Finger live anti-counterfeiting method based on optical coherence tomography with speckle variance

Abstract

The present invention relates to the technical field of image processing and fingerprint live anti-counterfeiting, and in particular to a finger live anti-counterfeiting method based on speckle variance optical coherence tomography. The present invention obtains a time-series B-scan image of a sample through B-scan scanning; extracts feature regions from the time-series B-scan image and retains the region of interest; uses speckle variance technology to remove interference from static structural information in the time-series B-scan image, extracts live information therein, and removes residual invalid information through threshold segmentation; compares the result of the threshold segmentation with a set live threshold to obtain a live anti-counterfeiting result of the sample. The present invention has a fast running speed, does not increase additional computing burden, and has a high degree of versatility and flexibility.

Description

Technical Field

[0001] The present invention relates to the technical fields of image processing and fingerprint liveness anti-counterfeiting technology, and particularly relates to a finger liveness anti-counterfeiting method based on speckle variance optical coherence tomography (OCT). Background Art

[0002] With the rapid development of information technology, identity recognition technology has been widely applied in fields such as social security management and privacy protection. Biometric recognition technology is an important way in identity recognition technology. Compared with traditional identity authentication methods (passwords, keys, etc.), biometric recognition technology has a series of unique advantages such as convenience, uniqueness, and confidentiality, and plays a very important role in the field of public security.

[0003] Among many biometric recognition technologies, fingerprints are widely used in security fields such as immigration management, criminal investigation, mobile payment, and access control due to their universality, uniqueness, and persistence. Most fingerprint recognition systems mainly rely on obtaining fingerprint images from the finger surface. The information extracted by this way of extracting surface information is extremely limited and is easily subject to forgery attacks. Optical coherence tomography (OCT) is a non-contact high-resolution optical imaging technology. As a non-invasive technology, it can perform high-resolution in-vivo cross-sectional real-time imaging of biological tissues and samples, can penetrate deep into the skin, and image the internal structure of the finger. Therefore, the OCT technology can obtain the tissue structure information at the depth under the finger skin, provide additional anti-counterfeiting features, and play an important role in the field of fingerprint recognition.

[0004] However, traditional OCT fingerprint anti-counterfeiting methods mainly use structural item information such as internal and external fingerprint information of the finger and the number of subcutaneous sweat glands for anti-counterfeiting. Although these anti-counterfeiting methods can resist different types of forgery attacks to a certain extent, there are still defects. These methods mainly focus on the structural item information in the finger, regard the flow item information coupled therein as noise in the picture, and ignore the liveness information contained therein. Therefore, traditional OCT fingerprint anti-counterfeiting methods can only classify forged materials and cannot achieve liveness anti-counterfeiting. If liveness anti-counterfeiting needs to be achieved, additional liveness detection devices and complex processing procedures often need to be introduced. Summary of the Invention

[0005] The purpose of the present invention is to solve the deficiencies existing in traditional OCT fingerprint anti-counterfeiting methods, and propose a finger liveness anti-counterfeiting method based on speckle variance optical coherence tomography. By processing and calculating the sequential B-scan images collected by the OCT device, liveness anti-counterfeiting is achieved.

[0006] To achieve the above purpose, the technical solution provided by the present invention is as follows:

[0007] The finger in-vivo anti-counterfeiting method based on speckle variance optical coherence tomography includes:

[0008] Collecting the sequential B-scan images of the current sample at a specific position by B-scan scanning;

[0009] Extracting the feature regions from each sequential B-scan image to obtain the final feature regions of all sequential B-scan images, and cropping each sequential B-scan image according to the final feature regions to obtain the corresponding feature region images;

[0010] Calculating the mean value of the brightness of all the feature region images, and calculating the speckle variance according to the mean value of the brightness;

[0011] Performing threshold segmentation on the feature region images according to the speckle variance, and calculating the in-vivo probability of the current sample according to the result of the threshold segmentation;

[0012] Comparing the in-vivo probability of the current sample with the in-vivo threshold to obtain the anti-counterfeiting result of the current sample.

[0013] Further, the extracting the feature regions from each sequential B-scan image to obtain the final feature regions of all sequential B-scan images includes:

[0014] Obtaining the length and width of the sequential B-scan image;

[0015] Calculating the horizontal gradient of each row of pixels of each sequential B-scan image, which is expressed by the formula as follows:

[0016]

[0017] where M x represents the horizontal gradient of the x-th row of pixels of the current sequential B-scan image, W represents the width of the sequential B-scan image, I(x, y) represents the brightness at the position (x, y), and I(x - 1, y) represents the brightness at the position (x - 1, y);

[0018] Traversing each row from top to bottom according to the length of the current sequential B-scan image, setting the row where the horizontal gradient is greater than the preset threshold encountered for the first time as the starting position of the feature region, and after setting the starting position, setting the row where the horizontal gradient is less than the preset threshold encountered for the first time as the ending position of the feature region to obtain the feature region of each sequential B-scan image;

[0019] Calculating the average value of the feature regions of all sequential B-scan images to obtain the final feature region.

[0020] Further, the calculating the mean value of the brightness of all the feature region images includes:

[0021] Traverse and read all the feature region images, and calculate the average brightness of all the feature region images, which is expressed by the formula as follows:

[0022]

[0023] Wherein, represents the average brightness at (x, y) of all the feature region images, N represents the number of the feature region images, and I i (x, y) represents the brightness at (x, y) of the i-th feature region image.

[0024] Furthermore, calculating the speckle variance according to the average brightness includes:

[0025] Traverse and read all the feature region images, and calculate the speckle variance, which is expressed by the formula as follows:

[0026]

[0027] Wherein, SV(x, y) represents the speckle variance at (x, y) of the feature region image.

[0028] Furthermore, performing threshold segmentation on the feature region image according to the speckle variance includes:

[0029] Calculate the adaptive threshold, which is expressed by the formula as follows:

[0030]

[0031] Wherein, c represents a fixed constant, and T represents the adaptive threshold;

[0032] Perform threshold segmentation according to the speckle variance and the adaptive threshold, which is expressed by the formula as follows:

[0033]

[0034] Wherein, B(x, y) represents the result of threshold segmentation at (x, y) of the feature region image.

[0035] Furthermore, calculating the liveness probability of the current sample according to the result of threshold segmentation is expressed by the formula as follows:

[0036]

[0037] Wherein, SSI represents the liveness probability of the current sample, R represents the width of the feature region image, and C represents the height of the feature region image.

[0038] Further, comparing the liveness probability of the current sample with the liveness threshold to obtain the anti-counterfeiting result of the current sample, which is expressed by the following formula:

[0039]

[0040] Among them, Result represents the anti-counterfeiting result of the current sample. When Result = true, the anti-counterfeiting result is liveness; when Result = false, the anti-counterfeiting result is non-liveness. P represents the liveness threshold.

[0041] Compared with the prior art, the significant advantages of the present invention are as follows: 1. For the first time, the internal relationship between the speckle variance result and finger liveness anti-counterfeiting is analyzed in the time dimension, and the liveness information in the changing speckles in the time series data is extracted as the anti-counterfeiting feature. 2. Only a small number of B-scan images need to be collected to achieve the purpose of liveness detection, with a small computational burden and a fast running speed. The detection result can be obtained within 0.5 seconds at the fastest. 3. No additional detection equipment is required, and the original OCT system does not need to be modified. It is seamlessly compatible with any OCT device, with high versatility and flexibility. 4. Calculate the liveness probability of the current sample according to the result of threshold segmentation, which can obtain a high discrimination degree and classification accuracy on true and false fingers, with high accuracy, and can accurately distinguish true fingers from various forgery materials, with high application value. Description of the Drawings

[0042] Figure 1 is a flowchart of the finger liveness anti-counterfeiting method based on speckle variance optical coherence tomography of the present invention;

[0043] Figure 2 is a physical diagram of the OCT system of the present invention;

[0044] Figure 3 is a schematic diagram of the principle of the OCT system for collecting time series B-scan data of the present invention;

[0045] Figure 4 is a schematic diagram of the algorithm flow of the robust liveness finger detection method based on improved speckle variance optical coherence tomography of the present invention. Detailed Embodiments

[0046] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0047] The finger in-vivo anti-counterfeiting method based on speckle variance optical coherence tomography provided by the present invention: Place the finger or sample in the acquisition window of the OCT system, perform B-scan scanning at a fixed position of the finger, and obtain sequential B-scan images. Extract the feature regions from the sequential B-scan images and retain the regions of interest. Use the speckle variance technique to remove the interference of static structural information in the sequential B-scan images, extract the in-vivo information therein, and after threshold segmentation, remove the remaining invalid information. Compare the result of the threshold segmentation with the set in-vivo threshold to obtain the in-vivo anti-counterfeiting result of the sample.

[0048] As Figure 1 shown, the specific steps are as follows:

[0049] 1) Set the acquisition mode of the OCT system, place the finger in the acquisition window, perform multiple B-scan scans at the same position of the finger, collect sequential B-scan images, and the processing steps are as follows:

[0050] (11) Set the OCT system parameters, and the specific steps are as follows:

[0051] (111) Set the OCT system trigger mode to Trigmodel, which means setting the CCD camera to external trigger. The physical diagram of the OCT system is as Figure 2 shown;

[0052] (112) Set the galvanometer scanning mode to fixed position scanning;

[0053] (113) Set the initial light spot position to (0, 0), which means scanning at the origin position;

[0054] (114) Set the system scanning frequency to 50 kHz.

[0055] (12) Set the OCT host computer software parameters, and the specific steps are as follows:

[0056] (121) Power on the OCT system and start the host computer software;

[0057] (122) Set the A-line scan length;

[0058] (123) Set the number of scans, that is, set the number of sequential B-scan images finally obtained;

[0059] (124) Set the file saving path.

[0060] (13) Perform B-scan scanning at a fixed position of the finger to collect sequential B-scan images. The principle of the OCT system is as Figure 3 shown, and the specific steps are as follows:

[0061] (131) Place a sample such as a finger on the collection window and keep the finger stationary.

[0062] (132) Run the host computer program to perform a B-scan on the fixed position of the finger and collect sequential B-scan images.

[0063] 2) According to the sequential B-scan images, use the feature region extraction algorithm to extract the region of interest to eliminate the influence of irrelevant regions and reduce the waste of subsequent computing resources. Use the speckle variance technique to remove the stationary structural information in the sample and extract the living body information existing in the sample. The specific process is as Figure 4 shown, and the processing steps are as follows:

[0064] (21) Extract the feature region from the sequential B-scan images according to formula (1). The specific steps are as follows:

[0065] (211) Read the sequential B-scan images and obtain the length and width of the sequential B-scan images.

[0066] (212) Calculate the average horizontal gradient of each row of the sequential B-scan images according to formula (1), that is:

[0067]

[0068] where M x represents the horizontal gradient of the pixels in the x-th row of the current sequential B-scan image, W represents the width of the sequential B-scan image, I(x, y) represents the brightness at position (x, y), and I(x - 1, y) represents the brightness at position (x - 1, y).

[0069] (213) Traverse each row from top to bottom according to the length of the sequential B-scan images, and set the position where the horizontal gradient first exceeds the preset threshold as the starting position of the feature region;

[0070] (214) After setting the starting position, set the position where the horizontal gradient first is less than the preset threshold as the ending position of the feature region;

[0071] (215) Calculate the feature region of each sequential B-scan image according to the above method, take the average value, and determine the starting position and ending position of the final feature region;

[0072] (216) Crop all the sequential B-scan images according to the final feature region, remove the regions with invalid information above and below the final feature region, and retain the middle final feature region;

[0073] (217) Save the cropped feature region images.

[0074] (22) Calculate the mean brightness of all feature region images according to formula (2), and the specific steps are as follows:

[0075] (221) Traverse and read the brightness at each position of the feature region images after all feature regions are extracted;

[0076] (222) Calculate the mean brightness of all feature region images according to formula (2), that is:

[0077]

[0078] Among them, represents the average brightness of all feature region images at (x, y), N represents the number of feature region images, and I i (x, y) represents the brightness of the i-th feature region image at (x, y).

[0079] (23) Calculate the speckle variance of all feature region images according to formula (3), and the specific steps are as follows:

[0080] (231) Traverse and read the brightness at each position of all feature region images;

[0081] (232) Calculate the speckle variance according to formula (3), extract the living body information of the sample, and remove the structural information therein, that is:

[0082]

[0083] Among them, SV(x, y) represents the speckle variance of the feature region image at (x, y).

[0084] 3) According to the speckle variance of all feature region images at (x, y), perform threshold segmentation on the feature region images, remove the remaining invalid information, calculate the living body probability of the sample, and compare it with the preset threshold to obtain the anti-counterfeiting result of the sample, including the following steps:

[0085] (31) Perform threshold segmentation on the feature region images according to formulas (4) and (5), and the specific steps are as follows:

[0086] (311) Calculate the adaptive threshold according to formula (4), that is:

[0087]

[0088] Among them, c represents a fixed constant, and T represents the adaptive threshold.

[0089] (312) Perform threshold segmentation on the feature region images according to formula (5), remove the interference of the remaining invalid information, and improve the accuracy of living body anti-counterfeiting, that is:

[0090]

[0091] Among them, B(x, y) represents the result of threshold segmentation at the feature region image (x, y).

[0092] (32) Calculate the liveness probability of the current sample according to formula (6), that is:

[0093]

[0094] Among them, SSI represents the liveness probability of the current sample, R represents the width of the feature region image, and C represents the height of the feature region image.

[0095] (33) Compare the liveness probability of the current sample with the preset liveness threshold according to formula (7) to obtain the anti-counterfeiting result of the sample, that is:

[0096]

[0097] Among them, Result represents the anti-counterfeiting result of the current sample. When Result = true, the anti-counterfeiting result is a live body; when Result = false, the anti-counterfeiting result is a non-live body, and P represents the liveness threshold.

[0098] The working principle of the present invention is as follows: The OCT technology is used to collect the sequential B-scan images of the sample. Through the speckle variance technology, the sequential B-scan images can be calculated. In this process, the static structural information such as fingerprints and sweat glands is removed, and the dynamic information such as blood flow carried by the sample is extracted. According to the extracted dynamic information, it can be judged whether the sample is a live body.

[0099] The above embodiments only represent one or several implementation manners of the present invention, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention patent shall be subject to the appended claims.

Claims

1. A finger live anti-counterfeiting method based on speckle variance optical coherence tomography, characterized in that: The finger live anti-counterfeiting method based on speckle variance optical coherence tomography includes: Acquire a time-series B-scan image of the current sample at a specific position by means of a B-scan scanning method; Extract the feature region of each time-series B-scan image to obtain the final feature region of all time-series B-scan images, and crop each time-series B-scan image according to the final feature region to obtain the corresponding feature region image; Calculate the mean brightness of all feature area images, and calculate the speckle variance based on the mean brightness; Perform threshold segmentation on the feature area image according to the speckle variance, and calculate the live probability of the current sample according to the result of the threshold segmentation; Compare the liveness probability of the current sample with the liveness threshold to obtain the anti-counterfeiting result of the current sample; in, The feature region extraction is performed on each time-series B-scan image to obtain the final feature region of all time-series B-scan images, including: Get the length and width of the time-series B-scan image; Calculate the horizontal gradient of each row of pixels in each time-series B-scan image, expressed as follows: Among them, M x represents the horizontal gradient of the xth row of pixels in the current sequential B-scan image, W represents the width of the sequential B-scan image, I(x,y) represents the brightness at the position (x,y), and I(x-1,y) represents the brightness at the position (x-1,y); Traverse each row from top to bottom according to the length of the current time series B-scan image, set the first row with a horizontal gradient greater than a preset threshold as the starting position of the feature area, and after setting the starting position, set the first row with a horizontal gradient less than the preset threshold as the ending position of the feature area, to obtain the feature area of ​​each time series B-scan image; The average value of the feature regions of all time-series B-scan images is calculated to obtain the final feature region.

2. The finger live anti-counterfeiting method based on speckle variance optical coherence tomography according to claim 1, characterized in that: The calculation of the mean brightness of all feature region images includes: Traverse and read all feature area images, calculate the mean brightness of all feature area images, and express it with the following formula: in, represents the average brightness of all feature region images at (x, y), N represents the number of feature region images, and I i (x,y) represents the brightness of the i-th feature region image at (x,y).

3. The finger live anti-counterfeiting method based on speckle variance optical coherence tomography according to claim 2, characterized in that: The calculating of the speckle variance according to the mean value of the brightness includes: Traverse and read all feature area images, and calculate the speckle variance, which can be expressed as follows: Where SV(x,y) represents the speckle variance at the feature region image (x,y).

4. The finger live anti-counterfeiting method based on speckle variance optical coherence tomography according to claim 3 is characterized in that: The step of performing threshold segmentation on the feature area image according to the speckle variance comprises: Calculate the adaptive threshold, expressed as follows: Where c represents a fixed constant and T represents an adaptive threshold; Threshold segmentation is performed based on speckle variance and adaptive threshold, which can be expressed as follows: Among them, B(x,y) represents the result of threshold segmentation at the feature area image (x,y).

5. The finger live anti-counterfeiting method based on speckle variance optical coherence tomography according to claim 4, characterized in that: The live probability of the current sample is calculated based on the result of the threshold segmentation, and is expressed as follows: Among them, SSI represents the live probability of the current sample, R represents the width of the feature area image, and C represents the height of the feature area image.

6. The finger live anti-counterfeiting method based on speckle variance optical coherence tomography according to claim 5, characterized in that: The liveness probability of the current sample is compared with the liveness threshold to obtain the anti-counterfeiting result of the current sample, which is expressed by the following formula: Wherein, Result represents the anti-counterfeiting result of the current sample. When Result=true, the anti-counterfeiting result is live, and when Result=false, the anti-counterfeiting result is non-live. P represents the live threshold.

Images