Palm vein image quality evaluation method, device and equipment and storage medium

Abstract

The application provides a palm vein image quality evaluation method and device, equipment and a storage medium, relating to the technical field of image processing. The method comprises: acquiring a palm vein image, positioning key points on the palm vein image to obtain a corresponding vein area image; acquiring an evaluation score corresponding to a preset evaluation parameter item of the vein area image, the evaluation parameter item including posture, area size, illumination, definition and integrity; determining a comprehensive score of the vein area image based on a preset score quantization formula and the evaluation score, and taking the palm vein image corresponding to the vein area image with a comprehensive score meeting a preset score as a qualified palm vein image. The scheme can select a palm vein image with excellent quality through quality evaluation of the palm vein image, so that the stability of the system is improved.

Description

Technical Field

[0001] This application relates to the field of image processing technology, and in particular to a method, apparatus, device and storage medium for evaluating the quality of palm vein images. Background Technology

[0002] Palm vein recognition uses the intensity of reflected near-infrared light to identify the location of veins, and then compares the read vein data with pre-stored palm vein data to perform identification. Due to its non-contact identification process, near-infrared palm vein recognition has been increasingly widely used as an effective biometric identification technology in recent years.

[0003] During palm vein recognition, quality variations can easily occur due to differences in imaging equipment and background environment, resulting in problems such as low resolution, uneven illumination, blurred images, and occlusion of vein features, severely affecting the stability of the palm vein identification system. Therefore, quality assessment should be performed before palm vein identification to filter out low-quality palm vein images and improve the stability of the recognition system. However, the quality assessment schemes for palm vein images in related technologies are not perfect, making it difficult to effectively filter out low-quality palm vein images, leading to low recognition accuracy and poor system stability. Summary of the Invention

[0004] This application provides a method, apparatus, device, and storage medium for evaluating the quality of palm vein images, which solves the problem of effectively filtering out low-quality palm vein images. By evaluating the quality of palm vein images, high-quality palm vein images can be selected, thereby improving the stability of the system.

[0005] In a first aspect, embodiments of this application provide a method for assessing the quality of palm vein images, including:

[0006] Acquire palm vein images and locate key points on the palm vein images to obtain corresponding vein region images;

[0007] The acquired vein region image corresponds to the evaluation score of preset evaluation parameters, including region size, pose, illumination, sharpness, and integrity.

[0008] Based on a preset scoring quantification formula and evaluation score, a comprehensive score for the vein region image is determined, and the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score is considered a qualified palm vein image.

[0009] Secondly, embodiments of this application provide a palm vein image quality assessment device, comprising:

[0010] The image acquisition module is configured to acquire palm vein images and locate key points on the palm vein images to obtain corresponding vein region images;

[0011] The score acquisition module is configured to acquire the evaluation score of the vein region image corresponding to the preset evaluation parameters, which include region size, pose, illumination, sharpness and integrity.

[0012] The image determination module is configured to determine the comprehensive score of the vein region image based on a preset scoring quantification formula and evaluation score, and to select the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score as the qualified palm vein image.

[0013] Thirdly, embodiments of this application provide an electronic device, including:

[0014] One or more processors;

[0015] Storage device for storing one or more programs.

[0016] When one or more programs are executed by one or more processors, the one or more processors implement the palm vein image quality assessment method of the embodiments of this application.

[0017] Fourthly, embodiments of this application also provide a storage medium for storing computer-executable instructions, which, when executed by a processor, are used to perform the palm vein image quality assessment method of embodiments of this application.

[0018] This application acquires vein region images by locating key points on palm vein images and obtains corresponding evaluation scores for multiple preset evaluation parameters using appropriate quantification methods. By covering multiple factors affecting palm vein image quality, such as region size, pose, illumination, sharpness, and integrity, and using these as evaluation parameters, the overall image score can be determined by calculating the evaluation scores of each parameter based on a preset quantification formula. This quantifies the image quality, enabling image quality assessment and more efficiently determining image quality, thus allowing for the removal of low-quality images, further improving palm vein recognition accuracy, and enhancing system stability. Attached Figure Description

[0019] Figure 1 A flowchart of the palm vein image quality assessment method provided in the embodiments of this application;

[0020] Figure 2 A flowchart for acquiring images of vein regions provided in an embodiment of this application;

[0021] Figure 3 A schematic diagram illustrating the principle of identifying key points provided in this application embodiment;

[0022] Figure 4 A flowchart for determining a first evaluation score corresponding to the size of the region provided in this application embodiment;

[0023] Figure 5 A flowchart for determining a second evaluation score corresponding to an attitude is provided for embodiments of this application;

[0024] Figure 6 A principle block diagram for determining the degree of tilt provided in an embodiment of this application;

[0025] Figure 7 A flowchart for determining a third evaluation score corresponding to illumination, provided for embodiments of this application;

[0026] Figure 8 A block diagram illustrating the principle of determining illumination uniformity as provided in an embodiment of this application;

[0027] Figure 9 A flowchart for determining a fourth evaluation score corresponding to sharpness, provided for embodiments of this application;

[0028] Figure 10 A flowchart for determining a fifth evaluation score corresponding to completeness, provided for embodiments of this application;

[0029] Figure 11 A schematic diagram of three images under different actions provided in the embodiments of this application;

[0030] Figure 12 A schematic diagram of a palm vein image quality assessment device provided in an embodiment of this application;

[0031] Figure 13 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0032] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the embodiments described herein are for illustrative purposes only and not for limiting the scope of the application. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present application are shown in the accompanying drawings, not the entire structure.

[0033] It should be noted that, due to space limitations, this application specification does not exhaustively list all possible implementation methods. Those skilled in the art should be able to conceive after reading this application specification that, as long as the technical features do not contradict each other, any combination of technical features can constitute an optional implementation method.

[0034] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity, operation, or object from another, and do not necessarily require or imply any actual relationship or order between these entities, operations, or objects. Furthermore, the number of objects distinguished by "first," "second," etc., is not limited; there can be one or more. It is conceivable that "multiple" in the description of this application means two or more. In addition, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0035] The palm vein image quality assessment method provided in this application can be applied to electronic devices with palm vein image recognition capabilities, such as palm vein recognition devices and other biometric devices, to effectively screen images before palm vein recognition. The application of this palm vein image quality assessment method in electronic devices enables the evaluation of image quality, facilitating the removal of low-quality palm vein images and thereby improving the stability of the recognition system.

[0036] Figure 1 The flowchart of the palm vein image quality assessment method provided in this application embodiment is shown in the figure. The method specifically includes the following steps:

[0037] Step S110: Obtain a palm vein image and locate key points on the palm vein image to obtain the corresponding vein region image.

[0038] Taking the application of the palm vein image quality assessment method to the aforementioned electronic device as an example, the palm vein image can be captured by a camera installed on the electronic device. It should be noted that the captured image of the user's palm can be used as the palm vein image for quality assessment. It should be understood that when the palm vein image meets the image quality requirements (i.e., is a qualified palm vein image), the electronic device can initiate near-infrared scanning to acquire the user's vein data. For example, the electronic device can acquire an image of the user's palm using an infrared CCD (Charge Coupled Device) camera, and then perform vein feature extraction for palm vein recognition. Alternatively, it is conceivable that testers can input pre-captured images into the electronic device.

[0039] Of course, after obtaining the palm vein image, it is necessary to process the palm vein image to determine the key points on it, that is, to locate the key points on the obtained image so as to extract the vein area image.

[0040] In one embodiment, for acquiring an image of a vein region, refer to Figure 2 , Figure 2The flowchart for acquiring vein region images provided in this application embodiment is shown in the figure. The palm vein image quality assessment method provided in this application further includes the following steps:

[0041] Step S210: Based on the Openpose algorithm, determine the location of key points in the palm vein image.

[0042] Step S220: Based on the location of the key points, determine the vein region on the palm vein image and crop the vein region to obtain a vein region image.

[0043] The acquired palm vein image can be input into a pre-defined recognition network, such as an Openpose network built based on the Openpose algorithm. This network identifies key points in the image, such as... Figure 3 As shown, Figure 3 The diagram below illustrates the principle of key point recognition in an embodiment of this application. The palm vein image is used as the input image of the Openpose network. Key points in the image are identified, such as the center point of the root of the five fingers, the center point of the palm, and the center point of the root of the palm in the palm vein image. The key points can also be marked in the image to determine their positions in the palm vein image.

[0044] After determining the corresponding key points, the vein region on the palm vein image can be determined accordingly. For example, the center point of the palm can be used as the center point of the vein region, and the distance between the center point of the base of the middle finger and the center point of the base of the palm can be used as the side length, thus selecting a regular rectangular area as the vein region; or the center point of the palm can be used as the center point of the vein region, and the lines connecting the other key points to adjacent key points can be used as the boundary lines, thus selecting an irregular area as the vein region. Therefore, the determined vein region is cropped from the palm vein image to obtain the corresponding vein region image.

[0045] By locating key points on the image, not only can the vein region be identified, but also the tilt of the current palm relative to the imaging plane can be determined, which helps in determining the evaluation scores of the vein region image corresponding to various evaluation parameters.

[0046] It should be noted that in some embodiments, key point detection can also be performed using Mediapipe. Mediapipe is a framework mainly used to construct multimodal audio, video, or any time-series data. It supports detection in many scenarios, such as face detection, facial mesh, gesture recognition (providing the coordinate information of key points on it), partial pose recognition, human pose estimation, target detection and tracking, etc.

[0047] Step S120: Obtain the evaluation score of the vein region image corresponding to the preset evaluation parameter item.

[0048] Once the vein region image is determined, multiple evaluations are performed on the vein region image. The preset evaluation parameters include region size, pose, illumination, sharpness, and integrity. For each evaluation parameter, the score of the vein region image in each evaluation parameter can be calculated according to the preset quantification method for each evaluation parameter, i.e., the evaluation score.

[0049] For the evaluation parameter item, which is the area size, corresponding to the size of the vein area, it can be understood that the larger the area of ​​the vein area, the higher the evaluation score for this item.

[0050] For the evaluation parameter of pose, the pose of the hand in the image can be determined by calculating the tilt of the palm plane relative to the imaging plane in the current palm vein image, thus calculating a corresponding score for that pose. It is understood that the imaging plane is the plane where the camera captures the image; the vein region is concentrated in the center of the palm, meaning the vein region is also within the palm plane. Pose affects palm vein recognition. When the user's palm pose does not meet the recognition requirements, such as excessive tilt leading to difficulty in extracting effective vein features or extracting few vein features, the device cannot recognize the image. Therefore, when evaluating the quality of palm vein images, pose is included as one of the evaluation parameters, which helps to filter out low-quality images.

[0051] For the evaluation parameter of illumination, illumination affects the brightness and darkness of the image. It's conceivable that the stronger the skin's absorption of light, the weaker the light returning to the skin surface through backscattering; conversely, the greater the absorption of near-infrared light by veins in the skin, the greater the difference in overall light intensity on the skin surface. Therefore, if the illumination intensity is too low, the difference between skin and veins in the image will be small, making effective differentiation difficult, and increasing the difficulty of extracting vein features, thus affecting the accuracy of palm vein recognition.

[0052] For the evaluation parameter of sharpness, the presence of blur in the image can be used to characterize its sharpness. It's understood that in images, sharpness and blurriness are relative; the sharper the image, the higher its quality, the greater the sharpness, and the smaller the blurriness. Conversely, the less sharp (blurred) the image, the lower its quality, the smaller the sharpness, and the greater the blurriness. Therefore, when describing the sharpness of an image, both sharpness and blurriness can be used, but the two indicators are inversely proportional. Blurripping in an image can be due to motion blur or defocus blur. For blurred images, especially when the vein region in a palm vein image is blurred, the difficulty of identifying the extracted vein features increases, and the accuracy decreases. Therefore, in the quality assessment of palm vein images, sharpness needs to be evaluated to filter out images with low sharpness.

[0053] For the evaluation parameter item of integrity, it can be calculated whether the target area in the image is occluded, such as whether the vein area is occluded. It should be noted that since palm vein recognition is a non-contact biometric recognition method, there is a gap between the palm and the electronic device when the user uses the above-mentioned electronic device for palm vein recognition. Obstacles such as hair and jewelry can easily fall into the gap, resulting in the image being occluded. If the vein area is occluded, the vein features are difficult to extract effectively, thus affecting the recognition accuracy.

[0054] Therefore, in the process of quality assessment of palm vein images, it is necessary to process the above-mentioned assessment parameters accordingly to obtain the assessment score for each item, so as to effectively assess the image quality and facilitate image screening.

[0055] Step S130: Based on the preset scoring quantification formula and evaluation score, determine the comprehensive score of the vein region image, and take the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score as the qualified palm vein image.

[0056] After determining the evaluation score for each image corresponding to the aforementioned evaluation parameters, the electronic device can use a preset scoring formula to determine the overall score of the vein region image. This overall score is then compared to a preset score. If the overall score of the vein region image is greater than the preset score, meaning the overall score meets the preset score, then the corresponding palm vein image is a qualified palm vein image. A qualified palm vein image indicates that the image quality is acceptable and has passed the quality assessment, requiring no rejection. It should be noted that the preset score can be set according to the image quality requirements; a higher preset score corresponds to a higher quality qualified palm vein image.

[0057] It should be considered that images of vein regions whose overall scores do not meet the preset threshold are determined to be low-quality images, and electronic devices need to filter them out. For example, in practical applications, after evaluating the quality of the captured images, if the electronic device determines that an image is low-quality, it outputs a prompt message to the user indicating that the current image quality is poor and there is a risk of it being unrecognizable. This prompt message can be delivered to the user through methods such as voice broadcast or display on a monitor.

[0058] As can be seen from the above scheme, this application evaluates each evaluation parameter item of the vein region image and obtains the corresponding evaluation score to quantify each evaluation parameter item, thereby obtaining a comprehensive score representing the image quality, so as to determine whether the image quality is qualified, so that low-quality images can be screened out before vein recognition, further improving the palm vein recognition accuracy and effectively improving the system stability.

[0059] Figure 4 The flowchart for determining the first evaluation score corresponding to the region size provided in the embodiments of this application is shown in the figure. In one embodiment, the determination of the first evaluation score for the evaluation parameter item of region size can be carried out using the following steps:

[0060] Step S410: Obtain the image area of ​​the vein region.

[0061] Step S420: Determine the area score associated with the vein region image based on the image area and the preset area.

[0062] Step S430: If the area score is greater than the first threshold and less than or equal to the second threshold, the area score is used as the first evaluation score of the static region image corresponding to the region size.

[0063] In the process of determining the first evaluation score, after the vein region is determined, such as by determining the vein region using the key points mentioned above, and thus obtaining the vein region image, it is also necessary to calculate the size of the vein region, that is, the image area of ​​the vein region image, such as by representing the image area by the number of pixels on the image.

[0064] It can be inferred that the first evaluation score is a value with upper and lower limits, such as the first threshold being its lower limit and the second threshold being its upper limit. Therefore, when the area score is greater than the first threshold and less than or equal to the second threshold, the area score is the first evaluation score. The area score is determined based on the image area and the preset area. The preset area is the minimum area in which the image can be effectively recognized. When the image area is less than or equal to the preset area, the corresponding first evaluation score is 0.

[0065] Specifically, the first assessment score can be determined using the following formula. A :

[0066]

[0067] Where S represents the image area corresponding to the vein region, and the area score is calculated using... The calculation is performed using S1 = 100 as the preset area value in the above formula. As can be seen from the formula, when the image area S is less than or equal to the preset area S1, the corresponding area score is less than or equal to 0. Moreover, the first threshold is 0 and the second threshold is 100. Therefore, when selecting the maximum value between the first threshold and the area score, the first threshold should be taken. Thus, when selecting the minimum value between the first threshold and the second threshold, the first threshold, i.e., 0, should be taken. Since two selections are required, the first evaluation score will be limited to between the first threshold and the second threshold. That is, when the area score is between the first threshold and the second threshold, the area score can be used as the first evaluation score.

[0068] The size of the vein region affects the palm vein information. For the evaluation parameter of region size, a corresponding preset threshold corresponding to the area is set so that when the image area corresponding to the vein region is less than the threshold, the first evaluation score is set to zero, indicating that the image quality is difficult to meet the requirements. That is, by quantifying the corresponding value, the image quality can be effectively and intuitively represented, which helps in the screening of images.

[0069] Figure 5 The flowchart for determining the second evaluation score corresponding to the attitude, provided for an embodiment of this application, is shown in the figure. In one embodiment, the determination of the second evaluation score for the attitude evaluation parameter can be performed using the following steps:

[0070] Step S510: Determine the horizontal rotation angle and pitch angle between the palm plane and the imaging plane corresponding to the vein region image.

[0071] Step S520: Based on the horizontal rotation angle and pitch angle, obtain the first attitude score value corresponding to the horizontal rotation angle and the second attitude score value corresponding to the pitch angle.

[0072] Step S530: Based on the first posture score and the second posture score, obtain the second evaluation score of the vein region image corresponding to the posture.

[0073] The degree of tilt between the palm plane and the imaging plane is determined by the horizontal rotation angle and the pitch angle. For the detection of the horizontal rotation angle and the pitch angle, refer to... Figure 6 The diagram shown illustrates the principle of determining the degree of tilt. After inputting the image into the Openpose network and determining the key points on the image, the pose of the hand in the image is estimated. For example, the degree of tilt between the hand plane and the imaging plane is determined based on the key points, expressed as horizontal rotation angle and pitch angle.

[0074] Understandably, pose estimation is a representation of an object's orientation in three-dimensional space, typically described using the Earth as a reference frame (standard coordinate system). When converting from the Global Coordinate System (GCS) to the Static Coordinate System (SCS), the angles the object rotates around the three axes of the SCS are the three-dimensional pose angles: the yaw angle (Z-axis), the pitch angle (Y-axis), and the roll angle (X-axis). When the palm plane is parallel to the imaging plane (i.e., parallel to the YOZ plane), the roll angle has no impact on the evaluation, and the quality assessment results are the same. Therefore, the corresponding yaw and pitch angles are chosen as indicators to represent the degree of tilt between the palm plane and the imaging plane.

[0075] For both horizontal rotation and pitch angles, corresponding attitude scores are calculated, such as a first attitude score for the horizontal rotation angle and a second attitude score for the pitch angle. The smaller the absolute values ​​of the horizontal rotation and pitch angles, the higher the second evaluation score, and the better the image quality. The second attitude evaluation score can be obtained from the first and second attitude scores.

[0076] Specifically, the first posture score yaw It can be calculated using the following formula:

[0077]

[0078] The second posture score pitch It can be calculated using the following formula:

[0079]

[0080] Second assessment score p It can be calculated using the following formula:

[0081]

[0082] Where, θ y The angle value θ represents the horizontal rotation angle. p The pitch angle is represented by the angle value. After determining the angle value, the corresponding first attitude score and second attitude score can be calculated according to the above formula. Finally, based on the calculation formula of the second evaluation score, the score of the evaluation parameter item of the corresponding attitude is determined.

[0083] Therefore, the tilt of the image is represented by the horizontal rotation angle and pitch angle between the palm plane and the imaging plane. The attitude is determined and quantified to obtain the corresponding score, such as the first attitude evaluation score corresponding to the horizontal rotation angle and the second attitude evaluation score corresponding to the pitch angle. Thus, the evaluation score related to the attitude evaluation parameter is calculated, thereby effectively realizing the quantification of attitude and contributing to the evaluation of image quality.

[0084] Figure 7 A flowchart for determining a third evaluation score corresponding to illumination, provided for embodiments of this application, is shown below. Figure 7 As shown, the palm vein image quality assessment method also includes the following steps:

[0085] Step S710: Obtain the proportion of pixels with brightness within a preset brightness range in the vein region image, and use the proportion as the pixel ratio.

[0086] Step S720: Obtain the similarity of the segmented regions divided by the preset segmentation lines on the vein region image, and use the similarity as the illumination uniformity.

[0087] Step S730: Determine the third evaluation score of the illumination corresponding to the vein region image based on the pixel ratio and illumination uniformity.

[0088] Understandably, the quality assessment of illumination as an evaluation parameter is based on pixel ratio and illumination uniformity, with a preset brightness range of [80, 150]. For grayscale images, brightness can be represented by grayscale values ​​ranging from [0, 255]. The number of pixels in the vein region image that satisfy the preset brightness range is counted, thus determining the corresponding proportion, i.e., the pixel ratio.

[0089] For obtaining illumination uniformity, you can refer to the following: Figure 8 The diagram illustrates the principle of determining illumination uniformity. For the extracted vein region image, vertical segmentation can be used, dividing the image into left and right halves along a vertical dividing line, as shown in the upper part of the diagram. The similarity (e.g., left-right similarity) between the two halves is calculated using methods such as MD5 or histograms. Similarly, for horizontal segmentation, the image is divided into upper and lower halves along a horizontal dividing line, and the corresponding similarity (e.g., upper-lower similarity) is calculated in the same way. The impact of illumination on image quality is determined by quantifying pixel ratios and illumination uniformity, resulting in a corresponding evaluation score for image quality assessment.

[0090] It is worth noting that in determining the uniformity of illumination, left-right similarity and / or top-bottom similarity can be selected. When left-right similarity and top-bottom similarity are selected, corresponding weights are set for the two similarities to finally determine the uniformity of illumination.

[0091] Specifically, for the third assessment score I The following formula can be used for calculation:

[0092]

[0093] Among them, Q N Q represents the pixel ratio. L This indicates the uniformity of illumination.

[0094] Figure 9 The flowchart for determining the fourth evaluation score corresponding to sharpness provided in the embodiments of this application is shown in the figure. The palm vein image quality evaluation method further includes the following steps:

[0095] Step S910: Determine the degree of blur of each pixel in the vein region image according to the blur detection algorithm.

[0096] Step S920: Normalize the blur level to determine the sharpness score corresponding to each pixel.

[0097] Step S930: Based on the sharpness score of each pixel, determine the fourth evaluation score of the sharpness corresponding to the vein region image.

[0098] It is understood that the specific blur detection algorithm can refer to the technical solutions described in existing literature, such as blur detection based on the Laplacian operator, which will not be elaborated in the embodiments of this application. Correspondingly, δ(i,j) represents the normalized sharpness score. For example, δ(i,j) = 0 indicates that the pixel is sharp and there is no blur or other interference, while δ(i,j) = 1 indicates that the pixel is blurry and may have motion blur or out-of-focus blur.

[0099] Specifically, for the fourth evaluation value, score S The calculation can be performed using the following formula:

[0100]

[0101] Where L represents the number of pixels in different directions, the fourth evaluation value of the corresponding sharpness is determined by accumulating the sharpness scores of all pixels in the captured vein region image.

[0102] The blur level of each pixel is quantified to obtain the sharpness score corresponding to each pixel in the current image. By accumulating the sharpness scores of each pixel, the sharpness of the entire image can be determined, thus realizing the quantification of image sharpness for subsequent image quality assessment.

[0103] Figure 10 The flowchart for determining the fifth evaluation score corresponding to integrity, provided for embodiments of this application, is shown in the figure. The palm vein image quality assessment method further includes the following steps:

[0104] Step S1010: Determine the degree of occlusion of each pixel in the vein region image according to the occlusion detection algorithm.

[0105] Step S1020: Normalize the degree of occlusion to determine the integrity score corresponding to each pixel.

[0106] Step S1030: Based on the integrity score of each pixel, determine the fifth evaluation score of the integrity of the vein region image.

[0107] Understandably, assessing the integrity of vein region pixels can be done by determining whether pixels are occluded. A higher occlusion percentage results in a lower integrity score, while a lower occlusion percentage results in a higher score. This applies to occlusion detection algorithms.

[0108] For specific occlusion detection algorithms, please refer to the technical solutions described in existing literature, such as classifying and locating the occluded parts based on the SSD (SingleShot MultiBox Detector) algorithm to achieve the detection of the occluded parts. This will not be elaborated in the embodiments of this application.

[0109] After determining the degree of occlusion of each pixel, it is normalized. For example, ο(i,j) represents the output result of the algorithm, i.e., the integrity score. ο(i,j) = 0 indicates that the pixel is complete and there is no occlusion or other interference, while ο(i,j) = 1 indicates that the pixel is not visible.

[0110] Specifically, for the fifth evaluation value, score O The calculation can be performed using the following formula:

[0111]

[0112] Where L represents the number of pixels in different directions, the fifth evaluation value of the corresponding sharpness is determined by accumulating the integrity scores of all pixels in the captured vein region image.

[0113] The degree of occlusion of pixels is quantified to obtain the integrity score corresponding to each pixel in the current image. By accumulating the integrity scores of each pixel, the integrity of the entire image can be determined, thus realizing the quantification of image integrity for subsequent image quality assessment.

[0114] In one embodiment, after obtaining the evaluation scores of each evaluation parameter item, the comprehensive score is determined based on the above evaluation scores, so as to realize the evaluation of image quality by combining the size, posture, illumination, sharpness and integrity of the palm vein area, and comprehensively consider the influence of each parameter on the image.

[0115] On the one hand, the size and integrity of the palm vein region determine the amount and completeness of the acquired palm vein information. On the other hand, pose, illumination, and sharpness reflect different aspects of the palm vein image quality and affect the extraction of vein features. Therefore, region size and integrity are given the same weight and are independent of pose, illumination, and sharpness. That is, pose, illumination, and sharpness are calculated in a weighted manner to obtain a score, such as the sixth evaluation score.

[0116] Based on preset weights, the second evaluation score corresponding to posture, the third evaluation score corresponding to illumination, and the fourth evaluation score corresponding to sharpness of the vein region image are configured. That is, different evaluation scores are configured with different weights, and a weighted calculation is performed to obtain the sixth evaluation score. It should be noted that the sum of the weights configured for the second, third, and fourth evaluation scores is 1.

[0117] The minimum value among the first assessment score corresponding to the size of the vein region image, the fifth assessment score corresponding to the integrity, and the sixth assessment score is selected as the comprehensive score.

[0118] Specifically, the overall score can be determined using the following score quantification formula:

[0119] score = min(score) A ,(score P ×a1+score S ×a2+score I ×a3), score O )

[0120] Here, a1, a2, and a3 are the weights corresponding to different evaluation parameters, and their values ​​range from 0 to 1. It should be noted that the specific values ​​can be adjusted according to the actual application scenario, for example, a1 = 0.4, a2 = 0.2, and a3 = 0.4.

[0121] Using the above calculation formula, the minimum value among the three scores is selected as the comprehensive score. In addition, for images with a comprehensive score greater than the preset score, such as a comprehensive score greater than 80, the electronic device determines that it is a qualified palm vein image and the user performs vein feature extraction.

[0122] For example, Figure 11 The image shows three images of the same user under different actions (Figures 11a, 11b, and 11c). Corresponding palm vein images were acquired under each action, and the image quality was evaluated using the palm vein quality assessment method described above. The results are shown in the table below.

[0123] Sample Area size attitude illumination Clarity Integrity Overall Score Sample I 99 95 90 95 100 94 Sample II 72 93 93 81 100 72 Sample III 95 68 75 83 54 54

[0124] Among them, sample I is associated with Figure 11 a, Sample II is associated with Figure 11 b, Sample III is associated with Figure 11 c. The values ​​in the table are the evaluation scores for the corresponding evaluation parameters. As shown in the table, if the preset score is set to 80, then the image of Sample I is a qualified palm vein image, and the comprehensive scores of the other images are all less than 80, thus they are screened out.

[0125] Therefore, a comparative experiment was conducted on palm vein recognition with and without quality assessment. First, 26,125 palm vein images from 238 IDs were collected as the test set and tested in a 1:1 ratio. After quality assessment using the aforementioned palm vein quality assessment method, 1,082 images that failed to meet quality standards were filtered out, retaining 25,043 test images. The test model used MobileNetv2 as the baseline network, which was trained using 11,488 classes and 1.23 million data points. The test results of the comparative experiment are shown in the table below, which tests the false rejection rate (FRR) of the algorithms with and without quality screening under different false acceptance rates (FAR):

[0126] Data / Error Reception Rate 0.1％ 0.01％ 0.001％ 0.0001％ No quality screening 1.46％ 2.29％ 3.2％ 10.9％ Quality screening 0.34％ 0.69％ 1.1％ 1.68％

[0127] The results above show that after quality assessment and screening, the algorithm significantly improved the false rejection rate under the same false acceptance rate. In particular, as the false acceptance rate decreased, the impact of low-quality samples on the recognition results became more obvious. For example, when the false acceptance rate was one in a million (i.e., 0.0001%), the false rejection rate without quality screening was 10.9%, while the false rejection rate with quality screening was only 1.68%.

[0128] Therefore, it can be seen that in the same palm vein recognition network, whether or not there is quality assessment and low-quality image filtering has a significant impact on recognition. That is, by applying the palm vein image quality assessment method of this application to filter input images, the false rejection rate can be significantly improved, and the stability of the system can also be effectively improved.

[0129] Figure 12 This is a schematic diagram of a palm vein image quality assessment device provided in an embodiment of this application. The device is used to execute the palm vein image quality assessment method provided in this embodiment, and specifically performs the corresponding functional modules and beneficial effects of the method. As shown in the figure, the palm vein image quality assessment device includes:

[0130] The image acquisition module 1201 is configured to acquire a palm vein image and locate key points on the palm vein image to obtain the corresponding vein region image;

[0131] The score acquisition module 1202 is configured to acquire the evaluation score of the vein region image corresponding to the preset evaluation parameters, including region size, pose, illumination, sharpness and integrity.

[0132] The image determination module 1203 is configured to determine the comprehensive score of the vein region image based on a preset scoring quantification formula and evaluation score, and to take the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score as a qualified palm vein image.

[0133] Based on the above embodiments, the image acquisition module 1201 is further configured as follows:

[0134] Based on the Openpose algorithm, the location of key points in the palm vein image is determined;

[0135] Based on the location of key points, the vein region on the palm vein image is determined, and the vein region is cropped to obtain the vein region image.

[0136] Based on the above embodiments, the image acquisition module 1201 is further configured such that the key points include the center point of the root of the five fingers, the center point of the palm, and the center point of the root of the palm in the palm vein image.

[0137] Based on the above embodiments, the score acquisition module 1202 is further configured as follows:

[0138] Obtain the image area of ​​the vein region;

[0139] Based on the image area and the preset area, determine the area score associated with the vein region image;

[0140] If the area score is greater than the first threshold and less than or equal to the second threshold, the area score is used as the first evaluation score for the vein region image corresponding to the region size.

[0141] If the image area is less than or equal to the preset area, the first evaluation score is 0.

[0142] Based on the above embodiments, the score acquisition module 1202 is further configured as follows:

[0143] Determine the horizontal rotation angle and pitch angle between the palm plane and the imaging plane corresponding to the vein region image;

[0144] Based on the horizontal rotation angle and pitch angle, obtain the first attitude score corresponding to the horizontal rotation angle and the second attitude score corresponding to the pitch angle;

[0145] Based on the first posture score and the second posture score, a second evaluation score corresponding to the posture is obtained from the vein region image.

[0146] Based on the above embodiments, the score acquisition module 1202 is further configured as follows:

[0147] Obtain the proportion of pixels with brightness within a preset brightness range in the vein region image, and use the proportion as the pixel ratio;

[0148] Obtain the similarity of segmented regions divided by preset segmentation lines on the vein region image, and use the similarity as the illumination uniformity.

[0149] Based on pixel ratio and illumination uniformity, the third evaluation score of the vein region image corresponding to the illumination is determined.

[0150] Based on the above embodiments, the score acquisition module 1202 is further configured as follows:

[0151] The degree of blurring of each pixel in the vein region image is determined based on the blur detection algorithm;

[0152] The blur level is normalized to determine the sharpness score corresponding to each pixel.

[0153] Based on the sharpness score of each pixel, the fourth evaluation score for sharpness corresponding to the vein region image is determined.

[0154] Based on the above embodiments, the score acquisition module 1202 is further configured as follows:

[0155] The degree of occlusion of each pixel in the vein region image is determined based on the occlusion detection algorithm.

[0156] The degree of occlusion is normalized to determine the integrity score corresponding to each pixel.

[0157] Based on the integrity score of each pixel, a fifth evaluation score for integrity is determined for the vein region image.

[0158] Based on the above embodiments, the image determination module 1203 is further configured as follows:

[0159] Based on preset weights, the second evaluation score corresponding to posture, the third evaluation score corresponding to illumination, and the fourth evaluation score corresponding to sharpness of the vein region image are configured to obtain the sixth evaluation score.

[0160] The minimum value among the first assessment score corresponding to the size of the vein region image, the fifth assessment score corresponding to the integrity, and the sixth assessment score is selected as the comprehensive score.

[0161] It is worth noting that in the embodiments of the palm vein image quality assessment device described above, the various functional modules are divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of each functional module are only for easy differentiation and are not used to limit the scope of protection of this application.

[0162] Figure 13 This is a schematic diagram of an electronic device provided in an embodiment of this application. This electronic device can be used to execute the palm vein image quality assessment method provided in the above embodiments, and has corresponding functional modules and beneficial effects for executing the method. As shown in the figure, the electronic device includes a processor 1301, a memory 1302, an input device 1303, and an output device 1304. The number of processors 1301 in the device can be one or more; one processor 1301 is shown as an example in the figure. The processor 1301, memory 1302, input device 1303, and output device 1304 in the device can be connected via a bus or other means; a bus connection is shown as an example in the figure. The memory 1302, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the palm vein image quality assessment method in the embodiments of this application. The processor 1301 executes various functional applications and data processing of the device by running the software programs, instructions, and modules stored in the memory 1302, thereby realizing the aforementioned palm vein image quality assessment method.

[0163] The memory 1302 may primarily include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the terminal device. Furthermore, the memory 1302 may include high-speed random access memory and non-volatile memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some instances, the memory 1302 may further include memory remotely located relative to the processor 1301, which can be connected to the terminal device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0164] Input device 1303 can be used to receive input digital or character information, and to generate key signal inputs related to user settings and function control of the device, such as inputting a palm vein image. Output device 1304 can be used to send or display key signal outputs related to user settings and function control of the device, such as outputting a valid palm vein image.

[0165] This application also provides a storage medium storing computer-executable instructions, which, when executed by a processor, are used to perform related operations in the palm vein image quality assessment method provided in this application.

[0166] Computer-readable storage media include both permanent and non-permanent, removable and non-removable media, and information storage can be achieved by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0167] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0168] Note that the above are merely preferred embodiments and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments, and may include many other equivalent embodiments without departing from the concept of this application, the scope of which is determined by the scope of the appended claims.

Claims

1. A method for assessing the quality of palm vein images, characterized in that, include: Acquire palm vein images and locate key points on the palm vein images to obtain corresponding vein region images; The evaluation score of the vein region image corresponding to preset evaluation parameters is obtained. The evaluation parameters include region size, pose, illumination, sharpness, and integrity. Based on the preset scoring quantification formula and the evaluation score, the comprehensive score of the vein region image is determined, and the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score is a qualified palm vein image. For the evaluation parameter item being illumination, the evaluation score of the vein region image corresponding to the preset evaluation parameter item includes: The proportion of pixels with a brightness within a preset brightness range in the vein region image is obtained, and the proportion is used as the pixel ratio. The similarity of the segmented regions divided by the preset segmentation line on the vein region image is obtained, and the similarity is used as the illumination uniformity. Based on the pixel ratio and the illumination uniformity, a third evaluation score for the illumination corresponding to the vein region image is determined.

2. The palm vein image quality assessment method according to claim 1, characterized in that, The step of acquiring a palm vein image and locating key points on the palm vein image to obtain a corresponding vein region image includes: Based on the Openpose algorithm, the location of the key points in the palm vein image was determined; Based on the location of the key points, the vein region on the palm vein image is determined, and the vein region is cropped to obtain the vein region image.

3. The palm vein image quality assessment method according to claim 2, characterized in that, The key points include the center point of the base of the five fingers, the center point of the palm, and the center point of the base of the palm in the palm vein image.

4. The palm vein image quality assessment method according to claim 1, characterized in that, For the evaluation parameter item being region size, the evaluation score of the vein region image corresponding to the preset evaluation parameter item includes: Obtain the image area of ​​the vein region; Based on the image area and the preset area, determine the area score associated with the vein region image; If the area score is greater than a first threshold and less than or equal to a second threshold, the area score is used as the first evaluation score for the vein region image corresponding to the region size. If the area of ​​the image is less than or equal to the preset area, the first evaluation score is 0.

5. The palm vein image quality assessment method according to claim 1, characterized in that, For the evaluation parameter item being posture, the evaluation score obtained for the vein region image corresponding to the preset evaluation parameter item includes: Determine the horizontal rotation angle and pitch angle between the palm plane and the imaging plane corresponding to the vein region image; Based on the horizontal rotation angle and the pitch angle, a first attitude score value corresponding to the horizontal rotation angle and a second attitude score value corresponding to the pitch angle are obtained; Based on the first posture score and the second posture score, a second evaluation score corresponding to the posture of the vein region image is obtained.

6. The palm vein image quality assessment method according to claim 1, characterized in that, For the evaluation parameter item being clarity, the evaluation score of the vein region image corresponding to the preset evaluation parameter item includes: The degree of blurring of each pixel in the vein region image is determined based on the blur detection algorithm. The blur level is normalized to determine the sharpness score corresponding to each pixel. Based on the sharpness score of each pixel, the fourth evaluation score of the sharpness of the vein region image is determined.

7. The palm vein image quality assessment method according to claim 1, characterized in that, For the completeness of the evaluation parameter item, the evaluation score of the vein region image corresponding to the preset evaluation parameter item includes: The degree of occlusion of each pixel in the vein region image is determined based on the occlusion detection algorithm. The degree of occlusion is normalized to determine the integrity score corresponding to each pixel. Based on the integrity score of each pixel, a fifth integrity assessment score is determined for the vein region image.

8. The palm vein image quality assessment method according to claim 1, characterized in that, The process of determining a comprehensive score for the vein region image based on a preset scoring quantification formula and the evaluation score, and defining the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score as a qualified palm vein image, includes: According to preset weights, the second evaluation score corresponding to posture, the third evaluation score corresponding to illumination, and the fourth evaluation score corresponding to sharpness of the vein region image are configured to obtain a sixth evaluation score. The minimum value among the first evaluation score corresponding to the size of the vein region image, the fifth evaluation score corresponding to the integrity, and the sixth evaluation score is selected as the comprehensive score.

9. A device for assessing the quality of palm vein images, characterized in that, include: The image acquisition module is configured to acquire a palm vein image and locate key points on the palm vein image to obtain the corresponding vein region image. The score acquisition module is configured to acquire the evaluation score of the vein region image corresponding to preset evaluation parameters, the evaluation parameters including region size, pose, illumination, sharpness and integrity; The image determination module is configured to determine the comprehensive score of the vein region image based on a preset scoring quantification formula and the evaluation score, and to take the palm vein image corresponding to the vein region image whose comprehensive score meets the preset score as a qualified palm vein image. For the evaluation parameter item being illumination, the score acquisition module is configured as follows: The proportion of pixels with a brightness within a preset brightness range in the vein region image is obtained, and the proportion is used as the pixel ratio. The similarity of the segmented regions divided by the preset segmentation line on the vein region image is obtained, and the similarity is used as the illumination uniformity. Based on the pixel ratio and the illumination uniformity, a third evaluation score for the illumination corresponding to the vein region image is determined.

10. An electronic device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the palm vein image quality assessment method as described in any one of claims 1-8.

11. A storage medium for storing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a processor, are used to perform the palm vein image quality assessment method as described in any one of claims 1-8.

Images