TW201945981A - Fingerprint template registration method based on sliding type sampling and fingerprint identification device - Google Patents

Abstract

The invention provides a fingerprint template registration method based on sliding type sampling and a fingerprint identification device. The fingerprint template registration method comprises a fingerprint template input process based on sliding type sampling, the process comprises the following screening process of n fingerprint images, and the screening process comprises an initial image selection process, a terminated image selection process and a fingerprint image selection process. According to the invention, the collected fingerprint images are screened; the method has the advantages of reducing coincident fingerprint images in the registered fingerprint image data processing process, eliminating incomplete pressing fingerprint images, screening the fingerprint images, reducing the data processing amount in the fingerprint image registration process, improving the data processing efficiency, being beneficial to realizing rapid input of a fingerprint template, and improving the user fingerprint input experience; a sliding and pressing combined sampling mode is adopted, the advantages of the sliding and pressing combined sampling mode are combined, the defects of the sliding and pressing combined sampling mode are overcome through the screening process, and user experience and the fingerprint recognition rate are better improved.

Description

Fingerprint template registration method and fingerprint recognition device based on sliding sampling

The present invention relates to a data processing method and device, and more particularly to a processing method and fingerprint identification device for obtaining fingerprint data.

In the implementation scheme of fingerprint recognition based on a small area sensor, the fingerprint entry methods used for fingerprint template registration include pressing entry and sliding entry.

The conventional press input method uses multiple presses to collect fingerprint data to be registered, and because of the difference in the position of each press, the collected fingerprint template is easy to form a deviation template, and multiple presses are easy to collect high coincidence. In addition, the push-type fingerprint entry step requires the user to perform a dozen or even dozens of pressing operations, so the user experience is poor.

The sliding input of the conventional technology is to collect a plurality of continuous fingerprint image data by the steps of sliding a finger across a fingerprint sensor, so a large amount of fingerprint information can be collected, which can improve the user experience. However, sliding entry still generates a large number of fingerprint images with high coincidence, which increases the difficulty of stitching fingerprint images in the fingerprint registration step, resulting in increased data processing time; and the large image area after stitching fingerprint images causes Fingerprint image recognition time also increased.

The technical problem to be solved by the present invention is to avoid the shortcomings of the conventional technology, and propose a method that can filter out fingerprint images with high coincidence and incompletely pressed fingerprint images to improve the efficiency of fingerprint entry and registration, which in turn can help improve The fingerprint recognition rate and the sliding unlocking experience are enhanced. The invention provides a fingerprint template registration method for improving the user experience effect, and a fingerprint recognition device for implementing the method.

The technical problem solved by the present invention can be achieved by adopting the following technical scheme: a fingerprint template registration method capable of quickly filtering images based on sliding sampling is proposed, which is based on a fingerprint acquisition sensor, a data processor, and a memory Fingerprint recognition device. The method includes the following steps:
Fingerprint template entry steps based on sliding sampling,
It prompts the user to enter a fingerprint template by sliding a finger across the fingerprint collection window;
The fingerprint collection sensor collects n fingerprint images;
Perform the following filtering steps on the n fingerprint images:
Select the starting image program,
It is based on the image coincidence discrimination criteria, and in the order of fingerprint collection time from front to back, to detect the degree of coincidence between two adjacent fingerprint images, and select the first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria. An earlier fingerprint image is the starting image;
Select Terminate Image Program,
It selects the first fingerprint image that meets the standard of full compression image as the termination image according to the standard of full compression image and the order of fingerprint collection time from back to front.
Fingerprint image selection procedure,
It is a set of fingerprint images from the start image to the end image. According to the screening criteria, a set of fingerprint image groups to be registered is selected;
Fingerprint template entry steps,
It processes and stores the fingerprint image groups to be registered in the memory;
Detect whether the sliding number threshold for sliding fingerprint entry is reached; if the sliding number threshold is not reached, return to prompt the user to enter a fingerprint template by sliding a finger across the fingerprint collection window; if the sliding number threshold is reached, terminate the fingerprint based on sliding sampling Template entry steps.

The procedure for selecting the starting image is as follows:
Set the image local area correlation coefficient threshold, the image local area correlation coefficient reflects the degree of overlap between the two images;
According to the order of fingerprint collection time from front to back, select the adjacent fingerprint images in turn, and perform the following image coincidence determination procedure:
It selects the same local area on the two adjacent fingerprint images, compares the two local areas, determines the degree of overlap between the two local areas, and calculates the local area correlation coefficient of the two fingerprint images.
If the correlation coefficient of the local area is greater than the threshold value of the correlation coefficient of the local area of the image, it is determined that the fingerprint images where the two local areas are located are coincident, which meets the image coincidence criterion; otherwise, the fingerprint images where the two local areas are located are respectively determined. Not coincident, does not meet the criteria for image coincidence;
When a first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria are found, an earlier one of the two adjacent fingerprint images is selected as the starting image.

The procedure for selecting the termination image is as follows:
Set a threshold for the proportion of non-pressed areas;
According to the order of fingerprint collection time from back to front, the non-pressing area determination step is performed for each fingerprint image.
The non-pressed area determination step is to calculate the pressed area and the non-pressed area of the fingerprint image according to the grayscale value of the fingerprint image, and the non-pressed area ratio of the non-pressed area to the fingerprint image area;
The non-pressed area ratio of the fingerprint image is compared with the non-pressed area ratio threshold. If the non-pressed area ratio is not greater than the non-pressed area ratio threshold, it is determined to meet the full compression image standard; If it is greater than the threshold value of the non-pressed area, it is determined that it does not meet the standard of fully pressed images;
When the first fingerprint image that meets the criteria of a full-press image is found, the fingerprint image is selected as the termination image.

The fingerprint image selection procedure is as follows:
In the candidate image group including sn fingerprint images obtained from the start image to the end image, each fingerprint image is sequentially numbered in the order of fingerprint collection time from front to back;
Determine the number of valid images valid_n according to the operating parameters of the fingerprint recognition device;
If sn≤valid_n, all sn fingerprint images are selected as the fingerprint image group to be registered;
If sn> valid_n, filtering and sampling are performed, and a fingerprint image with a serial number of sid is selected from the candidate image group as the i-th fingerprint image in the fingerprint image group to be registered.
sid = round (i ₀ + (i-1) × sf),
Where i ₀ is the serial number of the starting image;
i is the serial number of each fingerprint image in the fingerprint image group to be registered, and i is an independent variable, i is a natural number, 1≤i≤valid_n;
sid is the serial number of the fingerprint image in the candidate image group, and sid is a dependent variable, sid is a natural number, 1≤sid≤sn;
sf is the filtering sampling frequency, and sf = sn / valid_n, sf is a floating point number;
round (x) is a function that rounds the variable x to the integer obtained.

Specifically, the valid_n number of valid images determined according to the operating parameters of the fingerprint recognition device is:
Calculate the number of valid images valid_n = (tl / ll-1) / r + 1, where tl is the empirical value of the maximum total length of the human fingertips, and ll is the side length of a single square image collected by the fingerprint sensor , R is the coefficient of image coincidence, and 0.2≤r≤0.5.

Specifically, the screening step further includes the following procedures:
Create an image buffer area program,
Calculate the number of valid images valid_n,
valid_n = (tl / ll-1) / r + 1,
Among them, tl is the empirical value of the maximum total length of the human fingertips,
ll is the side length of a single square image collected by the fingerprint acquisition sensor,
r is the image overlap ratio coefficient, 0.2≤r≤0.5;
Calculate the number of stored images save_n,
save_n = q × valid_n,
Among them, q is the distribution ratio coefficient, 1.2≤q≤1.5;
In the memory, a storage space for storing save_n fingerprint images is allocated for the filtering step.

In addition, the fingerprint template registration method capable of quickly filtering images based on sliding sampling according to the present invention further includes the following fingerprint template entry steps based on push sampling:
Prompt the user to press with their fingers on the fingerprint collection window;
Obtaining a fingerprint image to be registered;
Detect whether the number of times of pressing fingerprint input threshold is reached. If the number of times of pressing fingerprint threshold is not reached, return to prompting the user to press the finger on the fingerprint collection window; if the number of times of pressing threshold is reached, perform the fingerprint template entry step;
Fingerprint template entry steps,
It processes and stores the fingerprint image group to be registered obtained by the above-mentioned press-type entry into the memory.

The technical problem solved by the present invention can also be achieved by adopting the following technical solutions:
Design and manufacture a fingerprint identification device based on a sliding sample registration fingerprint template and capable of quickly screening images, which includes a data processor, a fingerprint collection sensor electrically connected to the data processor, and a memory. The data processor includes more than two electronic components, and the composition and connection structure of each electronic component enables the fingerprint identification device to complete the following fingerprint template registration steps:
Fingerprint template entry steps based on sliding sampling,
It prompts the user to enter a fingerprint template by sliding a finger across the fingerprint collection window;
The fingerprint collection sensor collects n fingerprint images;
Perform the following filtering steps on the n fingerprint images:
Select the starting image program,
It is based on the image coincidence discrimination criteria, and in the order of fingerprint collection time from front to back, the degree of coincidence between two adjacent fingerprint images is detected, and the first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria are selected. The earliest fingerprint image is the starting image;
Select Terminate Image Program,
It selects the first fingerprint image that meets the standard of full compression image as the termination image according to the standard of full compression image and the order of fingerprint collection time from back to front.
Fingerprint image selection procedure,
It is a set of fingerprint images from the start image to the end image. According to the screening criteria, a set of fingerprint image groups to be registered is selected;
Fingerprint template entry steps,
It processes and stores the fingerprint image groups to be registered in the memory;
Detect whether the sliding number threshold for sliding fingerprint entry is reached; if the sliding number threshold is not reached, return to prompt the user to enter a fingerprint template by sliding a finger across the fingerprint collection window; if the sliding number threshold is reached, terminate the fingerprint based on sliding sampling Template entry steps.

In addition, the composition and connection structure of the electronic components of the data processor enable the fingerprint identification device to enable the fingerprint template registration step to further include the following fingerprint template entry procedure based on push-type sampling:
Prompt the user to press with their fingers on the fingerprint collection window;
Obtaining a fingerprint image to be registered;
Detect whether the number of times of pressing fingerprint input threshold is reached. If the number of times of pressing fingerprint threshold is not reached, return to prompting the user to press the finger on the fingerprint collection window; if the number of times of pressing threshold is reached, perform the fingerprint template entry step;
Fingerprint template entry steps,
It processes and stores the fingerprint image group to be registered obtained by the above-mentioned press-type entry into the memory.

Compared with the conventional technology, the technical effect of the "simplified fingerprint template registration method and fingerprint recognition device" of the present invention lies in:
By filtering the collected fingerprint images, the number of overlapping fingerprint images in the registered fingerprint image data processing step is reduced, and unless the fingerprint image is completely pressed, the fingerprint image is filtered, thereby reducing the number of fingerprint image registration steps. The amount of data processing, improve the data processing efficiency, help to realize the rapid entry of fingerprint templates, and improve the user's fingerprint entry experience; through the filtering step to eliminate duplicate and useless data and increase the effective fingerprint template registration information, improve fingerprint recognition rate.
The invention is based on a sliding sampling method, which helps to improve the fingerprint recognition sliding unlocking experience.
The present invention adopts a combined sampling method of sliding type and pressing type to capture the advantages of both, and overcomes the shortcomings of the two through a screening step, which is more helpful to improve the user experience and the fingerprint recognition rate.

Hereinafter, the embodiments shown in the drawings will be further described in detail.

The invention proposes a fingerprint recognition device based on a sliding sampling registration fingerprint template and capable of quickly filtering images. The first embodiment and the second embodiment of the present invention, as shown in FIG. 1, include a data processor 1, and a fingerprint collection sensor 2 and a memory 3 electrically connected to the data processor 1. The fingerprint identification device, as a biometric identification device, can be used in electronic devices such as computers, mobile communication terminals, tablet computers, electronic watches, video and audio players to implement unlocking functions, payment confirmation functions, and identity verification functions; Used to independently unlock devices, independent authentication devices, etc. When the fingerprint template is collected, fingerprint information is collected by the fingerprint collection sensor 2 and converted into original fingerprint data and sent to the data processor 1. The data processor 1 processes the original fingerprint data to form fingerprint template registration data and stores it in the memory 3. When performing biometric identification, the data processor 1 compares the fingerprint data collected by the fingerprint collection sensor 2 with the fingerprint template registration data stored in the memory 3 one by one, and then performs subsequent related processing based on the judgment result. step. The present invention focuses on the data processing steps for fingerprint template registration in the fingerprint template collection step. The data processor 1 includes more than two electronic components, and the composition and connection structure of each electronic component enables the fingerprint identification device to implement the fingerprint template registration method of the present invention based on sliding sampling that can quickly screen images.

In the first embodiment of the present invention, as shown in FIG. 2, in addition to identifying the fingerprint template registration start process 5 and fingerprint template registration end process 8 at the beginning of the method, the fingerprint based on the sliding sampling of the present invention can quickly filter the fingerprints of the image. The template registration method includes a fingerprint template entry step 6 based on sliding sampling.

Step 6 of fingerprint template entry based on sliding sampling includes the following procedures:
As shown in the flow 61 in FIG. 2, the user is prompted to input a fingerprint template by sliding a finger across the fingerprint collection window of the fingerprint collection sensor 2;
As shown in the flowchart 62 in FIG. 2, in the step of sliding a finger across the fingerprint collection window, the fingerprint collection sensor 2 collects n fingerprint images, as shown in FIG. 6;
As shown in a flow 63 in FIG. 2, a screening step is performed on the n fingerprint images. The filtering step includes a selection start image program 632, a selection termination image program 633, and a fingerprint image selection program 634. details as follows:
As shown in the flowchart 632 in FIG. 2, a starting image program is selected.
It is based on the image coincidence discrimination criteria, and in the order of fingerprint collection time from front to back, to detect the degree of coincidence between two adjacent fingerprint images, and select the first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria. The earlier fingerprint image is the starting image;
As shown in the flow 633 in FIG. 2, the termination image program is selected.
It selects the first fingerprint image that meets the standard of full compression image as the termination image according to the standard of full compression image and the order of fingerprint collection time from back to front.
As shown in the flowchart 634 in FIG. 2, a fingerprint image selection process is performed.
It is a set of fingerprint images from the start image to the end image. According to the screening criteria, a set of fingerprint image groups to be registered is selected. The screening criteria are used to eliminate as far as possible the fingerprint images that do not require subsequent processing. , And select the fingerprint image group to be registered, thereby reducing the amount of data processing in subsequent processes, and increasing fingerprint template registration information that is helpful for fingerprint identification.

After completing the above screening steps, perform the fingerprint template entry steps:
As shown in flow 64 in FIG. 2, each fingerprint image group to be registered is processed and stored in the memory 3;
As shown in the flow 65 in FIG. 2, it is detected whether the sliding number threshold of the sliding fingerprint entry has been reached; if the sliding number threshold has not been reached, the process returns to prompting the user to input a fingerprint template by sliding a finger across the fingerprint collection window, and then returns to FIG. Process 61 shown; if the threshold of the number of slides is reached, as shown in process 66 shown in FIG. 2, the fingerprint template entry step based on sliding sampling is terminated, and the fingerprint template entry based on sliding sampling is completed.

The present invention is based on a sliding sampling method, so that the fingerprint template stored in the fingerprint recognition device can be used for sliding fingerprint recognition and unlocking accordingly, which helps to improve the fingerprint recognition sliding unlocking experience. The present invention reduces the overlapped fingerprint images in the registered fingerprint image data processing step through the screening step 63 of the collected fingerprint images, removes the fingerprint image unless it is completely pressed, and filters the fingerprint image to reduce the fingerprint image. The amount of data processing in the registration step, thereby improving the data processing efficiency, is beneficial to the rapid entry of fingerprint templates, and improves the user's fingerprint entry experience; the filtering step 63 eliminates duplicate and useless data, and increases effective fingerprint template registration information To improve fingerprint recognition rate.

The execution order of the selection start image program 632 and the selection termination image program 633 in the screening step 63 may be interchanged.

The screening steps are mainly aimed at three practical problems encountered in sliding fingerprint entry:
1. Fingerprint images collected in the initial stage of finger sliding are prone to the problem of complete overlap of adjacent images;
2. Fingerprint images collected at the end of finger sliding are prone to problems with non-finger finger images;
3. Adjacent fingerprint images collected in the middle of finger sliding are still prone to problems with large coincident areas.

The above problems will bring more redundant data and increase the amount of data processing for subsequent fingerprint image stitching processing and fingerprint template processing. The screening step 63 is to overcome the above three problems, and pre-process the collected fingerprint images purposefully, thereby reducing the amount of data processing for subsequent processing, thereby improving data processing efficiency and user experience.

In order to solve the problem that the fingerprint images collected at the initial stage of finger sliding are prone to completely overlap with adjacent images, the initial image program 632 is selected to distinguish the fingerprint images collected at the initial stage. Speed and experience settings, discard the first few fingerprint images collected, or use image matching to determine the overlapping images and discard them. The first embodiment of the present invention is to select an initial image program 632 to determine the degree of image overlap by using a fixed local area correlation coefficient of the image. As shown in FIG. 5, at a fixed position on two adjacent fingerprint images S1 and S2, a local area Z is selected, and their correlation degree is judged according to the data of the local area Z; the data of the local area Z such as the number of feature points and features The value of points, the contrast of fixed local areas, etc., are used to express the degree of correlation between them using the image local area correlation coefficient Y. The closer the local area correlation coefficient Y approaches 1, it means that the two adjacent fingerprint images S1, S2 tend to completely coincide. Thus, the degree of coincidence of adjacent fingerprint images is determined by the correlation coefficient of the local area of the image.

In the first embodiment of the present invention, as shown in FIG. 6, the selection of the starting image program 632 is as follows:
The threshold value of the correlation coefficient of the local area of the image is set. In the first embodiment of the present invention, the threshold value of the correlation coefficient of the local area of the image is set to 0.9. The local area correlation coefficient of the image reflects the degree of overlap between the two images.

According to the order of fingerprint collection time from front to back, select the adjacent fingerprint images in turn, and perform the following image coincidence determination procedure:
Select the same local area on the two adjacent fingerprint images, compare the two local areas, determine the degree of overlap between the two local areas, and calculate the local area correlation coefficient of the two fingerprint images;
If the correlation coefficient of the local area is greater than the threshold of the correlation coefficient of the local area of the image, it is determined that the fingerprint images where the two local areas are located are coincident, and the image coincidence determination criterion is met; otherwise, the fingerprint images of the two local areas are determined respectively. Images do not coincide, do not meet the criteria for image coincidence;
When a first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria are found, an earlier one of the two adjacent fingerprint images is selected as the starting image.

In the first embodiment of the present invention, as shown in FIG. 6, among the n fingerprint images sorted according to the acquisition time, comparison is performed from the adjacent image 1 and image 2. For the respective local areas Z of image 1 and image 2, calculate their image local area correlation coefficients; the image local area correlation coefficients of image 1 and image 2 are greater than the image local area correlation coefficient threshold of 0.9, so as to judge Image 1 and image 2 are superimposed images. Next, for the respective local regions Z of image 2 and image 3, the image local region correlation coefficients are calculated. If the image local region correlation coefficients of image 2 and image 3 are smaller than the image local region correlation coefficient threshold 0.9 , Then image 3 has a significant shift from image 2 and it is judged that image 2 and image 3 are not coincident images, and in image 2 and image 3, the earlier collected image 2 is selected as the starting image image.

In the first embodiment of the present invention, the selection of the starting image program 632 uses local areas to complete the comparison data calculation, and completes the comparison with a small amount of data processing, which has high stability and execution efficiency.

In order to solve the problem that the fingerprint images collected at the end of the finger sliding are prone to non-finger finger images, the termination image program 633 is selected to distinguish the fingerprint images collected at the final stage from non-finger finger images; depending on the application scenario, Slide your finger over the speed and experience value settings to discard the last few fingerprint images collected, or judge the non-finger finger images by the number of image feature points and discard them. In the first embodiment of the present invention, the selection termination image program 633 distinguishes incomplete finger images that are not completely pressed according to the image gray value. The difference in gray value can be determined by setting a dynamic gray threshold based on the real-time lighting conditions of the application environment.

In the first embodiment of the present invention, the non-pressed area is determined by using the gray-scale threshold division method, which is as follows:
Set the non-pressed area ratio threshold. In the first embodiment of the present invention, set the non-pressed area ratio threshold to 10%.
According to the order of fingerprint collection time from back to front, perform non-pressing area determination steps for each fingerprint image:
Calculating the pressed area and the non-pressed area of the fingerprint image and the non-pressed area ratio of the area of the fingerprint image to the area of the fingerprint image according to the gray value of the fingerprint image;
The non-pressed area ratio of the fingerprint image is compared with the non-pressed area ratio threshold. If the non-pressed area ratio is not greater than the non-pressed area ratio threshold, it is determined to meet the full compression image standard; if the non-pressed area ratio is If it is larger than the threshold value of the non-pressed area, it is determined that the standard of the fully pressed image is not met;
When the first fingerprint image that meets the criteria of a full-press image is found, the fingerprint image is selected as the termination image.

In the first embodiment of the present invention, as shown in FIG. 6, among the n fingerprint images sorted according to the acquisition time, the termination image selection is performed in reverse order, that is, the selection step is started from the image n. The pressed area and the non-pressed area are calculated for the gray value of the image n, and the non-pressed area ratio of the area of the non-pressed area to the area of the image n is determined. Comparing this ratio with the threshold value of the non-pressed area ratio of 10%, as shown in FIG. 6, it is obvious that the non-pressed area ratio of the image n is greater than 10%, thereby determining that it does not meet the standard of the fully pressed image. Next, the image n-1 is judged, and so on, until the image n-4 whose non-pressed area ratio is greater than the non-pressed area ratio threshold 10% is found, that is, the image n-4 is the first to meet the safety Press the standard fingerprint image to select image n-4 as the termination image.

In the first embodiment of the present invention, as shown in FIG. 6, after selecting a start image program 632 and a stop image program 633, image 2 is selected as a start image, and image n-4 is a stop image. A total of n-5 candidate image groups.

In order to solve the problem that adjacent fingerprint images collected in the middle stage of finger sliding are still prone to have a large coincidence area, the fingerprint image selection program 634 should perform image registration difference on the images in the candidate image group. Image matching method for image comparison, feature point discrimination method for feature point data comparison, etc.

The first embodiment of the present invention adopts a sampling extraction method, which is as follows:
In the candidate image group including sn fingerprint images obtained from the start image to the end image, each fingerprint image is sequentially numbered in the order of fingerprint collection time from front to back;
According to the operating parameters of the fingerprint recognition device, the number of valid images is determined to be valid_n; the operating parameters of the fingerprint recognition device include the sampling frequency, the size of the fingerprint collection window, the set coincidence degree, the set finger length experience value, etc .; the number of valid images valid_n is limited The maximum number of fingerprint images in the fingerprint image group to be registered. If the value exceeds this value, redundant data may appear, increasing the amount of data processing.
If sn≤valid_n, all sn fingerprint images are selected as the fingerprint image group to be registered;
If sn> valid_n, perform filtering and sampling, and select a fingerprint image whose serial number is sid from the candidate image group as the i-th fingerprint image in the fingerprint image group to be registered,
sid = round (i ₀ + (i-1) × sf),
Where i ₀ is the serial number of the starting image;
i is the serial number of each fingerprint image in the fingerprint image group to be registered, i is an independent variable, and i is a natural number, 1≤i≤valid_n;
sid is the serial number of the fingerprint image in the candidate image group, sid is the dependent variable, and sid is a natural number, 1≤sid≤sn;
sf is the filtering sampling frequency, sf = sn / valid_n, sf is a floating point number;
round (x) is a function that rounds the variable x to the integer obtained.

In the first embodiment of the present invention, the number of valid images determined according to the operating parameters of the fingerprint recognition device is valid_n = 7, and if n = 11, the backup image selected according to the foregoing selection start image program 632 and selection end image program 633 is selected. Select the number of fingerprint images sn = n-5 = 6 in the image group, that is, images 2 to 7. Because sn <valid_n in this assumption, all 6 fingerprint images of the candidate image group are selected as the fingerprint image group to be registered.

If n = 20, then the number of fingerprint images of the candidate image group selected according to the foregoing selection start image program 632 and selection termination image program 633 is sn = n-5 = 15, that is, images 2 to 16 , And images 2 to 16 have been numbered according to the order of fingerprint collection time from front to back, that is, i ₀ = 2, the screening sampling is performed at the screening sampling frequency of sf = sn / valid_n = 15 / 7≈2.14:
When i = 1, sid = round (i ₀ + (i-1) × sf) = round (2+ (1-1) × 2.14) = 2, that is, the first fingerprint image in the fingerprint image group to be registered Like image 2 of the candidate image group;
When i = 2, sid = round (i ₀ + (i-1) × sf) = round (2+ (2-1) × 2.14) = 4, that is, the second fingerprint image in the fingerprint image group to be registered Like image 4 of the candidate image group;
When i = 3, sid = round (i ₀ + (i-1) × sf) = round (2+ (3-1) × 2.14) = 6, that is, the third fingerprint image in the fingerprint image group to be registered Like image 6 of the candidate image group;
When i = 4, sid = round (i ₀ + (i-1) × sf) = round (2+ (4-1) × 2.14) = 8, that is, the fourth fingerprint image in the fingerprint image group to be registered Like image 8 of the candidate image group;
When i = 5, sid = round (i ₀ + (i-1) × sf) = round (2+ (5-1) × 2.14) = 11, that is, the fifth fingerprint image in the fingerprint image group to be registered Like image 11 of the candidate image group;
When i = 6, sid = round (i ₀ + (i-1) × sf) = round (2+ (6-1) × 2.14) = 13, that is, the sixth fingerprint image in the fingerprint image group to be registered Like image 13 of the candidate image group;
When i = 7, sid = round (i ₀ + (i-1) × sf) = round (2+ (7-1) × 2.14) = 15, that is, the seventh fingerprint image in the fingerprint image group to be registered Like image 15 of the candidate image group.

In the fingerprint image selection program 634 of the present invention, the problem of having a large overlap area that may occur in adjacent fingerprint images collected in the middle stage of finger sliding can be avoided by screening and sampling as much as possible.

According to the first embodiment of the present invention, the specific method for determining the valid number of valid images according to the operating parameters of the fingerprint recognition device is:
As shown in Figure 3, the maximum total length of the human fingertips set in the fingerprint recognition device is tl. As shown in Figure 4, the fingerprint collection window of the fingerprint collection sensor 2 captures a single square image with a side length of ll. , Set the scaling coefficient r that reflects the degree of coincidence of the fingerprint image, 0.2≤r≤0.5, then the number of valid images valid_n = (tl / ll-1) / r + 1.

In the first embodiment of the present invention, as shown in FIG. 2, in order to have sufficient storage space to process the filtering step data, the filtering step 63 further includes establishing an image cache area program 631; the image cache area program 631 should generally be set before the filtering process. That is, before the start image program 632, the end image program 633, and the fingerprint image selection program 634 are selected.

The program for creating an image buffer area 631 includes:
Calculate the number of valid images valid_n,
valid_n = (tl / ll-1) / r + 1,
Among them, tl is the empirical value of the maximum total length of the human fingertips,
ll is the side length of a single square image collected by the fingerprint acquisition sensor,
r is the image overlap ratio coefficient, 0.2≤r≤0.5;
Calculate the number of stored images save_n,
save_n = q × valid_n,
Among them, q is the distribution ratio coefficient, 1.2≤q≤1.5;
In the memory 3, for the filtering step, a storage space for storing save_n fingerprint images is allocated.

According to the present invention, a fingerprint template registration method capable of quickly filtering images based on sliding sampling is based on a fingerprint template entry step 6 based on sliding sampling, and a fingerprint template entry step 7 based on push sampling is also added.

In the second embodiment of the present invention, as shown in FIG. 7, the fingerprint template entry step 7 based on the push-type sampling includes:
As shown in the flow 71 in FIG. 7, the user is prompted to press the fingerprint collection window of the fingerprint collection sensor 2 with a finger;
As shown in flow 72 in FIG. 7, a fingerprint image to be registered is acquired;
As shown in the flow 73 in FIG. 7, it is detected whether the threshold of the number of times of pressing fingerprint input has been reached. If the number of times of the pressure has not been reached, it returns to prompting the user to press the finger on the fingerprint collection window, that is, the flow 71 shown in FIG. 7; The threshold of the number of times of pressing performs the fingerprint template entry step, which is the process 74 shown in FIG. 7;
As shown in the flow 74 in FIG. 7, the fingerprint template entry step processes the fingerprint image group to be registered obtained by the above-mentioned push-type entry and stores it in the memory 3, thereby completing the fingerprint template entry step 7 based on the push-type sampling.

The fingerprint template entry step 6 based on sliding sampling and the fingerprint template entry step 7 based on push sampling can be performed interchangeably, that is, the execution order of the two can be adjusted according to usage, without affecting the ability to quickly filter images based on sliding sampling. Implementation of the fingerprint template registration method.

The composition and connection structure of the electronic components of the data processor 1 enable the fingerprint recognition device to implement a fingerprint template registration method based on sliding sampling that can quickly filter images based on the second embodiment of the present invention.

The invention adopts a combined sampling method of sliding type and pressing type to capture the advantages of both, and overcomes the shortcomings of the two through a screening step, which is more conducive to improving the user experience and the fingerprint recognition rate.

The above are only preferred embodiments of the present application. It should be noted that for those of ordinary skill in the art, without departing from the principles of the present application, several improvements and retouches can be made. These improvements and Retouching should also be considered as the protection scope of this application.

1‧‧‧data processor

2‧‧‧Fingerprint sensor

3‧‧‧Memory

5 ~ 8‧‧‧flow

61 ~ 66‧‧‧ flow

631 ~ 634‧‧‧Process

71 ~ 74‧‧‧flow

1, 2,…, n‧‧‧ images (several)

A‧‧‧Instructions

tl‧‧‧Maximum total length of human fingertips

ll‧‧‧side length

S1‧‧‧ fingerprint image

S2‧‧‧ fingerprint image

Y‧‧‧ local correlation coefficient

Z‧‧‧ local area

FIG. 1 is a schematic block diagram of the electrical principle of the first embodiment of the “swipe-based fingerprint template registration method and fingerprint identification device” according to the present invention.

FIG. 2 is a schematic flowchart of a fingerprint template registration method according to the first embodiment.

Figure 3 is a schematic diagram of a typical full fingerprint.

FIG. 4 is a partially enlarged schematic diagram of an indicated portion in FIG. 3.

FIG. 5 is a schematic diagram of a principle of judging a local area coincidence of an adjacent fingerprint image.

FIG. 6 is a schematic diagram of selecting a start image and a stop image according to the first embodiment.

FIG. 7 is a schematic flowchart of a fingerprint template registration method according to a second embodiment of the present invention.

Claims (9)

A fingerprint template registration method capable of quickly filtering images based on sliding sampling is based on a fingerprint recognition device including a fingerprint collection sensor, a data processor, and a memory. The method is characterized in that the method includes the following steps: Fingerprint template entry steps based on sliding sampling, It prompts the user to enter a fingerprint template by sliding a finger across the fingerprint collection window; The fingerprint collection sensor collects n fingerprint images; Perform the following filtering steps on the n fingerprint images: Select the starting image program, It is based on the image coincidence discrimination criteria, and in the order of fingerprint collection time from front to back, to detect the degree of coincidence between two adjacent fingerprint images, and select the first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria. An earlier fingerprint image is the starting image; Select Terminate Image Program, It selects the first fingerprint image that meets the standard of full compression image as the termination image according to the standard of full compression image and the order of fingerprint collection time from back to front. Fingerprint image selection procedure, It is a set of fingerprint images from the start image to the end image. According to the screening criteria, a set of fingerprint image groups to be registered is selected; Fingerprint template entry steps, It processes and stores the fingerprint image groups to be registered in the memory; Detect whether the sliding number threshold for sliding fingerprint entry is reached; if the sliding number threshold is not reached, return to prompt the user to enter a fingerprint template by sliding a finger across the fingerprint collection window; if the sliding number threshold is reached, terminate the fingerprint based on sliding sampling Template entry steps. The fingerprint template registration method based on sliding sampling that can quickly filter images as described in claim 1, wherein: Select the starting image program, It sets the threshold of the local area correlation coefficient of the image, and the local area correlation coefficient of the image reflects the degree of overlap between the two images; According to the order of fingerprint collection time from front to back, select the adjacent fingerprint images in turn, and perform the following image coincidence determination procedure: It selects the same local area on the two adjacent fingerprint images, compares the two local areas, determines the degree of overlap between the two local areas, and calculates the local area correlation coefficient of the two fingerprint images; If the correlation coefficient of the local area is greater than the threshold value of the correlation coefficient of the local area of the image, it is determined that the fingerprint images where the two local areas are located are coincident, which meets the image coincidence criterion; otherwise, the fingerprint images where the two local areas are located are respectively determined. Not coincident, does not meet the criteria for image coincidence; When a first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria are found, an earlier one of the two adjacent fingerprint images is selected as the starting image. The fingerprint template registration method based on sliding sampling that can quickly filter images as described in claim 1, wherein: Select Terminate Image Program, It sets the threshold for the proportion of non-pressed areas; According to the order of fingerprint collection time from back to front, the non-pressing area determination step is performed for each fingerprint image. The non-pressed area determination step is to calculate the pressed area and the non-pressed area of the fingerprint image according to the grayscale value of the fingerprint image, and the non-pressed area ratio of the non-pressed area to the fingerprint image area; The non-pressed area ratio of the fingerprint image is compared with the non-pressed area ratio threshold. If the non-pressed area ratio is not greater than the non-pressed area ratio threshold, it is determined to meet the full compression image standard; if the non-pressed area ratio is If it is greater than the threshold value of the non-pressed area, it is determined that it does not meet the standard of fully pressed images; When the first fingerprint image that meets the criteria of a full-press image is found, the fingerprint image is selected as the termination image. The method for registering a fingerprint template capable of quickly filtering images based on sliding sampling according to claim 1, wherein the fingerprint image selection process includes sn fingerprints obtained from a start image to a stop image. In the candidate image group of the image, each fingerprint image is sequentially numbered in the order of fingerprint collection time from front to back; the number of valid images is determined to be valid_n according to the operating parameters of the fingerprint recognition device; if sn≤valid_n, all The sn fingerprint images are all selected as the fingerprint image group to be registered; if sn> valid_n, filtering and sampling are performed, and the fingerprint image with the sid number is selected from the candidate image group as the fingerprint image group to be registered. The ith fingerprint image, sid = round (i ₀ + (i-1) × sf), where i ₀ is the serial number of the starting image; i is the fingerprint image of each fingerprint image in the fingerprint image group to be registered Serial number, and i is an independent variable, i is a natural number, 1≤i≤valid_n; sid is the serial number of the fingerprint image in the candidate image group, and sid is a dependent variable, sid is a natural number, 1≤sid≤sn; sf is the filtering sampling frequency, and sf = sn / valid_n, sf is a floating point number round (x) is a function of the variable x rounded to an integer obtained. The fingerprint template registration method based on sliding sampling that can quickly filter images as described in claim 4, wherein, According to the operating parameters of the fingerprint recognition device, the number of valid images is determined to be valid_n, Calculate the number of valid images valid_n = (tl / ll-1) / r + 1, where tl is the empirical value of the maximum total length of the human fingertips, and ll is the side length of a single square image collected by the fingerprint sensor , R is the coefficient of image coincidence, and 0.2≤r≤0.5. The fingerprint template registration method based on sliding sampling that can quickly filter images as described in claim 1, wherein: The screening step further includes the following procedures: Create an image buffer area program, It calculates the number of valid images valid_n, valid_n = (tl / ll-1) / r + 1, Among them, tl is the empirical value of the maximum total length of the human fingertips, ll is the side length of a single square image collected by the fingerprint acquisition sensor, r is the image overlap ratio coefficient, 0.2≤r≤0.5; Calculate the number of stored images save_n, save_n = q × valid_n, Among them, q is the distribution ratio coefficient, 1.2≤q≤1.5; In the memory, for the filtering step, a storage space for storing save_n fingerprint images is allocated. The fingerprint template registration method according to any one of claims 1 to 6, which can quickly filter images based on sliding sampling, wherein: It also includes the following fingerprint template entry steps based on push sampling: Prompt the user to press with their fingers on the fingerprint collection window; Obtaining a fingerprint image to be registered; Detect whether the number of times of pressing fingerprint input threshold is reached. If the number of times of pressing fingerprint threshold is not reached, return to prompting the user to press the finger on the fingerprint collection window; if the number of times of pressing threshold is reached, perform the fingerprint template entry step; Fingerprint template entry steps, It processes and stores the fingerprint image group to be registered obtained by the above-mentioned press-type entry into the memory. A fingerprint recognition device based on a sliding sampling registration fingerprint template and capable of quickly filtering images is characterized by: The data processor includes a data processor, a fingerprint collection sensor and a memory electrically connected to the data processor; The data processor includes more than two electronic components, and the composition and connection structure of each electronic component enables the fingerprint identification device to complete the following fingerprint template registration steps: Fingerprint template entry steps based on sliding sampling, It prompts the user to enter a fingerprint template by sliding a finger across the fingerprint collection window; The fingerprint collection sensor collects n fingerprint images; Perform the following filtering steps on the n fingerprint images: Select the starting image program, It is based on the image coincidence discrimination criteria, and in the order of fingerprint collection time from front to back, to detect the degree of coincidence between two adjacent fingerprint images, and select the first pair of adjacent fingerprint images that do not meet the image coincidence discrimination criteria The earliest fingerprint image is the starting image; Select Terminate Image Program, It selects the first fingerprint image that meets the standard of full compression image as the termination image according to the standard of full compression image and the order of fingerprint collection time from back to front. Fingerprint image selection step, It is a set of fingerprint images from the start image to the end image. According to the screening criteria, a set of fingerprint image groups to be registered is selected; Fingerprint template entry steps, It processes and stores the fingerprint image groups to be registered in the memory; Detect whether the sliding number threshold for sliding fingerprint entry has been reached; if the sliding number threshold has not been reached, return to starting sliding fingerprint entry; if the sliding number threshold is reached, terminate the fingerprint template entry step based on sliding sampling. The fingerprint recognition device based on the sliding sampling registration fingerprint template as described in claim 8 and capable of quickly filtering images, wherein: The composition and connection structure of the electronic components of the data processor, so that the fingerprint identification device allows the fingerprint template registration step to include the following fingerprint template entry procedure based on press-type sampling: Prompt the user to press with their fingers on the fingerprint collection window; Obtaining a fingerprint image to be registered; Detect whether the number of times of pressing fingerprint input threshold is reached. If the number of times of pressing fingerprint threshold is not reached, return to prompting the user to press the finger on the fingerprint collection window; if the number of times of pressing threshold is reached, perform the fingerprint template entry step; Fingerprint template entry steps, It processes and stores the fingerprint image group to be registered obtained by the above-mentioned press-type entry into the memory.