JP4430262B2 - Input fingerprint state determination method and apparatus - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an input fingerprint state determination method and apparatus.
[0002]
In recent years, personal authentication devices using fingerprints have been put into practical use and have been used. However, in a personal authentication system using biometrics (biometric information) using fingerprints, rough fingers, dry fingers, and wet fingers are input. If a fingerprint that is not suitable for fingerprint authentication due to factors such as a thin fingerprint is input for creating registration template data, it may be possible to create data that is not suitable as a registration template. In addition, when a similar fingerprint is input for identity authentication, it may be impossible to clear the matching rate for authenticating the identity.
[0003]
The present invention determines whether an input image is suitable for fingerprint authentication without performing image processing, which is generally said to take time like thinning an image obtained from a fingerprint sensor at the software level. At the same time, the fingerprint determined to be unsuitable for fingerprint authentication relates to a method for determining a fingerprint state such as dry, wet, rough finger, and thin fingers.
[0004]
[Prior art]
As a conventional technique for determining the state of a finger, for example, there is a “fingerprint recognition discrimination device dry finger detection method” disclosed in JP-A-5-309082. This method is shown as a configuration of the conventional example shown in FIG. In this conventional example, a finger 81 is placed on the fingerprint detection unit 80, the resistance value is converted into a voltage value by the resistance value detection electrode 82, and the digital signal is converted by the A / D converter or comparator 83 to be determined by the CPU 84. Measures the dryness or wetness of However, in this conventional method, detection is performed by hardware, and hardware (resistance value detection electrode 82, A / D converter or comparator 83, etc.) is used to change the degree of dryness or moisture to be detected. The cost must be changed and inevitably an increase in cost is inevitable.
[0005]
In order to avoid this, it is conceivable to control whether to capture an image obtained by an existing fingerprint sensor by software. For example, as a timing for capturing an input fingerprint, a process of permitting capture only by the area of the input image can be considered. Specifically, the image input from the fingerprint information input unit is subdivided (blocked) into a predetermined size, and a histogram of each block is calculated. The histogram is obtained by calculating the number of pixels distributed in each brightness, with the brightness of the pixels in each block being 256 for black and 0 for black, and the average value of the histogram clears a predetermined threshold value. If there is a fingerprint, it is recognized that there are fingerprints (ridge lines (mountain lines) and valley lines) on the block. If the ratio between the total number of blocks divided and the number of blocks in which a fingerprint exists clears a predetermined threshold, the input image is captured as having an appropriate fingerprint, and user authentication is performed.
[0006]
FIG. 16 shows an example of a fingerprint image used for explaining the conventional example. This image is a 256 gradation square of 300 pixels vertically by 300 pixels horizontally. Although this image is subdivided, the size of one block to be subdivided can be appropriately compared by the ratio of the total number of blocks and the number of blocks in which data existed finally, and the processing speed can be subdivided so finely If you only pay attention to the fact that there is no effect on, you can block it to any size. For example, in the case of an image as shown in FIG. 16 (vertical 300 pixels × horizontal 300 pixels), a total of 225 blocks including 15 vertical and horizontal blocks may be used. In this case, the size of one block is 400 pixels of 20 pixels × 20 pixels. FIG. 17 shows a state where the image of FIG. 14 is subdivided.
[0007]
Next, the histogram of each block is calculated, and the average value of the histogram is obtained. For example, FIG. 18 shows pixel distributions of two blocks, and A. 17 shows the seventh block in the seventh column of FIG. 17, and this block definitely has ridges and valleys. The average value of the histogram of this block is about 210. Conversely, B. of FIG. In the 13th block in the 14th column of FIG. 17, it is difficult to determine that there are data, but there are ridges (convex lines) and valley lines (concave lines). The average value of the histogram of this block is about 250 (a block of pixels close to white). Whether the average value obtained in this way exceeds a predetermined threshold is determined, and the block is determined to have a fingerprint, and the block that does not exceed is determined to have no fingerprint. The average value calculated in this way varies depending on the density of the ridges and the number of ridges in the block. However, if the size of one block is 400 pixels, a block in which ridges and valleys are present appropriately. The average value of the histograms is between about 100 and 230. From this, when the average value of the histogram is within this range (hereinafter referred to as a threshold range), it can be determined that a fingerprint exists. It is necessary to use an optimum value for the threshold range depending on the size of the block to be subdivided and the background color of the fingerprint image.
[0008]
In this way, each block is determined, and when the ratio of blocks determined to have a fingerprint exceeds a predetermined threshold (for example, 80%) with respect to all blocks, it is determined that there is a fingerprint.
[0009]
As described above, even with the conventional technology, it is possible to determine whether an image input from the fingerprint input device is input in an area suitable for collation, and an image having a small area can be discarded in advance. It was.
[0010]
[Problems to be solved by the invention]
As described above, even in the above-described prior art, it is possible to determine whether an image input from the fingerprint input device is input in an area suitable for collation, and an image with a small area can be discarded in advance. is there. However, since this conventional technique does not actually confirm the presence of the ridge, it must be said that the determination result is unstable. In addition, since only the area of the collected fingerprint is determined, the state of the fingerprint cannot be determined, and even when a finger that is rough enough to be recognized as a fingerprint is input, a certain area is obtained. Even if it satisfies, there is a problem that it will be taken in, and the authentication rate will be significantly reduced.
[0011]
The present invention solves the above-mentioned problems of the prior art and simplifies whether the fingerprint input from the fingerprint information input unit is an image suitable for fingerprint authentication without performing image processing that requires a lot of calculation time. The purpose is to provide a method and apparatus for determining an input fingerprint state that can be accurately discriminated with a simple configuration, and when it is determined that it is not suitable for fingerprint authentication, the reason is that it is dry, wet, rough finger, and thin fingers Another purpose is to detect and notify.
[0012]
[Means for Solving the Problems]
In the personal authentication technology of the present invention, as a pre-process for performing fingerprint authentication, an input image that is not suitable for fingerprint authentication is determined accurately from the image input to the fingerprint sensor, such as dry, wet, rough finger, and thin fingers. Can be excluded in advance. As a result, it is possible to avoid a case where fingerprint authentication is executed with a fingerprint image that is not suitable for fingerprint authentication and authentication failure occurs. In addition, it is possible to realize a high-accuracy fingerprint verification system without changing the existing fingerprint sensor, by refusing to capture images that are not suitable for fingerprint authentication and instructing countermeasures at the software level. It becomes.
[0013]
FIG. 1 shows a principle configuration of the present invention, and FIG. 2 shows a processing flow according to the present invention.
[0014]
In FIG. 1, 1 is a fingerprint information input unit, 2 is an image subdivision unit, 3 is a subdivision image ridge presence determination unit, 4 is a subdivision image state determination unit, 5 is an image capture determination unit, and 6 is an image state determination. Part.
[0015]
The operation of FIG. 1 will be described with reference to the processing flow of FIG. 2. A person who wants to register a fingerprint or a person who has already registered a fingerprint and wants to receive authentication places his finger at the input position of the fingerprint information input unit 1. Then, a predetermined area is scanned and fingerprint information (fingerprint image) is input (S1 in FIG. 2). Next, the image subdivision unit 2 subdivides the fingerprint image into blocks with a predetermined size (S2 in FIG. 2). Next, in the subdivision image ridge presence determination unit 3, existence of ridges in all subdivision images (blocks) focusing on the continuity of pixel black (ridges or noise) and white (valley lines or background) Is determined (S3 in FIG. 2). Further, the subdivided image state determination unit 4 determines the image state from the state of the pixel of the subdivided image, whether it is wet, dry, or wet or dry (S4 in FIG. 2). ). The result of the determination for each block by the segmented image ridge presence determining unit 3 and the segmented image state determining unit 4 is supplied to the image capture determining unit 5, and the total number of blocks and the number of blocks for which the presence of the fingerprint can be confirmed. 2 and the determination result of the image state, the input fingerprint state is determined (S5 in FIG. 2), and it is determined whether the image is suitable for fingerprint authentication (S6 in FIG. 2). As a result, when it is determined that it is suitable for fingerprint authentication, the process proceeds to a predetermined process by a fingerprint authentication mechanism or a fingerprint registration mechanism (not shown) (S7 in FIG. 2). When it is determined that it is not suitable for fingerprint authentication, the image state determination unit 6 determines the reason why it is not suitable for fingerprint authentication, and again proceeds to fingerprint input processing. At this time, by providing image state notification means (display means or sound output means not shown), the reason why it is determined that it is not suitable for fingerprint authentication and the countermeasures for improving it are shown to the user. The user can perform fingerprint authentication without feeling stressed.
[0016]
As described above, by using the method of the present invention, it is possible to quickly determine whether an input image is suitable for fingerprint authentication without performing complicated image processing, and thus the determination result is an image not suitable for fingerprint authentication. If there is, it is possible to perform fingerprint authentication only with an image suitable for fingerprint authentication by discarding the image and repeating collection.
[0017]
DETAILED DESCRIPTION OF THE INVENTION
FIG. 3 shows a configuration of the embodiment. In the figure, 1 to 5 are the same as those of the same reference numerals in FIG. 1, 1 is a fingerprint information input unit, 2 is an image segmentation unit, and 3 ′ is a segmentation for determining whether a fingerprint exists for a segmented image An image fingerprint presence determination unit (corresponding to the segmented image ridge presence determination unit 3 in FIG. 1), 4 is a segmented image state determination unit, 5 is an image capture determination unit, and 1a is an image binarization unit. 6a to 6c correspond to the image state determination unit 6 in FIG. 1, 6a is a thin finger determination unit, 6b is a dry fingerprint determination unit, and 6c is a wet fingerprint determination unit.
[0018]
FIG. 4 is a processing flow of the embodiment, and the operation of the embodiment will be described below. First, a fingerprint image is input to the fingerprint information input unit 1 (S1 in FIG. 4), and the input fingerprint information is binarized by the image binarization unit 1a. At this time, there are various methods for determining the threshold value for binarization processing. For example, a mode method is used in which an intermediate value is obtained using a histogram and binarization is performed depending on whether the intermediate value is larger or smaller. However, any method may be used. Next, the binarized image is subdivided into a predetermined size in the image subdivision unit 2 (S2 in FIG. 4). All of the subdivided images are subjected to the presence determination of the fingerprint by determining the presence of the ridge in the subdivision image fingerprint presence determination unit 3 ′, and at the same time, the subdivided image state determination unit 4 wets the image from the pixel state. It is determined whether it is in a tendency, dry, or neither wet nor dry (S3 in FIG. 4). A detailed method for determining the presence of a ridge will be described later with reference to FIGS.
[0019]
Each block determined in this manner is calculated by calculating the ratio of the total number of subdivisions and the number of blocks in which the presence of the fingerprint can be confirmed in the image capture determination unit 5 (S4 in FIG. 4), and determining whether the image is suitable for fingerprint authentication. (S5). As a result, if it is determined that it is suitable for fingerprint authentication, the process proceeds to a predetermined registration verification process by a fingerprint authentication function unit (not shown) or a fingerprint registration unit (S6 in FIG. 4). If it is determined that it is not suitable for fingerprint authentication, the process proceeds to determination of the state of the input image in order to investigate the reason why it is determined not to be suitable for fingerprint authentication (S7 in FIG. 4). Determination of whether the finger is thin (the fingerprint area is small) (S8, S9 in FIG. 4), determination of whether it is a dry fingerprint by the dry fingerprint determination unit 6b (S10, S11 in FIG. 4), wet fingerprint by the wet fingerprint determination unit 6c If it is not a wet fingerprint, it is determined that the finger is rough (S14 in FIG. 4), and if it matches this determination element, it is not suitable for fingerprint authentication. The determination reason is output to output means (a display device not shown), the user is notified, and the input of the fingerprint from the fingerprint information input unit 1 is awaited again. Details of the process for determining the reason for determining that it is not suitable for the fingerprint authentication will be described later with reference to FIGS. Needless to say, the processing order for obtaining the reason why it is determined not to be suitable for fingerprint authentication can be freely changed.
[0020]
Here, the detailed operation of the embodiment will be described using a specific example of a fingerprint image. FIG. 5 shows an example of an input fingerprint image. In this fingerprint image, an image of 15 mm length × 15 mm width, which is a necessary size as a fingerprint image, is a square of 300 pixels × 300 pixels, and the background color is white. By binarizing the input fingerprint image shown in FIG. 5 by the image binarization unit 1a, an image shown as an example of the binarized image in FIG. 6 can be obtained. In this example, the threshold is binarized using 230. When the binarized image shown in FIG. 6 is subdivided by the image subdivision unit 2, the ridge interval of a person's fingerprint varies depending on the person, and a wide part and a narrow part coexist on the same finger. It is known that 1 to 4 distributions are formed in 5 mm. In this embodiment, 0.5 mm × 0.5 mm square is subdivided into 10 pixels × 10 pixels (100 pixels) area (block), and the image size of this 100 pixel area is adopted. That is, FIG. 7 shows a state where the image of FIG. 6 is subdivided, and is subdivided into 30 (vertical) × 30 (horizontal) (= 900) areas (blocks), and one area (block) is 10 pixels × 10 pixels. In this embodiment, processing is performed with a square area of 10 pixels × 10 pixels. However, if this area has a size that allows confirmation of one to four ridges, the subdivision method and area shape are as follows. It is not limited.
[0021]
FIG. 8 shows as an example the contents of the region in the sixth column and the 18th row in the subdivided region of FIG. Since the image has already been binarized, FIG. 8 represents the background color and valley line as 0 (white) and data such as fingerprints or noise as 1 (black).
[0022]
In order to determine whether or not a ridge exists in a block, the present invention employs a method of determining pixel continuity by providing determination lines in the vertical and horizontal directions in the block. Therefore, in this embodiment, since one block pixel is 10 pixels × 10 pixels, the determination lines in the block are 20 lines vertically and horizontally. However, as described above, the ridge interval between human fingerprints is 1 to 4 in 0.5 mm, so that there are 4 ridges in the block (0.5 mm square 100 pixels) described in this embodiment. If present, the ridge width per line is 2.5 pixels. Therefore, if black (pixels representing ridges) is continuous for 3 pixels or more in the determination line and white (pixels indicating valleys) is also continuous for 3 pixels or more, the fingerprint (ridges) is displayed on the determination line. And valley line).
[0023]
FIG. 9 is a diagram illustrating determination line positions according to the embodiment. As described above, since the continuity is confirmed to be 3 pixels, it is possible to evaluate sufficiently if it is judged every 2 lines, so 3 judgments (1) to (3) in the horizontal direction per block. The line and the three judgment lines (4) to (6) in the vertical direction are combined to make 6 lines. By limiting the determination lines in this way, the processing speed can be improved. In this case, if the continuity of the ridges can be confirmed in 3 or more lines in the 6 determination lines, the ridge is determined to be present. In FIG. 9, the continuity (“1” or “0”) of the three pixels of the ridge and valley lines is added to the total of four lines of the judgment line (2), judgment line (4), judgment line (5), and judgment line (6). It can be confirmed that there are ridges in this block. The presence of fingerprints (ridges and valleys) can be confirmed in the image shown in FIG. 9 (6th column, 18th row).
[0024]
FIG. 10 and FIG. 11 are determination processing flows (part 1) and (part 2) for each block. Processing is performed for each block using the determination line shown in FIG. 9, and data storage areas indicated by (1) to (5) are provided corresponding to each block as shown on the right side of FIG. . In FIG. 10, the first block is selected as a processing target (S1 in FIG. 10), and the block number (No) is stored in (1) of the corresponding area (S2). Next, it is determined whether this block is suitable for determination processing (S3 in FIG. 10). This determination is made by identifying whether the block contains only background pixels (when the background is white, all pixels are white) or by identifying whether the block contains more than 3 ridges. Do. It is assumed that such a block is not suitable for determination processing, and a set indicating that it is not a processing target is set in (2) of the data storage area (same S4), and the next block is set as a processing target (S5 in FIG. 10). , S2 is returned. If it is determined in S3 that the block is suitable for the determination process, (2) is set to indicate that (S6 in FIG. 10), i = 0, k = 1 (where i is the determination line) Represents the count value, and k represents the number of lines in which the presence of ridges and valleys is confirmed (S7). Next, it is determined whether i has exceeded 7 (S8). If it has exceeded, the process proceeds to S13 in FIG. 11. If it has not exceeded, the i-th determination line has three or more pixels and three or more pixels. It is determined whether or not black is present (S9).
[0025]
Here, it is determined whether or not the presence of three or more pixels is confirmed (S10 in FIG. 10). If it is confirmed, k is updated (+1) (S11), and then i is updated (+1). (S12). If it cannot be confirmed in S10, the process proceeds to S12, i is updated, and the process returns to S8. In this way, S9 to S12 are repeatedly executed until i becomes 1 to 6, and when i becomes 7, when the process proceeds to S13 in FIG. 11, it is determined whether k ≧ 3, and is not applicable (white or more of 3 pixels or more). If the number of black lines is 2 or less), it is determined that the block to be processed does not have a fingerprint (S14), and the fact that no fingerprint exists in (3) of the data storage area corresponding to this block is set (S15). ). If it is determined as YES in step S13, it is determined that there is a fingerprint in the block to be processed (S16), and a state indicating that a fingerprint exists in (3) of the data storage area corresponding to this block is set. (S17), the coordinates of the block (with respect to the entire screen) are set in (5) of the data storage area (S18), and then the state of the block (wet state, dry state, or wet or dry) Any of the above states is set in the data storage area (4) (S19). The state of the block is determined by detecting the ratio of black in all the pixels in the block, the ratio of white, and the ratio of black and white, and determining whether the ratio exceeds a preset ratio.
[0026]
Subsequently, it is determined whether or not the processing of all blocks has been completed (S20 in FIG. 11). If the processing is not completed, the next block is set as a processing target (S21 in FIG. 11), and the process returns to S2 in FIG. Repeated.
[0027]
Even if the determination lines are not as shown in FIG. 9, the determination positions are not particularly defined as long as the determination lines can be determined every two lines.
[0028]
The states of the blocks divided in this way are individually determined. With respect to the block to be determined, it may be performed on all the divided 900 blocks, but it is useless to process a block (all pixels are background color) in which no data exists as shown in FIG. For this reason, it is possible to improve the processing speed by excluding such an image from being processed in advance because no ridge exists.
[0029]
In addition, since the method of determining the presence of a fingerprint based on pixel continuity is adopted, a block in which data (black) of 3 pixels or more does not exist in a block may be excluded from a processing target as a ridge does not exist in advance. I can do it. In this way, it is possible to further improve the processing speed by excluding unnecessary processing blocks from processing targets and defining them in advance as blocks having no ridges.
[0030]
The block to be processed is the state of the block to be processed in three stages: wet, dry, dry or not wet, depending on the pixel white / black ratio and continuity status, along with the fingerprint presence confirmation process by connecting pixels Recognize Specifically, for example, regardless of the presence or absence of continuity, if the ratio of black is about 80%, it is determined that there is a tendency to get wet. Conversely, if the ratio of white is about 80%, the tendency to dry It is determined that If neither of these is present, it is determined that neither dry nor wet. Needless to say, the severity of judgment can be freely set by slightly changing the ratio for judging the state.
[0031]
Also, as absolute coordinates for the input image, regardless of the presence of the ridges of the block to be processed, if the upper, lower, left, and right coordinates of the block in which data exists are retained by the process of S18 in FIG. When the determination is completed, input the upper coordinate of the block located at the top, the left coordinate of the block located at the left, the right coordinate of the block located at the right, and the lower coordinate of the block located at the bottom The fingerprint coordinates and input fingerprint area can be recognized.
[0032]
Next, a method for determining whether or not the input fingerprint image is suitable as a fingerprint image for fingerprint authentication based on the information of 900 blocks thus determined will be described.
[0033]
In the present embodiment, it is assumed that a fingerprint verification function that can perform fingerprint authentication without any problem if a ridge and a valley exist in about 80% of the whole is incorporated in the system. Therefore, if the presence of fingerprints (ridges and valleys) in 80% of 900 blocks, that is, 720 blocks or more, can be confirmed, an input image is taken as an image suitable for fingerprint authentication and the fingerprint collation function is shifted to.
[0034]
However, elements that are determined to be suitable for fingerprint authentication must be defined based on the performance of the fingerprint verification function that actually performs fingerprint verification.
[0035]
Depending on the type of fingerprint sensor, there are white and black background colors of the image to be collected. There are also various input image sizes. However, even in such an image, it is possible to perform the same processing based on this embodiment by changing the size of the image or changing the background color recognition method.
[0036]
Next, processing until the state of the input image is determined when it is determined that the input fingerprint is not suitable for fingerprint authentication will be described.
[0037]
FIG. 12 shows a flowchart up to the determination of a thin finger, which is executed by the thin finger determination unit 6a of FIG. 3 and shows details of S8 and S9 of FIG.
[0038]
In this case, the area ratio of the input fingerprint (block including ridges and valleys) to the entire image is set as a variable n (S1 in FIG. 12). Next, it is determined whether n> 60% (S2 in FIG. 12). If it does not reach 60%, the process proceeds to a wet or dry fingerprint determination process. If it is 60% or more, the number of input fingerprint blocks and the number of ridge confirmation blocks Is set as a variable m (S3), and it is determined whether m exceeds 70% (S4). If not, the process proceeds to a wet or dry fingerprint determination process. It is determined that the finger is thin (or the area of the input fingerprint is small) (S5), and the process ends.
[0039]
In the process of FIG. 12, if the ratio of the processing target block is large with respect to the input image area, there is no problem in the input area, and the process proceeds to a wet / dry determination process. If the ratio of the processing target block to the input image area is small, it is considered that a fingerprint with a thin finger has been input. The input fingerprint area described above may be used as the area of the processing target block as the determination element. However, since the input image area is small and the finger is thin, the block to be processed is extremely wet or extremely dry. Therefore, the ratio of the blocks that are judged to have ridges to the block to be processed is If it is smaller, the final judgment shifts to wet finger judgment and dry finger judgment. On the other hand, if the ratio of blocks determined to have ridges to the processing target block is large, it can be determined that the finger is definitely thin. As described above, when the input fingerprint area is small with respect to the input image area and the existence ratio of the ridges of the processing target block is high, it is determined that the finger is thin. When it is determined that the finger is thin, there are a case where the finger is lightly placed on the fingerprint input unit and only a part of the fingerprint is input, or a case where the finger is actually thin. In the former case, there is a possibility that it will be adopted as a normal one by inputting the fingerprint again, and in the latter case, it can be dealt with by changing to use another thick finger for authentication.
[0040]
In FIG. 12, the ratio of the processing target block to the input image area is set to 60%, and the ratio of the block determined to have a ridge to the processing target block is set to 70%. For this ratio, the fingerprint is actually authenticated. It goes without saying that some settings can be changed depending on factors such as engine performance.
[0041]
FIG. 13 is a flowchart until it is determined that the fingerprint is a dry fingerprint. In FIG. 13, it is considered that the determination of the thin finger shown in FIG. 12 is completed before this processing. First, for all the blocks to be processed, the ratio of dry (white (valley or background) in the above example is about 80% distribution) and wet (black (ridge) in the above example) is 80. The number determined to be neither dry nor wet is counted and set as i, j, k (S1 in FIG. 13). It is determined whether the ratio of blocks determined to be dry in the determination processing target block exceeds a predetermined threshold (60%) (S2 in FIG. 13), and if it exceeds, it is determined as a dry fingerprint (S4). Further, even when the ratio of blocks determined to be dry does not exceed a predetermined threshold (60%), it is determined whether the ratio of i, j, k is 50%, 30%, or 30% (FIG. 13). S3), if applicable, it is determined as a dry fingerprint, and if not, the process proceeds to a wet fingerprint determination process.
[0042]
In FIG. 13, the threshold value for determining a dry fingerprint is 60%, and the block ratio determination ratio is 50% or more for dryness: 30% or less for wetness: 30% or less for non-dry or wetness. It goes without saying that some settings can be changed depending on factors such as engine performance that actually authenticates fingerprints.
[0043]
FIG. 14 shows a flow chart for determining a wet fingerprint. In FIG. 14, it is assumed that the thin finger determination process shown in FIG. 12 and the dry fingerprint determination process shown in FIG. 13 are completed before this process.
[0044]
First, the ratio of blocks to be processed and the blocks determined to have ridges, and the number of blocks determined to be dry, wet, and neither dry nor wet for all the blocks to be processed are counted as i, j , K (S1 in FIG. 14). In this process, the value calculated by the dry fingerprint determination can be used as it is. It is determined whether the ratio of the block determined to be wet with respect to the processing target block exceeds a predetermined threshold (60%) (S2 in FIG. 14), and if it exceeds, it is determined as a wet fingerprint (S4). Then, even when the ratio of the blocks determined to be wet does not exceed the predetermined threshold, it is determined whether the ratio of i, j, k is 30% or less, 50% or more, and 30% or less (S3 in FIG. 14). ). That is, if the ratio of the block determined to be wet is clearly larger than the block determined to be wet and the block determined to be neither dry nor wet, it is determined as a wet fingerprint and the process ends. If this condition is not met, it is determined that the finger is rough (S5 in FIG. 14), and the process ends.
[0045]
In FIG. 14, the threshold value for determining a wet fingerprint is 60%, and the block state determination ratio is 50% or more wet: 30% or less dry: 30% or less is neither dry nor wet. It goes without saying that the settings can be changed somewhat depending on factors such as the engine performance for authenticating fingerprints.
[0046]
If it is not possible to determine that the fingerprint is a wet fingerprint, the collected fingerprint may have a dry portion and a wet portion mixed at the same ratio. In this case as well, the fingerprint is not suitable for fingerprint authentication. However, such a determination is often made with a finger that is extremely rough, and thus it is determined that the finger is rough.
[0047]
In this embodiment, the processing order for determining the state of the image has been described as performing processing in the order of processing for determining a fingerprint with a thin finger, processing for determining a dry fingerprint, and processing for determining a wet fingerprint. It goes without saying that this determination order can be changed depending on how the determination processing is combined. For example, the determination process can be freely replaced, such as determining a dry fingerprint with an image that does not require wet fingerprint determination.
[0048]
That is, the presence of ridges in each block (subdivision area) into which the input image is divided, and the state of the image such as whether each block is wet, dry, wet or dry, If the information of the input fingerprint coordinate and the input fingerprint area from the coordinates of each block is determined regardless of the presence of ridges, the contents of the input image can be used to freely determine the state of the input image. It becomes possible. In addition, if the function that creates and compares fingerprint feature points tends to reduce the authentication rate if images near the center of the fingerprint cannot be collected accurately, the coordinates near the center are identified from the above input fingerprint coordinates. It is also possible to determine whether or not the fingerprint in the coordinates is wet or dry and determine whether the fingerprint is unsuitable for authentication. As described above, according to the characteristics of the fingerprint collation function to be used, it is possible to increase the degree of freedom of whether the fingerprint is an abnormal fingerprint or a normal fingerprint as a result of partial state determination.
[0049]
(Appendix 1) In an input fingerprint state determination method for determining whether an input fingerprint image is suitable for authentication, the input fingerprint image is subdivided into regions of a predetermined size, and ridges and valleys are formed in each subdivided region. It is determined whether there is a fingerprint consisting of, and the skin surface state in the subdivided area is determined, and whether the fingerprint image is suitable for fingerprint authentication from the fingerprint presence determination result and the state determination result for each area And determining an input fingerprint state.
[0050]
(Supplementary note 2) A method for determining an input fingerprint state according to supplementary note 1, wherein a size to be subdivided into a size that can identify at least one ridge or valley line is selected.
[0051]
(Supplementary Note 3) In Supplementary Note 1, for each of the subdivided images of each region, for each pixel arranged vertically and horizontally, a determination is made by separating at least the width of the ridge or valley in the vertical and horizontal directions. A method for determining an input fingerprint state, characterized in that it is determined whether or not a fingerprint exists depending on whether or not a predetermined number of pixels representing ridges or valleys continue by line.
[0052]
(Supplementary Note 4) In Supplementary Note 1, for each image of each of the subdivided regions, the ratio of pixels representing ridges in the image and the ratio of pixels representing valleys or backgrounds are obtained to represent the ridges. When the pixel ratio is greater than or equal to a predetermined value, the wet state is set, and when the ratio of pixels representing valleys or backgrounds is greater than or equal to the predetermined value, the dry state is set. A method for determining an input fingerprint state, wherein a ratio of regions that are neither wet nor dry is obtained, and a fingerprint state is determined by determining a ratio of each of these states.
[0053]
(Additional remark 5) The additional fingerprint state determination method in Additional remark 4 characterized by determining each state whether the fingerprint is thin, dry, wet, or rough.
[0054]
(Additional remark 6) In the input fingerprint state determination apparatus which determines whether an input fingerprint image is suitable for authentication, a subdivision part that subdivides the input fingerprint image into regions of a predetermined size, and a ridge in each subdivision region A fingerprint presence determination unit that determines whether or not a fingerprint composed of a line and a valley line exists, a state determination unit that determines a skin surface state in a subdivided region, and the fingerprint presence determination unit for each region An input fingerprint state determination device comprising: a fingerprint image state determination unit that determines whether the fingerprint image is suitable for fingerprint authentication based on the determination result and the determination result of the state determination unit.
[0055]
(Additional remark 7) In additional remark 6, the said fingerprint presence determination part is the width | variety of at least a ridge or a valley line in the vertical direction and a horizontal direction about each pixel arrange | positioned vertically and horizontally for every image of each subdivided area | region. An input fingerprint state determination apparatus, wherein a determination is made as to whether or not a fingerprint exists based on whether or not a predetermined number of pixels representing ridges or valleys are continued by determination lines spaced at intervals.
[0056]
(Additional remark 8) In additional remark 6, the said fingerprint image state determination part is a ratio in which the pixel showing a ridge in the image exists, and the ratio in which the pixel showing a trough line or a background exists for every image of each subdivided area | region. When the ratio of pixels representing ridges is greater than or equal to a predetermined value, the wet state is set, and when the ratio of pixels representing valleys or backgrounds is greater than or equal to the predetermined value, the wet state is set. An input fingerprint state determination apparatus characterized by determining a ratio, a ratio of a dry state, and a ratio of a region that is neither wet nor dry, and determining each of these ratios to determine a fingerprint state.
[0057]
(Supplementary note 9) In the supplementary note 6, when the fingerprint image state determination unit determines that the fingerprint image state is not suitable for fingerprint authentication, the reason why the fingerprint is thin, dry state, wet state, etc. is output. Judgment device.
[0058]
【The invention's effect】
According to the present invention, an image that is not suitable for fingerprint authentication is confirmed at high speed without performing processing that requires a large calculation time such as thinning the image input to the fingerprint image input unit. In this case, it is possible to determine the reason for determining that the image is not suitable, such as dry, wet, rough finger, and thin finger, and to provide a function of rejecting the image capture.
[0059]
In addition, in a personal authentication system equipped with a fingerprint information input unit that has the function of collecting fingerprint information as an image, the collected image is suitable for fingerprint authentication as preprocessing for extracting fingerprint feature data from the input image. If it is determined that it is not suitable for authentication on the spot, it is possible to capture a fingerprint image suitable for fingerprint authentication by determining the cause of the determination that it is not suitable It becomes. For users who are prone to fingerprint authentication failure (users who tend to input wet or dry fingerprints), feedback necessary measures to improve the situation based on the judgment results are fed back to the user. Then, it is possible to perform fingerprint collation with a stable authentication rate without stressing the user.
[Brief description of the drawings]
FIG. 1 is a diagram showing a principle configuration of the present invention.
FIG. 2 is a diagram showing a processing flow according to the present invention.
FIG. 3 is a diagram illustrating a configuration of an example.
FIG. 4 is a diagram illustrating a processing flow of an embodiment.
FIG. 5 is a diagram illustrating an example of an input fingerprint image.
FIG. 6 is a diagram illustrating an example of a binarized image.
7 is a diagram showing a state in which the image of FIG. 6 is subdivided. FIG.
8 is a diagram showing the contents of a region in the sixth column and the 18th row in the subdivided region in FIG. 7; FIG.
FIG. 9 is a diagram illustrating determination line positions according to an embodiment.
FIG. 10 is a diagram showing a determination processing flow (No. 1) for each block;
FIG. 11 is a diagram showing a determination processing flow (No. 2) for each block;
FIG. 12 is a diagram illustrating a flowchart for determining a thin finger.
FIG. 13 is a flowchart illustrating determination as a dry fingerprint.
FIG. 14 is a flowchart illustrating determination as a wet fingerprint.
FIG. 15 is a diagram showing a configuration of a conventional example.
FIG. 16 is a diagram showing an example of a fingerprint image used for explaining a conventional example.
17 is a diagram illustrating a state where the image of FIG. 16 is subdivided.
FIG. 18 is a diagram illustrating pixel distribution of two blocks.
[Explanation of symbols]
1 Fingerprint information input section
2 Image segmentation
3 Subdivision image ridge presence judgment part
4 Subdivision image state determination unit
5 Image import decision section
6 Image state determination unit

Claims (2)

In an input fingerprint state determination method for determining whether an input fingerprint image is suitable for authentication,
The input fingerprint image is subdivided into regions of a predetermined size , and the width of at least the ridges or valleys in the vertical and horizontal directions for each pixel arranged vertically and horizontally for each subdivided region image. A judgment line with a certain interval is used to judge whether there is a fingerprint depending on whether a predetermined number of pixels representing ridges or valleys continue , and to determine the state of the skin surface in the subdivided area And determining whether or not the fingerprint image is suitable for fingerprint authentication from the fingerprint presence determination result and the state determination result for each region. In an input fingerprint state determination device for determining whether an input fingerprint image is suitable for authentication,
A subdivision part for subdividing the input fingerprint image into a predetermined size area;
For each image in each segmented area, for each pixel arranged vertically and horizontally, a ridge or valley is defined by a judgment line spaced at least by the width of the ridge or valley in the vertical and horizontal directions. A subdivided image fingerprint presence determination unit that determines whether or not a fingerprint exists depending on whether or not a predetermined number of pixels to be represented are continuous;
A subdivided image state determination unit for determining the state of the skin surface in the subdivided region;
A fingerprint image state determination unit that determines whether the fingerprint image is suitable for fingerprint authentication based on a determination result of the subdivision image fingerprint presence determination unit and a determination result of the subdivision image state determination unit for each region;
An input fingerprint state determination apparatus comprising:

Images