KR20060131141A - Apparatus and method of fingerprint recognition - Google Patents

Abstract

A device and a method for recognizing a fingerprint is provided to perform precise fingerprint recognition based on composition of the partial fingerprint without respect to a non-uniform sliding speed when the fingerprint is slid. A calculator calculates sliding information among partial fingerprint images inputted from a fingerprint sensor(10). A storing part(40) stores the partial fingerprint images and the corresponding sliding information. A corrector(50) corrects the sliding information of each partial fingerprint image by using the stored partial fingerprint images and sliding information. A composer(60) composes one fingerprint image from the stored partial fingerprint images by using the corrected sliding information. A fingerprint recognizer(70) performs the fingerprint recognition based on the composed fingerprint image. The sliding information is a relative speed and direction information between the continuously neighboring partial fingerprint images.

Description

Fingerprint recognition device and method {APPARATUS AND METHOD OF FINGERPRINT RECOGNITION}

1 is a view showing an example of a slide type fingerprint sensor exterior structure;

Fig. 2 shows an example of a local fingerprint image obtained by a slide type fingerprint sensor.

Figure 3 is a block diagram of a fingerprint recognition device according to the present invention

4 illustrates an example of a local fingerprint image according to the present invention.

5 is a flowchart of a fingerprint recognition method according to the present invention.

<Explanation of symbols for the main parts of the drawings>

10: fingerprint sensor 20: fingerprint image processing unit

30: sliding information extraction unit 40: memory unit

50: correction unit 60: matching unit

70: fingerprint recognition unit

The present invention relates to a fingerprint recognition system.

In the security and security system technology, a method of using biometric information is used as a means for authenticating legitimacy of a person entering a building or a user of a specific device. Iris information or fingerprint information is used as biometric information for authentication, and various biometric authentication methods are used, ranging from authentication methods based on iris recognition to authentication methods based on fingerprint recognition. The biometric information-based authentication system is a method for not only authentication of the user of the building but also information protection or personal identification, and is used instead of the existing password method or ID card method.

In a system for authentication based on fingerprint recognition, a means for acquiring fingerprint information, a means for extracting a feature from the acquired fingerprint information, and a feature for analyzing the extracted feature and comparing the registered fingerprint information with the pre-registered fingerprint information are authenticated. Means;

Such a fingerprint recognition system is applied to a mobile terminal such as a PDA or a mobile phone because it has advantages such as mechanical simplicity of a sensor for authentication and high recognition accuracy of an authentication algorithm.

A fingerprint recognition sensor is used as a means for obtaining fingerprint information. Fingerprint sensors are classified into optical, thermal and semiconductor types, and mobile terminals are mainly equipped with semiconductor sensors due to space constraints. Semiconductor sensors can be classified into area type and slide type. However, since the space fingerprint recognition sensor has a large space occupied by the mobile terminal on which the sensor is mounted, the slide fingerprint recognition sensor having a relatively small occupied space is mainly mounted on the mobile terminal to exhibit an effect.

The slide-type fingerprint sensor reads local fingerprint images by sliding a finger, and then registers the fingerprint image based on obtaining an intact fingerprint image by composing the local fingerprint image into one image. Perform.

Conventional partial fingerprint image matching methods sequentially superimpose the partial fingerprint image input from the fingerprint sensor onto the matched fingerprint image and update the matched fingerprint image by matching relative brightness values of the overlapped regions. We are taking a sequential composition.

In this sequential matching method, the sampling rate of the slide type sensor is faster than the sliding speed, and the gain control of the sensor is ideal, so that the brightness values of the pixels corresponding to one-to-one corresponding to the overlapping fingerprint area are the same, If the geometric distortion of the fingerprint is in a negligible range, accurate fingerprint images can be matched.

However, the sliding operation relative to the actual slide sensor may have various forms according to a user's habit. For example, when performing a sliding action with a finger on the fingerprint sensor, in most cases the sliding speed is not uniform and variable, so that the sliding speed at the end of the sliding step tends to be considerably faster than the sliding speed at the start step. do.

Extremely, the sliding speed may be partially faster than the sampling rate of the sensor, and in this case, there is no overlap between the newly input partial fingerprint image and the matched fingerprint image, so that brightness value matching is fundamentally possible. There will be no.

However, the fingerprint sensor cannot read it, resulting in a false matching result, which appears as a discontinuous flow of fingerprint ridges in the matched fingerprint image.

Therefore, in the conventional partial fingerprint image matching method, in order to solve the above problems, it is determined whether to accept the matching result through a relative comparison between the threshold value and the matching error specialized in the fingerprint sensor. Depending on the amount of pressure exerted on the fingerprint sensor during sliding and the characteristics of the fingerprint (eg wet and dry), the matching error is still significantly affected.

As a result, the sequential fingerprint matching method has a fundamental problem in image matching when the sampling rate is not significantly faster than the sliding speed.

In addition, in the case of matching using the brightness value of the image, the possibility of false matching is always inherent, and detection thereof will generally use a threshold value specialized for the fingerprint sensor as described above. However, when the possibility of mismatching is detected, there is no method other than predicting the sliding speed of the previous sliding step by using a sequential fingerprint matching method, and there is a change in sliding speed at the time of mismatching. In this case, this statistical prediction method is inevitable in matching errors.

SUMMARY OF THE INVENTION An object of the present invention is to recognize a fingerprint based on the composition of a partial fingerprint image, and to perform accurate fingerprint recognition in spite of non-uniform sliding speed when sliding for fingerprint recognition. An apparatus and a method thereof are provided.

Another object of the present invention is a registration error in a system for recognizing a fingerprint based on the composition of a partial fingerprint image, a sliding error due to sliding faster than the sampling rate of the fingerprint sensor which is influenced by the sliding habit of the user in the sequential matching method, The present invention provides a fingerprint recognition device and a method for solving a mismatch problem caused by an algorithmic matching error and performing accurate fingerprint recognition based on an active and adaptive error correction type out of order matching method.

It is still another object of the present invention to calculate a relative sliding speed between adjacent neighboring partial fingerprint images in a system for recognizing a fingerprint based on the composition of partial fingerprint images, when sliding for fingerprint recognition, By correcting the velocity information of the speeding section or mismatching section by using the statistical characteristics of, it is possible to perform accurate fingerprint recognition by reconstructing partial fingerprint images into matching images using the sliding information supplemented from the final matching start point. A fingerprint recognition device and a method thereof are provided.

The fingerprint recognition apparatus of the present invention includes a fingerprint image obtaining means, means for calculating sliding information between partial fingerprint images input from a fingerprint image obtaining means, means for storing the partial fingerprint images and sliding information, and the stored partial fingerprint. Means for correcting sliding information of each partial fingerprint image by using the image and its sliding information; means for matching partial fingerprint images stored in the storage means to a single fingerprint image by using the corrected sliding information; And means for performing fingerprint recognition based on the image.

The fingerprint recognition method of the present invention may include obtaining a partial fingerprint image, obtaining relative sliding information between the obtained partial fingerprint images, correcting relative sliding information between the partial fingerprint images, and correcting the sliding information. And matching the partial fingerprint images to reflect the reflection.

EXAMPLE

Figure 1 shows an example of the slide structure of the fingerprint sensor appearance. Here, the structure in which the fingerprint sensor 102 of the slide type is mounted on the mobile terminal 101 is shown. The fingerprint sensor 102 acquires a partial and continuous partial fingerprint image by sliding a user's finger.

2 shows an example of a local fingerprint image obtained by a slide type fingerprint sensor. When the user slides the thumb 103 against the fingerprint sensor 102 at a predetermined speed, partial and continuous fingerprint images 104 to 107 are obtained.

As shown in FIG. 2, the fingerprint image of the user is obtained by being sliced into partial, adjacent, and continuous forms. The partial fingerprint images read from the fingerprint sensor 102 are matched with one fingerprint image, and the intact fingerprint image matched with one fingerprint image is compared with a registered fingerprint image (which may be a database). It will be used as a means of justification.

The present invention will provide a means for accurately matching and recognizing sliding partial fingerprint images based on the sequential matching scheme irrespective of the change in sliding speed.

3 shows a configuration of a fingerprint recognition device according to the present invention. The fingerprint recognition device shown in Fig. 3 can be mounted in the mobile terminal. The fingerprint recognition apparatus according to the present invention includes a fingerprint sensor 10 for acquiring a fingerprint image, a fingerprint image processor 20 for processing partial fingerprint images of a user read by the fingerprint sensor 10, and the fingerprint image processor ( A sliding information extracting unit 30 for extracting relative sliding information between partial fingerprint images from 20, a memory unit 40 storing the partial fingerprint images and respective sliding information, and the memory unit 40. A corrector 50 for correcting a fingerprint image using the stored partial fingerprint images and sliding information, a matching unit 60 constituting a single fingerprint image using the corrected fingerprint images, and a match The fingerprint recognition unit 70 performs recognition of the fingerprint image using the fingerprint image.

The fingerprint sensor 10 is a slide type semiconductor fingerprint sensor. The slide type semiconductor fingerprint sensor scans a fingerprint and reads the fingerprint in the form of a partial fingerprint image when the user slides the finger on the recognition window of the sensor at a predetermined speed.

The partial fingerprint image read from the fingerprint sensor 10 is input to the fingerprint image processor 20. The fingerprint image processing unit 20 performs sampling driving of the fingerprint sensor 10, and converts the partial fingerprint images read into digital signals to provide a basis for performing pixel-based fingerprint recognition. In addition, as will be described later, the fingerprint image processing unit 20 applies the same gain to all areas of the image so that the average brightness is the same before the search for the sliding speed and the direction of the reference image and the target image. It also performs the process of adjusting.

The sliding information extractor 30 extracts the speed and the direction as sliding information of the partial fingerprint images. That is, relative speed information and direction information between two adjacent partial fingerprint images are extracted.

The memory unit 40 stores sliding fingerprints for partial fingerprint images and respective partial fingerprint images that are sequentially input.

The correction unit 50 corrects the relative sliding information between the partial fingerprint images stored in the memory unit 40 to overcome the problem of occurrence of a matching error due to an uneven sliding speed or a problem between the sliding speed and the fingerprint sensor sampling speed. Do it.

The matching unit 60 matches each partial fingerprint image in which relative sliding information is corrected into one fingerprint image, and the fingerprint recognition unit 70 recognizes and authenticates the corresponding user fingerprint using the matched final fingerprint image. Processing will be performed.

4 shows an example of a local fingerprint image according to the present invention. Two partial fingerprint images are represented in FIG. 4, which are adjacent and consecutive partial fingerprint images. The first partial fingerprint image 110 is a current partial fingerprint image input from the fingerprint sensor 10, and the second partial fingerprint image 120 is a previous partial fingerprint image input from the fingerprint sensor 10. Here, the current partial fingerprint image and the previous partial fingerprint image mean one of the following two cases.

First, when real-time correction and registration is performed, the partial fingerprint image currently input is the current partial fingerprint image, and the partial fingerprint image already input at the previous time point corresponds to the previous partial fingerprint image.

Second, when correction and registration are performed collectively after completion of the input of all partial fingerprint images, if one partial fingerprint image selected from several partial fingerprint images is the current partial fingerprint image, immediately before the selected partial fingerprint image input This is a case where the partial fingerprint image inputted at the point of time is a previous partial fingerprint image.

In either case, the matching and recognition method based on the relative sliding information correction of the partial fingerprint image, which will be described later, will be applied in the same manner.

When the first partial fingerprint image 110 is used as a reference image, the second partial fingerprint image 120 becomes a target image. In contrast, when the second partial fingerprint image 120 is used as the reference image, the first partial fingerprint image 110 becomes a target image. In the present invention, in order to extract and correct relative sliding information between partial fingerprint images, a portion of the target image 120 is a target-search slice 121 and a portion of the region of the reference image 110 is a reference slice 111. Set each to. The detailed method using the target image-search slice, the reference image-reference slice will be described later.

5 is a flowchart of a fingerprint recognition method according to the present invention. A first step S10 is a step of acquiring a partial fingerprint image, and a second step S20 is a step of determining whether an input partial fingerprint image is an initial input. The third step S30 is a step of extracting sliding information relative to the previously inputted partial fingerprint image when the inputted partial fingerprint image is the second and subsequent partial fingerprint images. The fourth step S40 is to store the extracted partial fingerprint image and sliding information in the memory unit. The fifth step S50 is to continuously determine whether there is a fingerprint sensor input. If there is no further fingerprint sensor input in this step, the process moves to the sixth step S60 to correct sliding information between the partial fingerprint images. Do this. The seventh step S70 is a fingerprint image registration and recognition step in which each gag configures partial fingerprint images into one intact fingerprint image by reflecting the corrected sliding information.

The fingerprint recognition method represented in FIG. 5 will be described in more detail with reference to the accompanying drawings.

As the sliding operation of the user starts and the first partial fingerprint image input is detected from the fingerprint sensor 10, the fingerprint recognition system according to the present invention operates.

The partial fingerprint image is acquired by sensing an input from the fingerprint sensor 10 and storing the partial fingerprint image input from the fingerprint sensor 10 in a predetermined region of the predetermined memory unit 40 when sliding is detected in the sensor.

Since the sliding information cannot be extracted when the first partial fingerprint image is input, it waits for the next partial fingerprint image input.

If there is more than one input partial fingerprint image, sliding information extraction is performed.

That is, the partial fingerprint image T (N) currently input and the partial fingerprint image T (N-1) input in the previous step stored in the memory are read to extract relative sliding directions and speed information between the two partial fingerprint images. The sliding direction and the speed are extracted by comparing brightness values of overlapping images between two adjacent partial fingerprint images.

To this end, one image is used as a reference image and the other image is used as a target image, and a point where the difference in brightness values of overlapping pixels between the reference image and the target image is minimized is found. Here, the reference image and the target image are continuous partial fingerprint images input from the fingerprint sensor 10 and have the same image size. In order to efficiently perform the minimum point search of the brightness values between the adjacent partial fingerprint images, some specific regions of the target image are set as the search slices, and some specific regions of the reference image are set as the reference slices.

For example, in FIG. 4, the specific image region of the first partial fingerprint image 110 is designated as the reference slice 111, the specific image region of the second partial fingerprint image 120 is designated as the search slice 121, and While moving the two-part fingerprint image 120 in the long axis (X) and short axis (Y) directions of the fingerprint sensor 10, the brightness value of each pixel overlapping with the reference slice 111 is obtained, and the difference is minimized. Search for a point.

Taking a partial fingerprint image on the memory unit 40 shown in FIG. 5 as an example, T (N-1) and T (N) are read, and T (N) is in the long axis (X) and short axis (Y) directions. Moving toward, it searches for a point where the difference in brightness value of each pixel overlapping the reference slice of T (N-1) becomes the minimum. That is, when the target image is superimposed on the reference image, the point at which the difference in brightness between the pixels of the target image superimposed on the reference slice image and each pixel of the base slice image is minimized is searched for.

Next, when the difference between the brightness values of the overlapped areas becomes minimum, the speed and direction of sliding are calculated by calculating the relative degree of pixel movement between the reference slice coordinate system and the target slice coordinate system. That is, the two partial fingerprint images are calculated by calculating the relative degree of movement in the long axis (X) and short axis (Y) directions of the target image when the difference between the brightness values of the overlap regions is minimum based on the reference slice image. The relative speed and direction of the liver can be obtained.

As such, two adjacent partial fingerprint images may have a common overlapping area, or may not have an overlapping area when the sliding speed of the user is faster than the sampling rate of the fingerprint sensor.

In addition, the position and size of the reference slice image 111 may be variable or fixed. The position and size of the reference slice image 111 are fixed while the sliding direction and the speed are obtained, and the sliding of the partial fingerprint image obtained in the previous step is performed. When the new partial fingerprint image is input from the fingerprint sensor 10 according to the speed and the direction information, its position and size are variably adjusted. This is to more efficiently search for new partial fingerprint images and extract sliding information efficiently. In addition, prior to the search for the sliding speed and the direction, by adjusting the brightness value by applying the same gain over the entire area of the image such that the average brightness value of the reference image and the target image are the same, the image is adjusted. It increases the efficiency and accuracy of the search.

Next, the calculated sliding speed, the direction information, and the brightness difference information at that time are stored in the memory unit 40 in synchronization with the input partial fingerprint image.

Repeat the above process until the last partial fingerprint image is inputted, and all partial fingerprint images T (1), T (2), ..., T (N-1), T (N) and their The speed and direction, which are relative sliding information between the partial fingerprint images and the adjacent fingerprint images, are extracted and stored in the memory unit 40.

When the sliding operation of the user is finished, the sliding information correcting unit 50 checks whether the partial fingerprint image is misaligned using the sliding information of the fingerprint images stored in the memory unit 40.

In the present invention, two adjacent partial fingerprint images are not matched immediately after the relative sliding direction and speed are calculated, and the sliding direction and velocity information misaligned after calculating the sliding direction and speed information between all adjacent partial fingerprint images. The correction is then performed, and after that, it is matched collectively using the corrected sliding direction and speed information. Here, the sliding direction and speed information between all consecutive adjacent partial fingerprint images starts from the partial fingerprint image input from the fingerprint sensor 10 at the start of sliding of the user, and the fingerprint sensor 10 at the end of sliding of the user. Relative sliding directions and velocities obtained from combinations of partial fingerprint images sequentially generated from the partial fingerprint images to the input partial fingerprint images.

The determination of misalignment may include determining that a mismatch occurs when the relative sliding speed of adjacent partial fingerprint images has a discontinuity that is greater than the error range allowed by the fingerprint sensor 10, and the minimum brightness error at the time of mismatch is determined by the fingerprint. The sensor 10 determines whether the sensor 10 is within the allowable range, and when the error is greater than or equal to the error range, the sliding information of the slice (partial fingerprint image) is corrected using the sliding speed information of the surroundings.

In other words, detection of misalignment is performed by applying statistics on the sliding speed, direction continuity, and brightness value difference between the partial fingerprint image combinations obtained in the entire sliding. In other words, if the sliding speed and direction of adjacent partial fingerprint images are discontinuous, mismatch is determined. If the brightness value error at the time of mismatch is outside the allowable range, the sliding speed information of the partial fingerprint image is determined as the partial fingerprint image. This is corrected by using the sliding speed information of the surrounding partial fingerprint image.

When the correction for the misalignment is completed as described above, the fingerprint image matching unit 60 collectively matches the partial fingerprint images stored in the memory unit 40 into one fingerprint image using the corrected sliding speed and direction information. In this case, by using the sliding speed and the direction information of the adjacent partial fingerprint image, matching is limited to the fingerprint image region except for the region overlapping with the partial fingerprint image to be subsequently matched, thereby reducing unnecessary calculation time and forming the entire matched image. Can be.

When the matching is completed, the matched image is transmitted to the fingerprint recognition unit 70, and the fingerprint recognition unit 70 extracts the characteristic information of one intact fingerprint image, and based on the characteristic information, the authenticity of the user. Confirm and authenticate.

According to the present invention, in the fingerprint recognition using a slide type fingerprint sensor, the user's sliding information is generated as a single completed fingerprint image using only the input partial fingerprint image information, and at this time, a partial fingerprint image in which misalignment can occur. By effectively correcting, the shortcomings of the conventional mismatches using the sequential matching method can be effectively solved without increasing the computation time, thereby improving the partial fingerprint image matching performance.

The present invention can recognize partial fingerprint images input from a slide type fingerprint sensor with a minimum of matching error in a mobile terminal environment in which computing resources are limited, and a sliding speed according to a user's sliding habit when a sliding fingerprint sensor is mounted. Enables accurate fingerprint matching and recognition despite nonuniformity.

Claims (21)

By means of fingerprint image acquisition means, means for calculating sliding information between partial fingerprint images inputted from the fingerprint image acquisition means, means for storing the partial fingerprint images and their sliding information, and using the stored partial fingerprint image and its sliding information Means for correcting sliding information of each partial fingerprint image, means for matching partial fingerprint images stored in the storage means with one fingerprint image using the corrected sliding information, and means for performing fingerprint recognition based on the matched fingerprint image Fingerprint recognition device comprising a. The fingerprint recognition device according to claim 1, wherein the sliding information is relative speed and direction information between successive adjacent partial fingerprint images. The method of claim 1, wherein the sliding information calculates relative sliding information by comparing one of two adjacent partial fingerprint images as a reference image and the other as a target image by comparing the reference image and the target image. Fingerprint recognition device characterized in that for correcting the sliding information based on this. The target image of claim 1, wherein the correcting means comprises one of two consecutive partial fingerprint images as a reference image and a partial region of the reference image as a reference slice image, and the other as a target image. And taking a partial region of the image into a target slice image, calculating relative sliding information by comparing the reference slice image with the target slice image, and correcting the sliding information based on the reference slice image. The fingerprint recognition apparatus according to claim 1, wherein the acquiring means, the sliding information calculating means, the storing means, the correcting means, and the matching / recognition means are mounted on the mobile terminal. Fingerprint sensor for sequential and partial continuous acquisition of fingerprint images according to sliding, storage means for storing partial fingerprint images detected by the fingerprint sensor, sliding misaligned by calculating relative speed and direction between the stored partial fingerprint images Correction means for correcting direction and velocity information, matching means for collectively matching the partial fingerprint images into one fingerprint image using the corrected sliding direction and velocity information, and based on the matched fingerprint image Fingerprint recognition device comprising a recognition means for performing the user authentication. Obtaining a partial fingerprint image, obtaining relative sliding information between the obtained partial fingerprint images, correcting relative sliding information between the partial fingerprint images, and matching the partial fingerprint images by reflecting the corrected sliding information Fingerprint recognition method comprising the steps of. The fingerprint recognition method according to claim 7, wherein the sliding information is a relative speed and direction between partial fingerprint images. 8. The fingerprint recognition method according to claim 7, wherein the relative sliding information between the partial fingerprint images is obtained by comparing two adjacent partial fingerprint images. The fingerprint recognition method according to claim 7, wherein the sliding information is obtained for both the long axis direction and the short direction of the fingerprint sensor. The method of claim 7, wherein the relative sliding information between the partial fingerprint images is a comparison between the target image and the reference image using any one of a current partial fingerprint image and a previous partial fingerprint image as a target image and the other as a reference image. Fingerprint recognition method characterized in that obtained through. The method of claim 7, wherein the relative sliding information between the partial fingerprint images includes a search slice set in one of a current partial fingerprint image and a previous partial fingerprint image, and a reference slice set in the other one. Fingerprint recognition method characterized in that obtained through the comparison between slices. The fingerprint recognition method according to claim 7, wherein the relative sliding information between the partial fingerprint images is obtained by comparing brightness values of overlapping image regions between two adjacent partial fingerprint images. The method of claim 13, wherein the comparing of the brightness values comprises: setting one of two consecutive partial fingerprint images as a reference image and setting the other as a target image; Setting a reference slice image, superimposing the target image on the reference image while moving the target image in a long axis and a short axis direction, and in the overlapping, each pixel and the reference slice image of the target image one-to-one superimposed on the reference slice image And comparing the brightness difference between the pixels. 15. The method of claim 14, wherein the sliding direction and the speed are calculated by calculating the relative degree of movement in the long axis and short axis of the target image when the difference in the brightness value of the overlap region is minimized based on the reference slice image. Fingerprint recognition method characterized in that. The fingerprint recognition method according to claim 12 or 14, wherein the reference slice image has a fixed position and size in the reference image or a variable adjustment of the position and size. 12. The fingerprint of claim 11, wherein the reference image and the target image are adjusted by applying the same gain to all areas of the image so that the average brightness value is the same before the sliding speed and the direction search. Recognition method. The method of claim 7, wherein after the sliding information between all adjacent partial fingerprint images is calculated and the correction of the misaligned sliding direction and the speed information is performed, the batch is corrected using the corrected sliding direction and the speed information. Fingerprint recognition method characterized in that the matching of the partial fingerprint images. 19. The method of claim 18, wherein the direction and speed information as sliding information between successive adjacent partial fingerprint images are sequentially generated from the partial fingerprint image acquired at the sliding start point to the partial fingerprint image acquired at the sliding end point. Fingerprint recognition method characterized in that each is obtained from a combination of partial fingerprint images. 19. The fingerprint recognition method according to claim 18, wherein the misalignment detection is performed by applying statistics on the sliding speed, direction continuity, and brightness value difference between combinations of the partial fingerprint images. 19. The full registration image of claim 18, wherein each partial fingerprint image in the batch registration uses only sliding speeds and direction information between adjacent partial fingerprint images, except for an area except an overlapping portion with a partial fingerprint image to be matched later. Fingerprint recognition method characterized in that the matching.

Images