An adaptive dual-mode identification method for finger vein and external phalangeal pattern

Abstract

The application discloses a self-adaptive dual-mode identity recognition method for finger vein and external finger joint vein, and belongs to the field of biometric recognition technology.The method comprises the following steps: S1, collecting finger vein and external finger joint vein images; S2, pre-processing the finger vein and external finger joint vein images; S3, quality evaluation of the finger vein and external finger joint vein images and giving corresponding scores, and determining whether the images meet the requirements according to the scores; if the images meet the requirements, the next step is performed, otherwise, re-collection is prompted; S4, advantage feature coefficient assignment according to the quality of the corresponding finger vein and external finger joint vein images; S5, using an improved SIFT algorithm for the finger vein image and using a PHASH algorithm for the finger joint vein image to perform similarity matching with features in a database to obtain corresponding similarity; S6, combining the corresponding weight and the similarity through a weighted similarity function to obtain a corresponding weighted matching degree; S7, fusing the finger vein and finger joint vein weighted matching degrees through a fusion function to obtain a final fusion matching degree; and S8, threshold value judgment.The application has high recognition accuracy and robustness.

Description

Technical Field

[0001] This invention relates to biometric technology, and more particularly to an adaptive dual-modal identity recognition method using finger veins and external finger joint prints. Background Technology

[0002] With the rapid development of 5G and the emergence of the "metaverse," various technologies are required to be more intelligent and convenient, which places higher demands on information security and identity authentication. Gradually, more and more scenarios require identity authentication. These scenarios all require accurate and reliable identity authentication to ensure the stable operation of social order. Currently, mainstream single-modal identity recognition technologies are increasingly unable to meet the requirements of accurate and convenient identity authentication in various complex environments. Single-modal identity authentication has limitations for specific groups and is easily counterfeited. Therefore, multimodal biometric recognition technology has emerged. When a specific user's biometric feature is not obvious or defective, single-modal biometric recognition often cannot handle it well, but multimodal biometric recognition technology effectively compensates for the limitations of single-modal recognition technology for specific groups and enhances the reliability and accuracy of recognition.

[0003] Currently, there are a few studies on the fusion recognition of finger veins and knuckle prints both domestically and internationally. Kumar et al. designed a recognition system combining finger veins and knuckle prints. This system uses both infrared cameras and ordinary webcams to acquire images of finger veins and knuckle prints. Unlike other knuckle print images, this system mainly acquires the knuckle print image data closest to the fingertip. In finger vein recognition, it employs a feature extraction algorithm combining Gabor filter enhancement and morphological processing, and finally uses a template matching algorithm to calculate similarity. Domestic scholars have also conducted research on multimodal recognition systems for finger veins and knuckle prints. Kang et al. from SUSTech designed a multispectral biometric system that acquires images of finger veins, fingerprints, and knuckle prints using imaging light sources of different wavelengths and multiple imaging cameras. For finger veins and fingerprints, the LBP algorithm is used to extract features, while for knuckle prints, the ORB algorithm is used for feature extraction, and finally, fusion recognition is performed at the matching value level. Tests were conducted on a self-built dataset, and the equal error rate after multimodal fusion was significantly lower than that of single-modal biometric systems. Experiments have shown that by fusing features of finger veins and knuckle patterns, a higher accuracy rate can be achieved than that of single finger vein recognition.

[0004] The prominence of various biometric features often varies among different individuals. Some individuals have prominent finger veins, making them easily recognizable by algorithms and suitable as primary features for identification; while others have thinner veins with less distinct features, resulting in low accuracy when using finger veins as the primary identification method. The increased number of features necessitates multiple light sources and camera devices, leading to higher instrument costs and larger sizes. Different acquisition modules are required for different features. User verification often requires multiple devices, actions, and attempts to acquire the necessary features, resulting in lengthy waiting times during identification. This reduces the convenience of identification and degrades the user experience, contradicting the portability advantage of biometric technology. Summary of the Invention

[0005] The present invention aims to overcome the above-mentioned problems existing in the prior art and proposes an adaptive dual-modal identity recognition method based on finger vein and external finger joint print.

[0006] Since the features of finger veins and external phalangeal joint prints are both located on the fingers, they are easily acquired using a single device, making acquisition and recognition relatively simple. Therefore, this invention optimizes the combined recognition algorithm for finger veins and external phalangeal joint prints. The "adaptive" nature of this invention is mainly reflected in: assigning corresponding advantageous feature coefficients to the finger veins and external phalangeal joint prints of each individual based on their individual characteristics. That is, dynamically allocating advantageous features based on the relative advantages of different individuals' finger veins and external phalangeal joint prints. Based on the robustness of different enhancement algorithms to certain special groups (the elderly, some groups with unclear vascular imaging), the redundancy of feature extraction under information fusion conditions is automatically reduced to further improve recognition speed and enhance the overall performance of the algorithm.

[0007] An adaptive dual-modal identity recognition method based on finger vein and lateral phalanx prints is disclosed. The method comprises five parts: finger vein and lateral phalanx print image acquisition, image preprocessing, image quality assessment, assignment of dominant feature coefficients, and fusion matching. The specific steps are as follows:

[0008] S1. Use an integrated feature acquisition device for finger veins and external finger joint prints to acquire images of finger veins and external finger joint prints;

[0009] S2. Preprocess the images of digital veins and external phalangeal joint prints;

[0010] S3. Evaluate the quality of the digital vein and external phalangeal joint images and assign corresponding scores. Determine whether the images meet the requirements based on the scores. If they meet the requirements, proceed to the next step; otherwise, prompt for re-acquisition.

[0011] S4. Assign dominant feature coefficients based on the image quality of the corresponding finger veins and external finger joint prints;

[0012] S5. The finger vein image uses the improved SIFT algorithm, and the knuckle print image uses the PHASH algorithm to perform similarity matching with features in the database to obtain the corresponding similarity.

[0013] S6. The final fusion matching degree is obtained by fusing the dominant feature matching degree of finger veins and finger joint prints through a fusion function;

[0014] S7. Perform a threshold check. If the threshold is passed, the identity authentication is successful; otherwise, the identity authentication fails.

[0015] This invention acquires dual-modal biometric features using an integrated feature acquisition device for finger veins and external phalangeal joint patterns. After acquiring the images, it dynamically allocates advantageous features based on the relative prominence of finger veins and external phalangeal joint patterns for each individual, thereby enhancing the robustness of the algorithm, reducing the redundancy of feature extraction, and improving the overall performance of the algorithm.

[0016] The adaptive dual-modal identity recognition method based on finger vein and external finger joint print of the present invention has the following beneficial effects:

[0017] 1. By leveraging adaptive advantage features to highlight the characteristics of finger veins and external finger joint patterns, the robustness and accuracy of finger vein bimodal recognition technology can be effectively improved.

[0018] 2. The bias coefficient can be dynamically adjusted based on the number of times the user has used the finger vein and outer finger joint print recognition. The more times the user uses the device, the more accurate the bias coefficient will be for this user group, which can effectively improve the recognition rate. Attached Figure Description

[0019] To more clearly illustrate the technical method of the present invention, the accompanying drawings used in the system will be briefly introduced below. The following drawings only show some embodiments of the present invention and should not be regarded as limiting the scope. For those skilled in the art, other drawings can be obtained based on the structure of the drawings without creative effort.

[0020] Figure 1 This is a flowchart of the present invention;

[0021] Figure 2 Schematic diagram of the original images of digital veins and external phalangeal joint patterns;

[0022] Figure 3 This is a schematic diagram of an image enhanced by CLAHE in an embodiment of the present invention; Detailed Implementation

[0023] To provide a clearer understanding of the purpose, technical features, and effects of the invention, the implementation method of the invention will be described in detail below with reference to the accompanying drawings.

[0024] like Figure 1 The flowchart shown is an adaptive dual-modal identity recognition method for finger vein and external phalanx prints in an embodiment of the present invention. The overall system structure includes five parts: finger vein and external phalanx print image acquisition, image preprocessing, image quality assessment, assignment of advantageous feature coefficients, and fusion matching. Specific implementation steps include S1 to S7.

[0025] S1. Using an integrated feature acquisition device for finger veins and external phalangeal joint prints, images of finger veins and external phalangeal joint prints are acquired. The acquired images are as follows: Figure 2 As shown;

[0026] S2. Using CLAHE for image preprocessing transforms the histogram distribution of an image into an approximately uniform distribution, thereby enhancing the image contrast. The processed image is as follows: Figure 3 As shown;

[0027] S3. Image quality assessment is performed using a weighted summation method. The specific steps are as follows:

[0028] S31. Calculate the image contrast difference. The calculation process is as follows:

[0029]

[0030] Where: Q1 is the image contrast score, C s Here, n is the standard value, and x is the number of pixels in the image. o x represents the grayscale value of a pixel. m This represents the average grayscale value of the image;

[0031] S32. Calculate the image information entropy. The calculation process is as follows:

[0032]

[0033] Where Q2 represents the information entropy score, and p(x) is the probability of the distribution of the x-th gray-level pixel in the finger vein or external finger joint print image;

[0034] S33. Calculate the image position offset. The calculation process is as follows:

[0035]

[0036] Where Q3 represents the image position offset, D represents the total number of pixels in the image region, y[i] is the ordinate of the i-th pixel in the image region D, and x[i] is the abscissa of the i-th pixel in the image region D;

[0037] The overall image score Q is obtained through a weighted summation function. If Q meets the requirements, the image acquisition is successful and the next step is initiated; otherwise, a re-acquisition prompt is given. The formula for the weighted summation function is as follows:

[0038]

[0039] S4. Assign dominant feature coefficients based on the image quality of the corresponding finger veins and external phalangeal joint prints; the formula for the dominant feature coefficient α is as follows:

[0040]

[0041] S5. The finger vein image uses the improved SIFT algorithm, and the knuckle print image uses the PHASH algorithm to perform similarity matching with features in the database to obtain the corresponding similarity.

[0042] S6. The final fusion matching degree F is obtained by fusing the dominant feature matching degrees of finger veins and finger joint prints using a fusion function; the formula is as follows:

[0043]

[0044] Where S1 is the finger vein similarity obtained by the SIFT algorithm, and S2 is the finger joint pattern similarity obtained by the PHASH algorithm. Because the information content in finger veins is richer than that in external finger joint patterns, when the difference between the overall quality scores of the finger vein image and the external finger joint pattern is less than 10, a bias coefficient β1 is assigned to the finger vein, which can effectively improve the recognition rate. In this invention, β1 = 1.07, and the formula for β1 is as follows:

[0045]

[0046] Among them, A V K indicates the number of times the finger vein was successfully identified with emphasis. V Indicates the total number of finger vein recognition attempts with a preference for other methods; A F K indicates the number of successful recognitions of the outer finger joint print. F This indicates the total number of times the outer finger joint lines are identified.

[0047] S7. After obtaining the fusion matching degree F, a threshold judgment is performed. If the threshold is passed, the matching is successful; otherwise, the matching fails.

[0048] The embodiments described in this specification are merely examples of implementations of the inventive concept. The scope of protection of this invention should not be considered as limited to the specific forms stated in the embodiments. The scope of protection of this invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. An adaptive dual-modal biometric recognition method for finger veins and external phalangeal joint prints, comprising the following steps: S1. Use an integrated feature acquisition device for finger veins and external finger joint prints to acquire images of finger veins and external finger joint prints; S2. Preprocess the images of digital veins and external phalangeal joint prints; S3. Evaluate the quality of the digital vein and external phalangeal joint images and assign corresponding scores. Determine whether the images meet the requirements based on the scores. If they meet the requirements, proceed to the next step; otherwise, prompt for re-acquisition. S4. Determine the overall score of the finger vein image. Combined score with outer finger joint print image The size of the feature coefficient is used to assign a dominant feature coefficient to the feature with a higher overall image score. Dominance Feature Coefficient The formula is as follows: ⑸ S5. The finger vein image uses the improved SIFT algorithm, and the knuckle print image uses the PHASH algorithm to perform similarity matching with features in the database to obtain the corresponding similarity. S6. The final fusion matching degree is obtained by fusing the dominant feature matching degrees of finger veins and knuckle prints using a fusion function. The formula is as follows: ⑹ in, The finger vein similarity is obtained from the SIFT algorithm. The similarity score of the finger veins is obtained from the PHASH algorithm. Because the information content in the finger veins is richer than that in the outer finger veins, a bias coefficient is assigned to the finger veins when the difference between the overall quality score of the finger vein image and the overall quality score of the outer finger vein is less than 10. , The system dynamically adjusts the recognition rate based on the number of successful recognitions of the user's finger vein emphasis and external finger joint print emphasis. The more times a user uses the system, the more accurate the emphasis coefficient becomes for this user group, which can effectively improve the recognition rate. The formula is as follows: ⑺ in, This indicates the number of times the finger vein was successfully identified with a bias. This indicates the total number of times the finger vein recognition was emphasized; This indicates the number of times the outer finger joint print was successfully recognized. This indicates the total number of times the outer finger joint lines were identified; S7. Perform a threshold check. If the threshold is passed, the identity authentication is successful; otherwise, the identity authentication fails.

2. The adaptive dual-modal biometric recognition method for finger veins and external phalangeal joint prints according to claim 1, characterized in that, In step S2, CLAHE is used for image preprocessing to transform the histogram distribution of an image into an approximately uniform distribution, thereby enhancing the image contrast.

3. The adaptive dual-modal biometric recognition method for finger veins and external phalangeal joint prints according to claim 1, characterized in that, In step S3, image quality is assessed using a weighted summation method. The quality scores are accumulated according to set weights and used as feedback parameters to assess the image quality of finger veins and external phalangeal joint prints. The specific steps are as follows: S31. Calculate the image contrast difference. The calculation process is as follows: ⑴ in: Image contrast score. For standard values, The number of pixels in the image. This represents the grayscale value of a pixel. This represents the average grayscale value of the image; S32. Calculate the image information entropy. The calculation process is as follows: ⑵ in The information entropy score represents the information entropy. For images of finger veins or external phalangeal joints The probability of pixel distribution at each gray level; S33. Calculate the image position offset. The calculation process is as follows: ⑶ in This indicates the image position offset. This represents the total number of pixels in the image region. For image region The Middle The ordinate of each pixel. For image region The Middle The x-coordinate of each pixel; S34. After obtaining the above evaluation parameters for the finger vein and lateral phalanx print images, the corresponding image comprehensive score is obtained by summing the values ​​using a weighted summation function. The formula is as follows: ⑷ Final image The larger it is, the higher its quality; S35, the obtained This means comparing the image with a set threshold; if the condition is met, proceed to the next step; otherwise, re-acquire the image.

4. The adaptive dual-modal biometric recognition method for finger veins and external phalangeal joint prints according to claim 1, characterized in that, In step S5, the similarity matching of finger veins uses the improved SIFT algorithm, and the similarity of the knuckle print image is obtained using the PHASH algorithm.

Images