template for each sensing area. U.S. Patent No. 6,681,034, entitled ","

Board, method and New Zealand" money details · fingerprint matching device. β 6A ® two different patterns of fingerprint images, graphics and Figure 2. Compared with f, the different patterns are scanned by the same finger The data representation in the generated template. 6A_ shows how to determine whether the detail point (such as point B) in Figure i has the corresponding detail point in Figure 2 (such as point B). As explained below, this correspondence can be determined - Whether or not the confidence value of the detail point is adjusted. In the example of '° - the detail point (point B in the figure!) has 1 should be, and the field phase in the other figure (point B in Figure 2) , the confidence value can be increased. Otherwise, no. As explained in more detail below, the confidence value can also be adjusted according to the detail position: one = the increments of these large _ packs in the box to adjust for many reasons For example, the adjustment process and the simplification of the matching algorithm are used to make the I box, and the amount of processing that must be performed during the matching is reduced. The sixth figure also shows that there are corresponding detail points B, C, F, , , ίί in FIG. And the corresponding detail points B, c, F, G, Ή and I in Figure l. The figure shows the picture of Figure 6A! and Figure 2 However, each map has two predetermined 1 packets for all the minutiae contained in T—, and the confidence value 13 200807306 is assigned and updated as follows: • If the minutiae is matched or not connected to the bounding box, the confidence value is increased by CjioMateh • If the minutiae is successfully matched, the confidence value is increased

CjtnatchFinal • Confidence value is increased c_extemal if the minutiae point is outside the bounding box at the last matched minutiae position

The bounding box in the preferred embodiment is defined as a rectangle that minimally encompasses all of the matched minutiae points. The skilled artisan will be able to clarify other definitions, including that the rectangle is rotated, converted, or oriented toward the east to match the two fingerprints and the other 'corner' where the trapezoid or other multiple (four) is used to surround the matched point. If the minutiae is failed to match but is outside the bounding box, it may not match successfully because it is not part of the image that was matched to the image. Therefore, it is not taken into account.

Confidence value CjuatchFinal C matchFinal C matchFinal Refer to Figure 6B, assuming all the minutiae points start with the confidence value '' then Table 1 shows the confidence of the confidence dispatch process. 200807306 G C_matchFinal G, CjnatchFinal Η CmatchFinal H, CmatchFinal I CmatchFinal Γ CjmatchFinal r C An external table 1 In the preferred embodiment, a matcher has two orders. The first order uses loose valleys to match the fine points' and then tougher (such as stricter or tighter) tolerances. This first order is less likely to delete the minutiae that should be successfully matched, so it can add a right stem to the far heart. It is also possible to operate a single-order matcher twice to obtain similar results with different parameter settings, one loose and one tight. It will be apparent to those skilled in the art that any order and parameter settings, either alone or in combination, can be used to produce a finer granularity of confidence values, such as many matching metrics of matching errors that can be calculated per detail basis. In the latter case, C-external ’ C—matchFinal and C-noMatch are not fixed constants in the letter. In all of these cases, the bounding box is still defined as a rectangle that minimally surrounds the last matched minutiae point. However, in an alternative embodiment, the bounding box is defined by these minutiae points of the first order, except for the most. Or use different subsidies based on the results of each class/or parameter setting. In the preferred embodiment, the confidence values are assigned and updated for all the minutiae points contained in τ1 and 一1: • If the minutiae is matched or failed in the order 1 or the last order, In the bounding box, the confidence value is increased. C noMatCh 15 200807306 • If the detail point is successfully matched in the first stage, the confidence value is increased by C matchFinal. • If the detail point is successfully matched in the last stage, the confidence value is increased. —matchFinal • If the minutiae is outside the bounding box, the confidence value is increased. C external is also in the case, CjioMatch=0, c external^,

C_matchStagel=8 ’ and C_matchFinal=32. These incremental values represent the relative importance of each order and characteristics. However, other values and weights are also possible, including values less than zero. The process can also be repeated for multiple templates only by matching each other to the other board and assigning trusts. For example, if three templates can be used together - Figure i, ^_ Figure 2 and T - Figure 3 replace only two templates, then all confidence values can be set to zero, then all dependable group pairs are assigned by the following confidence Operation: In any order: • τ一图1 and Fig. 2 • τ一图1 and τ—Fig. 3 • Τ一图2 and 丁_图3 Generally speaking, all the models of the contact system are secret. It is advantageous to skip several pairs to save computer resources. Can,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, In the alternative embodiment, the above detail confidence adjustment technique is used in combination with the prior art 16 200807306 to form a more precise technique than the individual. Still further, in an alternative embodiment, the detail confidence adjustment technique is only applied to the slate, and the severance verification is used on the template obtained in the basin where only the single template is available. ^ Figure 7 is a flow chart of step 400 in accordance with the present gamma - implementation of the sequel - or more templates. This process begins with a step. Next, in step 410, all the detail confidence values in all templates are initialized to a value, such as 〇. Next, in step 415, the first pair of templates are matched, and in step 42, the confidence value is assigned based on the matching result and the bounding box as described above. Step 42 is based on the present invention, by sorting the details on the basis of their reliability values, sorting the details on the basis of their confidence values, deleting the details on the basis of their confidence values, or any combination of these template. In the step milk, decide if there are more detail templates to match. If there are no more, the process proceeds to its end step n n the process proceeds to step 430 where the next pair of templates is retrieved, and then proceeds to step 420. Figure 8A shows the match diagram! And the template 5 of Figure 1 of Figure 6a produced after the template of Figure 2. The template 5〇〇 contains a detail point of 52 〇 (for example, the data indicating the detail point), corresponding to one of the confidence values column 525, and the detail point and the corresponding confidence value column or record 501-510. Column 501 shows that detail point A (block 520) has a corresponding confidence value of 2 (column 525), while column 5.2 shows that thin node B (column 520) has a corresponding confidence value of 32 (block 525). The confidence values in the other columns are interpreted the same. Figure 8B shows the template 5〇〇 after it has been matched with the pepper plate corresponding to other images of the object (such as its image), so that the confidence value is adjusted. Similar components are labeled by similar numbers for all figures in the application. Figure 8B shows the corresponding confidence value of the _ a _ 441 (column 50), which is sufficient, the detail point β has a corresponding confidence value of 48 (column view, block 52 〇), and the detail point c has a corresponding confidence value (column 503) , block 520), etc., during operation, when the finger image is scaled, the record in the "candidate" template generated when the finger is scanned is matched to the record in the template 5. The scanned candidate The minutiae points are matched to the minutiae points in the template 5〇〇 to determine if there is sufficient minutiae between the successful matches. The template is considered to be a sinful finger-image-reflection detail. Matching to the details in the template 500, the green is successfully matched by several detail points above the threshold. The scanned system is verified. Usually, the record is continuously compared, and the 1 finger refers to a certain number of details with a heart value. The details are matched before being matched. Since the verification process can be performed more quickly, this is not enough to determine the details that are more likely to be located in the two templates. Furthermore, the entire template 500 is stored in the memory. - the system can store each package Many templates are in many detail points, so the memory is wasted. According to the present invention, the template leak is organized to create the template 500 of Figure 8C. The board 500" has been organized so that the detail points can be based on their confidence values from the highest So the lowest ordering. Therefore, the detail point B with the highest confidence value of 48 is ranked in the column 5〇1 + is sorted the most. The detail point A with the second highest confidence value of μ is then sorted in column 5〇2 Preferably, the template 500" is also deleted, so that the detail points having a confidence value lower than a threshold are deleted from the template 5", thereby saving money. For example, 18 807306 if the deletion threshold is 10 And the confidence value of the detail point, _胄·uf has less than 10. The preferred embodiment is wide;;::, from the table, the section and the truncation of the lowest confidence point or just the hunting by the confidence value sorting the detail to delete. It is based on the addition of factors and confidence values to attach the template table to supplement the ^ maximum coverage for the verification of the B picture (such as the first and delete threshold can be used as in the initialization of the cattle system Memory based on two: two, as it is, The threshold can be increased to limit the use of the café. The threshold can also be - system security!; = stored in the memory at a low security level, the syllabus is lower than the base. In this application, the door can also be determined by the sum of the confidence values. The sum of the heart values is at or near - the predetermined details of the letter ^=_存_ is added to the letter. When scanning, you can increase this threshold = base, not just the total sum to store two scans. If the confidence scale is the same, then the size of the cone is changed. The private bottle is the edge of the shot. The deletion calculus can be performed in accordance with the present invention. The ninth diagram is a step of verifying the fingerprint using the verification module in accordance with the present invention. The _ secret is initialized and the first step is 6 〇 ι. In step 603, the fingerprint image generation-candidate template is captured from the platform as described above and it is mapped to the next template in the template (four) library. Bu, the next "template is for the first time to be stored. Preferably, it is the detail point in the next template. Therefore, it is preferred that 'when the minutiae in the candidate template is continuously and the details in the candidate template=timed' is matched using the smallest (deleted) group minutiae point with acceptable values from the highest to the most niche value The point of detail. In step 6〇5, if the number of details of the candidate template towel is successfully matched with the next template (if the template is successfully matched), the process proceeds to step 6〇7, yielding, and successful template matching (“verification”). ) The result, and then the milk to the end of the process. Step 605 + ' If the detail threshold number fails to match, the process continues to step 609 to determine if there are more templates in the template repository for checking. If the template database has no more templates, the process proceeds to step 613, which results in a "failed module match, ("verification failed") result, and then proceeds to step 615. Step 6〇9, if If it is determined that there are more templates in the template database for checking, the process proceeds to step 611 to retrieve the lower panel, and the process then proceeds to step 603. It should be understood that although the above description corresponds to a template image of a single object, The template of the multi-patterned object can also be used in accordance with the present invention. For example, the present invention can be used to verify different fingers of a single user, or fingers of different users. Figure 1 is a verification mode of Figure 3. A block diagram of the component 220. The verification module 220 includes an input that can receive the template T1. The verification module 22 includes 20 200807306 a matcher 230 configured to match the detail template and update the confidence value as described above. The template database 225 stored in a computer readable medium receives the template 270, and matches the template in the template database 225 to the template T1. The matching zhai 230 has a confidence adjustment by the Moma. a first output, which can use the matching result to adjust the confidence value in the template. The confidence adjustment state 260 can organize the template based on the confidence value, and input the organized template back to the template bin 225. Figure 9 illustrates that the matcher 230 has its ability to generate a template match (success or failure) that is used to verify the second output of the patterned object, such as a finger. Preferably, confidence adjuster 260 is the same component discussed elsewhere as the confidence dispatcher. When a biological additional image is scanned, the single component assigns values during system initialization and adjusts these values later. In some embodiments, the single element is dependent on the processing steps (initialization or update) performed, involving a confidence adjuster or a confidence dispatcher. In other embodiments, the letter-adjustment state or the confidence dispatcher is a different component. Φ A 11A shows a matcher 230 in accordance with the invention of the present invention. Matching zarn 230 includes a first order 13^ coupled to a second order 24〇. Preferably, the first order 235 is a loose matcher that uses a wide valley limit to perform the match, and the second order is a strict matcher that uses a tighter tolerance to perform the match. Figure 11 shows a matcher 23A according to another embodiment of the present invention. The matcher 230 includes a single order 245 that matches the template in a multi-pass using a feedback loop. Preferably, the first pass can perform a loose match, and the first and subsequent pass uses a gradual match. 21 200807306 ^ Example Exercise (4), the image multiple sweep nMi θ ^ corresponds to her system is generated and stored in the template database μ 2: ° 夕 重 重 重 以 以 以 以 以 以 以 以 以 以 以 以 以 指 指 指 指 指 指 指 指 指 指 指 指 指 指Fine pointing to the ride, when it is called by r Tian; Ά 'system exists - corresponding template. The templates are matched to each other using pairwise matches, and the confidence values for each detail in each template are determined. The template is _ to make no job less, so its success can be matched to other; Preferably, the (four) scan is performed during the pointing or initializing step.

The template is later listened to as the authenticator. The user places her finger across a finger sensor or swipes her finger. - Candidate templates are generated from this scan and are matched to templates in the template repository. The match is performed based on the plate organization in the template database. The user is only verified if the match between the template and the template in the template database is successful. For a number of implementations, the thief template is added to the template repository. Once it is tested, it can be accessed. The system can perform certain functions, such as accessing protected files or launching specific programs. While the above discussion has focused on fingerprint sensors, it should be understood that the present invention can be used to verify other organisms such as the visual network model using identification features such as visual mode details and palm prints. The skilled artisan will appreciate that the invention can be modified in many ways. For example, a template in accordance with the present invention can be directed to different data structures that can actually store identification characteristics. In this way, the template will contain the components that correspond to the characteristics. These are the elements that can be arranged, sorted or deleted in accordance with the present invention. 22 200807306 It should be understood that the above description of the steps of the embodiments of the invention may be modified in many green forms in accordance with the invention. Repair can add a few spicy, you can delete the ^ dry steps, and can be called Tong Yuqing, the county to perform this step. Those skilled in the art will readily appreciate that other modifications may be made to the embodiment without departing from the spirit and scope defined by the scope of the application patent.

23 200807306 [Simple Description of the Drawings] Fig. 1 is a diagram showing a system high level pattern of a user using a fingerprint sensor in accordance with an embodiment of the present invention. The second diagram is a system high level pattern for verifying a user using a visual mode scanner in accordance with an embodiment of the present invention. Figure 3 is a block diagram of a template builder and a matcher in accordance with an embodiment of the present invention. • Figure 4 shows a fingerprint with a circle detail point. Figure 5 is a fingerprint sensor, a detail picker, and a template builder in accordance with an embodiment of the present invention. Figure 6A shows one of the two figures matching the fingerprint, and the line is between the corresponding details of the two figures. Figure 6B shows the bounding box of the two figures in Figure 6A. Figure 7 is a flow chart showing the processing steps for assigning detail confidence values in accordance with an embodiment of the present invention. _ Figures 8A through C show templates after the initialization and adjustment steps in accordance with an embodiment of the present invention. Figure 9 is a flow chart showing the processing steps of authenticating a user in accordance with an embodiment of the present invention. The figure is a high level block diagram of a template matcher in accordance with an embodiment of the present invention. 11A and B are diagrams showing a high level block diagram of a matcher in accordance with an embodiment of the present invention. 24 200807306

[Description of main component symbols] 100, 200, 150, 160 system 105 ^ 311 finger sensor 110 140, 157, 220 155 205 210 > 325 225 230 235 240 245 260 270 300

301A-I 310 315 320 Finger Validation Template Retina Scanner Biosensor Template Generation Template Lean Library Matcher First Order Second Order Single Order Confidence Receiver Receive Template Fingerprint Image Detail Point Component Detail Picker Template Builder 25

Claims (1)

200807306 X. Patent application scope: L A method for organizing the recognition characteristics of a patterned object, which is used to verify a patterned object, the method comprising: · the pattern in one or more groups The recognition characteristic of the object is matched with the recognition characteristic of the patterned object in a candidate group; and based on the result of the matching, a confidence value of each recognition characteristic in the one or more groups is adjusted. 2. The method of claim i, further comprising organizing the one or more groups based on the k~ value. 3. The method of claim 2, wherein organizing the one or more groups comprises ranking the identification characteristics in the one or more groups based on the confidence value. 4. The method of claim 2, wherein the one or more groups are organized to include any identifying characteristic from the one or more _ in addition to having a confidence value below one. 5• If the method described in item 4 of the patent consumption, the door broadcast system is 6. If the patent application scope is the fourth dynamic. The method of claim 7, wherein the threshold is 7·fingerprint detail template 26 8. 200807306 forming a template corresponding to the visual mode details, such as the method described in claim i, further comprising including the candidate set in The one or more groups. A method of verifying a patterned object, comprising: generating a template of candidate identification characteristics from an image of the patterned object; and mapping the candidate identification characteristic to the patterned one or more of the other templates The stored identification characteristics of the object are matched, wherein the matching system = the confidence value associated with the stored identification characteristic is used to perform the verification of the matching object. The method of claim 10, wherein the identifying characteristic t is a finger detail, and the method further comprises using the finger to capture an image of the patterned object. The method of claim 10, wherein the identifying characteristic comprises a visual mode detail. The method further comprises reading the image of the patterned object using a visual mode sensor. 13. The method of claim 10, wherein the template is from the patterned object - or more images. 14. The method of claim 10, wherein In a basic manner, the candidate identification characteristics XD~values are matched. 4 Gu Cunlang characteristics 15. If the method described in the _na 14 item, the method is 9. 10. 11· 12 27 200807306 from the highest to the lowest confidence value - the order. The method of U.S. Patent Application Serial No. 1G, further comprising calibrating a template of the candidate identification characteristics with the - or more other templates. A method for generating a map corresponding to image data, comprising: binding one or more pairs of matches of each of the two disparity templates containing the data of the sage; and f is based on a match between the identification characteristics, Dispatch-confidence values to each of the recognition characteristics in two or more templates. The method of claim 1, wherein the identification characteristic is a fingerprint detail point. 19. The method of claim π, wherein the identifying characteristic is a visual mode detail. The method of claim 17, wherein assigning a confidence value to an identification characteristic comprises updating the confidence value associated with the identification characteristic by an incremental value. 21. The method of claim 2, wherein the incremental value is dependent on a matching type between the identification characteristics matched during the matching period. 22. The method of claim 21, wherein the matching type is a failed match, a successful loose match, a successful strict match, a successful match, or a combination of these. The method of claim 17, wherein the two or more templates include a record of each confidence value to a recognition characteristic, the method further comprising organizing each of the corresponding confidence values. The 28 200807306 record. 24. The method of claim 23, wherein organizing each of the records comprises sorting the records according to their confidence values. 25. The method of claim 23, wherein organizing each of the records comprises deleting any record having a corresponding confidence value below one threshold. 26. The method of claim 17, wherein each pairwise match comprises a first order and a second order. 27. The method of claim 26, wherein the first order comprises - loose, and the second order comprises a strict match with the loose match. 28. Apparatus for assigning confidence values to the identified characteristics of a patterned object, comprising: - a slab builder, a thin & template for constructing identification features, wherein each template contains a mapping - confidence value a record of the phase-to-phase characteristics; and a dispatcher' configured to assign a confidence value to each of the recognition characteristics, from the recognition feature in a template to the patterning in one or more other templates A match is made between the recognition characteristics of the object. ~ y, two - are used to organize each record according to the confidence value. 30. The device of claim 29, wherein the confidence value is based on a sequence of _ records. Related devices such as the device described in claim 29, 29 200807306 32. 33. • 34· 35· 36. 37. 38. 39. Any record with a confidence value below one threshold to organize the record . The device of claim 28, further comprising a biosensor coupled to the template builder and configured to read a patterned object corresponding to the identification characteristic and to retrieve data. The device of claim 32, wherein the biosensing system is a finger image sensor and the identification characteristic is a finger image detail. The device of claim 3, wherein the biosensing system is a visual mode sensor, and the identification characteristic comprises a visual pattern detail. The apparatus of claim 28, further comprising a template library configured to store one or more templates, the one or more templates comprising data mapping confidence values to identifying characteristics. The device of claim 32, further comprising a matching device that can determine a recognition characteristic of the patterned object read by the biometric finger and an identification characteristic of the template database, thereby Verify the patterned object. If the device of claim 30 is applied, the predetermined method of the towel includes matching the identification characteristics in an order determined by the confidence value. The device of claim 36, wherein the matching device comprises a first order and a second order. The method of claim 38, wherein the first-order hit-loose match, and the second-order execution is a device according to the match 40 item of the strict match 30 200807306, wherein The first order: = work: match, then the matcher is configured to increase - confidence = 'second order successful match, then increase - "the value reaches a second value greater than the first value. The device described in claim 35, further comprising a main broken 'configured for age-appropriate---------------------------------------------------------------------- The miscellaneous item described in item 41, wherein the main device is a personal computer... digital camera, digital voice = benefit, digital voice/video player, or a health/monitor device. A device for identifying characteristics of a 77-spot & heart-to-sampling object, comprising: Means for constructing a template for identifying characteristics, wherein each template includes a record mapping a confidence value to a related identification characteristic; and, = assigning a confidence value to the identification characteristic from one template to one or more other templates A device for each of the identification characteristics generated by the matching of the identification characteristics of the patterned object. 44. A device for matching details of a patterned object, comprising: a camera sensor for reading an image of the patterned object, a template builder configured to The image constructs a detail template; a storage, a database containing one or more templates, wherein each of the 31 200807306 one or more templates can capture confidence values from one of the patterned objects to the details. Wherein the detail in the template or templates is an organization having the confidence value; and ', _ is used to match any template pair on the basis of the _ to verify the patterned object. 45. In the case of the device described in claim 44, the step further comprises a score of 46. :: the touch assigns a confidence value to the sire based on the matching result between the templates, and organizes the confidence value based on the confidence value. template. Top: A computer with a data structure including multiple records can be used. The second record contains the first column and the second column. The second = indicates the details, and the second block contains the information. Hai,,, Tian Lang points a piece of confidence value information. 32