Vein image acquisition method, acquisition system and acquisition device

Abstract

The present application discloses a vein image acquisition method, which includes: setting a first polarizer on one side of the light emitting surface of the light emitting array and setting a second polarizer on the side of the image sensor sensing light, setting the polarization directions of the first polarizer and the second polarizer to form a preset angle; using a distance sensor to detect the distance from the human body to be captured of the vein image to the vein acquisition device; using a three-color display light and a speaker to prompt the correct acquisition distance during registration or identification according to the distance detected by the distance sensor; controlling the light emitting array to emit near-infrared light of different brightness combinations at the distance according to the distance detected by the distance sensor; using an image sensor to capture m vein images of near-infrared light of different brightness combinations; and performing image quality evaluation on the m vein images, selecting the vein image corresponding to the maximum image quality evaluation score as the acquisition result for registration or identification. The present application also provides a vein image acquisition system and a vein image acquisition device.

Description

Technical Field

[0001] The present application relates to the field of biometric feature acquisition, and more specifically, to a vein image acquisition method, acquisition system and acquisition device. Background Art

[0002] In the context of the information society, information security plays an increasingly important role in life. People have put forward higher requirements for the security of personal information and privacy. In the fields of finance, online shopping, customs, etc., there is an urgent need for an effective, convenient and secure means of identity recognition. Authentication methods based on passwords or instructions are easy to be stolen or cracked, and can no longer meet the urgent security requirements of the current society. With the development of computer, network and other technologies, identity authentication using biometrics has emerged and gradually replaced traditional identity authentication methods.

[0003] The veins of the human body are hidden under the epidermis and cannot be illuminated under visible light. They need to be photographed and collected under near-infrared light. The characteristic of human hemoglobin absorbing near-infrared light is used to collect the vein feature image under the palm skin. When using an image sensor to collect vein images, the reflection of the human epidermis affects the contrast between the vein area and the non-vein area of ​​the collected image; when the actual application of the collection device to collect vein images includes vein collection during registration and recognition, the distance and brightness of the collection process will affect the quality of the vein image. For example, when collecting palm veins, the distance is too far, resulting in unclear images. The palm vein image collected at a distance of too far is not complete. Excessive brightness will cause overexposure, and insufficient brightness will easily cause insufficient image brightness. However, the quality of the collected vein image often affects the subsequent algorithm analysis process of the vein image and the accuracy of the recognition and authentication results using vein features. Summary of the invention

[0004] In view of the prior art, the technical problem solved by the present application is to provide a method, system and device for capturing venous images with no reflection, clear images, uniform brightness and rich venous features.

[0005] In order to solve the above technical problems, in a first aspect, the present application provides a vein image acquisition method, which is based on a vein acquisition device to acquire a human vein image during registration or identification, wherein the vein acquisition device includes a light emitting array and an image sensor, wherein the light emitting array is used to emit near-infrared light irradiating a human body to be imaged with a vein, and the image sensor is used to sense the light emitted by the light emitting array to acquire a vein image, and the vein image acquisition method includes:

[0006] A first polarizer is arranged on one side of the light emitting surface of the light emitting array and a second polarizer is arranged on the side of the image sensor sensing light, and the polarization directions of the first polarizer and the second polarizer are arranged to form a preset angle;

[0007] Using a distance measuring sensor to detect the distance between the human body whose vein image is to be collected and the vein collection device;

[0008] According to the distance detected by the distance measuring sensor, a three-color display lamp is used for light display and a loudspeaker is used for voice broadcast to prompt the correct collection distance during registration or recognition;

[0009] Controlling the light-emitting array to emit near-infrared light with different brightness combinations at the distance detected by the distance-measuring sensor;

[0010] Using the image sensor to sense near infrared light of different brightness combinations emitted by the light emitting array, passing through the first polarizer, reflected by the human body and passing through the second polarizer to collect m vein images, where m is greater than 1; and

[0011] An image quality evaluation is performed on the m vein images in terms of brightness, clarity and vein feature information, and a vein image corresponding to a maximum image quality evaluation score is selected as an acquisition result for registration or recognition.

[0012] In one embodiment of the application, the distance detected by the distance measuring sensor is recorded as S, in cm, and the steps of using a three-color display lamp to display light and a speaker to broadcast voice to prompt personnel to collect the correct distance during registration or identification according to the distance detected by the distance measuring sensor include:

[0013] When registration and collection are being performed and the distance range detected by the distance measuring sensor is 5≤S≤10, the three-color display light is used to display green, and the controller announces a voice indicating "the distance is suitable for registration and collection";

[0014] When recognition and collection are being performed and the distance range detected by the distance measuring sensor is 5≤S≤30, the three-color display light is displayed in green and the announcer is used to announce a voice indicating that "the distance is suitable for recognition and collection".

[0015] In one embodiment of the application, the light array is set to include a first light emitting element to an nth light emitting element that respectively emit 1 to n near-infrared wavelengths of light. When performing registration or identification collection, the step of controlling the light array to emit near-infrared light of different brightness combinations at the distance detected by the ranging sensor includes:

[0016] When the distance range detected by the distance measuring sensor is: 5≤S≤8, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 10% and 30% of the maximum brightness of the light emitting element, 10% to 30% of the maximum brightness of the light emitting element is divided into A levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is A nkind;

[0017] When the distance range detected by the distance measuring sensor is: 8<S≤10, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 30% and 50% of the maximum brightness of the light emitting element, 30% to 50% of the maximum brightness of the light emitting element is divided into B levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is B n kind;

[0018] When the distance range detected by the distance measuring sensor is: 10<S≤20, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 50% and 80% of the maximum brightness of the light emitting element, 50% to 80% of the maximum brightness of the light emitting element is divided into C levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is C n kind;

[0019] When the distance range detected by the distance measuring sensor is: 20<S≤30, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 80% and 100% of the maximum brightness of the light emitting element, 80% to 100% of the maximum brightness of the light emitting element is divided into D levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is D n kind.

[0020] In an application embodiment, when registering and collecting, the m vein images for image quality evaluation are multiple vein images of the same human body collected when the light array emits light with different brightness combinations at different collection distances, or are multiple vein images collected when the light array emits light with different brightness combinations at the same collection distance;

[0021] When the recognition acquisition is performed, the m vein images for image quality evaluation are multiple vein images acquired when the light array emits light with different brightness combinations at the same acquisition distance.

[0022] In one embodiment of the application, image quality evaluation is performed on the m vein images to obtain a final registration or recognition palm vein image result during registration or recognition, including:

[0023] Train a neural network evaluation model for vein image clarity and brightness;

[0024] During registration or recognition acquisition, m vein images of the same human body at different brightness acquired by the image sensor during registration or recognition are input into the neural network evaluation model to obtain evaluation scores C1, C2...C j ;

[0025] The vein feature information contained in the m vein images of the same human body under different brightness combinations collected by the image sensor during registration or recognition is evaluated respectively to obtain evaluation scores P1, P2, ..., P j ;

[0026] For the m vein images, the neural network evaluation score and the vein feature information evaluation score of each vein image are weighted averaged to obtain the quality evaluation scores S1, S2, ..., S of the m vein images. j ;

[0027] Set the quality evaluation scores S1, S2, ... S of m vein images. j The vein image corresponding to the maximum value in is the final vein image acquisition result;

[0028] Among them, C j represents the score of the neural network model evaluation of the jth image among m vein images, P j Represents the vein feature information evaluation score of the jth image among m vein images, 1≤j≤m.

[0029] In one embodiment of the application, the vein feature information contained in the jth vein image is evaluated to obtain an evaluation score P j ,include:

[0030] Perform image segmentation on the j-th vein image to obtain vein features;

[0031] The evaluation score P is obtained by calculating the ratio of the number of pixels occupied by the vein features after image segmentation to the total number of pixels in the jth vein image. j , P j =p t / p s ;

[0032] Among them, p t is the number of pixels occupied by the vein features after image segmentation, p s The total number of pixels in the jth vein image.

[0033] In one embodiment of the application, training a neural network evaluation model for vein image clarity and brightness includes:

[0034] Obtain a training vein image sample set;

[0035] The training vein image sample set is annotated with labels about image clarity and brightness quality, including:

[0036] The overall image clarity of each vein image sample in the sample set is respectively counted using the Laplacian kernel of Gaussian, and a clarity quality label of the vein image sample is obtained according to the overall image clarity of the vein image sample;

[0037] The brightness mean of each pixel of each vein image sample is calculated, and the brightness quality label of the vein image sample is obtained according to the brightness mean of each pixel of the vein image sample.

[0038] In one embodiment of the application, the neural network evaluation model is trained based on a CNN neural network model.

[0039] In the vein image acquisition method, during the registration acquisition or identification acquisition process, the first polarizer and the second polarizer with different polarization directions are used to eliminate reflections; the appropriate distance during registration or identification acquisition is prompted and the different brightness of the light-emitting array is adjusted according to the distance to obtain a plurality of vein images with a suitable registration or identification acquisition distance and different brightness; finally, the image quality of the plurality of vein images is evaluated to obtain a vein image with appropriate brightness, clarity and rich vein features.

[0040] In a second aspect, the present application further provides a vein image acquisition system, comprising a controller, a light emitting array for emitting near-infrared light for irradiating a human body, an image sensor, a first polarizer, a second polarizer, a distance measuring sensor, a three-color display light, a speaker, and an image processor; the light emitting array, the image sensor, the distance measuring sensor, the three-color display light, the speaker, and the image sensor are all connected to the controller, the first polarizer is used to convert the light emitted by the light emitting array into polarized light, the second polarizer is located on a side of the image sensor sensing light, covers the image sensor, and its polarization direction forms a preset angle with the polarization direction of the first polarizer;

[0041] The controller controls the distance measuring sensor to detect the distance from the human body whose vein image is to be collected to the image sensor, controls the three-color display light to display light and controls the speaker to make voice announcements according to the distance detected by the distance measuring sensor to remind the personnel of the correct collection distance during registration or identification, controls the light emitting array to emit near-infrared light with different brightness combinations according to the distance detected by the distance measuring sensor, and controls the image sensor to collect multiple vein images with different brightness combinations; the image processor is used to evaluate the image quality of the multiple vein images collected by the image sensor in terms of brightness, clarity and vein feature information.

[0042] In the vein image acquisition system, a first polarizer and a second polarizer with different polarization directions are used to eliminate reflections during the image acquisition process; a controller, a distance sensor, a three-color display light and a speaker cooperate to indicate a suitable acquisition distance and control the luminous brightness combination of the light-emitting array at different acquisition distances; an image processor is used to evaluate the image quality of multiple vein images acquired by the image sensor to obtain a vein image with appropriate brightness, clarity and rich vein features.

[0043] In a third aspect, the present application further provides a vein image acquisition device, comprising:

[0044] a housing; and

[0045] The above-mentioned vein image acquisition system is located in the housing.

[0046] In the vein image acquisition device, the vein image acquisition system is used to eliminate reflections during the vein image acquisition process, and the controller, ranging sensor, three-color display light and speaker cooperate to indicate the appropriate acquisition distance and control the luminous brightness combination of the light-emitting array at different acquisition distances. The image processor is used to evaluate the image quality of multiple vein images acquired by the image sensor to obtain a vein image with appropriate brightness, clarity and rich vein features. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0048] Figure 1 This is an overall flow chart of the vein image acquisition method of the first embodiment of the present application;

[0049] Figure 2 This is a flow chart of performing image quality evaluation on the m vein images according to the first embodiment of the present application;

[0050] Figure 3 This is a structural block diagram of a vein image acquisition system according to Embodiment 2 of the present application;

[0051] Figure 4 This is a structural diagram of the vein image acquisition device of Example 3 of the present application. DETAILED DESCRIPTION

[0052] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0053] It should be noted that when an element is referred to as being "fixed to" or "disposed on" another element, it can be directly on the other element or indirectly on the other element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or indirectly connected to the other element.

[0054] It should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0055] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0056] The vein image acquisition method, vein image acquisition system and vein image acquisition device provided in the embodiments of the present application are now described in conjunction with the accompanying drawings.

[0057] Embodiment 1

[0058] The vein image acquisition method provided in the first embodiment of the present application is based on a vein image acquisition device to acquire a human body vein image for registration or identification, and the human body vein image includes palm vein, finger vein and arm vein images. The vein acquisition device includes a light emitting array and an image sensor, the light emitting array is used to emit near-infrared light to illuminate the human body whose vein image is to be acquired, and the image sensor is used to sense the light emitted by the light emitting array to acquire the vein image.

[0059] Reference Figure 1 The vein image acquisition method provided in the first embodiment includes:

[0060] Step S100: a first polarizer is disposed on one side of the light emitting surface of the light emitting array and a second polarizer is disposed on the side of the image sensor sensing light, and polarization directions of the first polarizer and the second polarizer are set to a preset angle.

[0061] Step S200: using a distance measuring sensor to detect the distance between the human body whose vein image is to be collected and the vein collection device.

[0062] Step S300: according to the distance detected by the distance measuring sensor, a three-color display lamp is used for light display and a loudspeaker is used for voice broadcast to prompt the correct collection distance during registration or recognition.

[0063] Step S400: controlling the light-emitting array to emit near-infrared light of different brightness combinations at the distance detected by the distance-measuring sensor.

[0064] Step S500: using the image sensor to sense near-infrared light of different brightness combinations emitted by the light-emitting array, passing through the first polarizer, reflected by the human body and passing through the second polarizer to collect m vein images, where m is greater than 1.

[0065] Step S600: performing image quality evaluation on the m vein images in terms of brightness, clarity and vein feature information, and selecting the vein image corresponding to the maximum image quality evaluation score as the acquisition result for registration or recognition.

[0066] In step S100, the preset angle between the polarization direction of the first polarizer and the polarization direction of the second polarizer is set to be 600-900. Therefore, since the polarization directions of the first polarizer and the second polarizer are different, the light entering the image sensor is further attenuated after passing through the second polarizer; thus, the reflection of the palm surface or the finger surface is eliminated, and the palm print and fingerprint information is filtered out, effectively improving the contrast between the vein area and the non-vein area of ​​the captured image.

[0067] In this embodiment, the distance detected by the distance measuring sensor is denoted as S, in cm, and step S300 specifically includes the following steps:

[0068] When performing registration and collection and when the distance range detected by the ranging sensor is 5≤S≤10, the three-color display light is displayed green, and the control announcer broadcasts a voice indicating "the distance is suitable for registration and collection"; when the distance range detected by the ranging sensor is S<5, the three-color display light is displayed red and the speaker broadcasts a voice indicating "the distance is too close, please move away from the device"; when the distance range detected by the ranging sensor is S>30, the three-color display light is displayed blue and the speaker broadcasts a voice indicating "the distance is too far, please move closer to the device";

[0069] When performing identification and collection: when the distance range detected by the ranging sensor is 5≤S≤30, the three-color display light displays green and the announcer broadcasts a voice indicating "the distance is suitable for attention identification and collection"; when the distance range detected by the ranging sensor is S<5, the three-color display light displays red and the speaker broadcasts a voice indicating "the distance is too close, please move away from the device"; when the distance range detected by the ranging sensor is S>30, the three-color display light displays blue and the speaker broadcasts a voice indicating "the distance is too far, please move closer to the device".

[0070] For example, when collecting palm vein images, in order to collect complete and clear palm vein images, it is necessary to be within a suitable distance. If the distance is too close, the complete palm veins may not be collected, and the collected palm vein features may not be clear enough. Therefore, step S300 can prompt the person to be collected to the correct collection distance during registration or identification.

[0071] The light-emitting array is set to include a first light-emitting element to an nth light-emitting element that respectively emit 1 to n kinds of near-infrared wavelengths of light. In one embodiment, the light-emitting array is set to emit near-infrared light of mixed wavelengths, specifically, to simultaneously emit light of 760nm, 850nm and 940nm wavelengths. At this time, n is 3, and the first light-emitting element can be set to emit near-infrared light of 760nm, the second light element can emit near-infrared light of 850nm, and the third light-emitting element can emit near-infrared light of 940nm. In other embodiments, the light-emitting array can emit near-infrared light of a single wavelength or near-infrared light of other mixed wavelengths; when the light-emitting array emits near-infrared light of a single wavelength, n is 1. Due to the differences in human biological tissues, the use of mixed-wavelength near-infrared light to collect vein images can better adapt to different groups of people.

[0072] In this embodiment, when performing registration or identification collection, the step of controlling the light-emitting array to emit near-infrared light of different brightness combinations at the distance detected by the distance measuring sensor includes the following steps, that is, step S400 includes:

[0073] Step S410: When the distance range detected by the distance measuring sensor is: 5≤S≤8, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 10% and 30% of the maximum brightness of the light emitting element, 10% to 30% of the maximum brightness of the light emitting element is divided into A levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is A n kind;

[0074] Step S420: When the distance range detected by the distance measuring sensor is: 8<S≤10, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 30% and 50% of the maximum brightness of the light emitting element, 30% to 50% of the maximum brightness of the light emitting element is divided into B levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is B n kind;

[0075] Step S430: When the distance range detected by the distance measuring sensor is: 10<S≤20, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 50% and 80% of the maximum brightness of the light emitting element, 50% to 80% of the maximum brightness of the light emitting element is divided into C levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is C n kind;

[0076] Step S440: When the distance range detected by the distance measuring sensor is: 20<S≤30, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 80% and 100% of the maximum brightness of the light emitting element, 80% to 100% of the maximum brightness of the light emitting element is divided into D levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is D n kind.

[0077] It can be understood that when the distance range detected by the distance measuring sensor is: 5≤S≤8, the combination of the light emitting array emitting multi-spectrum mixed wavelength near-infrared light is A n Species; for the distance range detected by the ranging sensor: 8 < S ≤ 10, the light emitting array emits a multi-spectrum mixed wavelength near-infrared light combination mode B n In the type; for the distance range detected by the ranging sensor: 10 <S ≤ 20, the light emitting array emits a multi-spectrum mixed wavelength near-infrared light combination mode C n Species; for the distance range detected by the ranging sensor: 20 <S ≤ 30, the light emitting array emits a multi-spectrum mixed wavelength near-infrared light combination mode D n In this way, vein images with rich vein features can be collected for different distances. n , B n , C n and D n The n in the figure is the number of light emitting elements in the light emitting array.

[0078] In this embodiment, step S400 is used to control the light brightness of the light array according to the acquisition distance, so that the problem of overexposure or too dark brightness is not easy to occur when collecting vein images at different distances. At the same time, due to the different combinations of emission brightness of the light array, it can adapt to different groups of people, making the image vein acquisition method of the embodiment of the present application highly universal.

[0079] In one embodiment, the A level brightness is 4 levels of brightness, and the A level brightness is 15%, 20%, 25% and 30% of the maximum brightness respectively; the B level brightness is 4 levels of brightness, and the B level brightness is 35%, 40%, 45% and 50% of the maximum brightness respectively; the C level brightness is 4 levels of brightness, and the C level brightness is 50%, 60%, 70% and 80% of the maximum brightness respectively; the D level brightness is 4 levels of brightness, and the D level brightness is 85%, 90%, 95% and 100% of the maximum brightness respectively. In other embodiments, the A, B, C and D level brightness can be other number of level brightness.

[0080] In this embodiment, when performing registration acquisition, the m vein images for image quality evaluation are multiple vein images collected when the light array emits light with different brightness combinations at different acquisition distances of the same human body, or are multiple vein images collected when the light array emits light with different brightness combinations at the same acquisition distance; when performing identification acquisition, the m vein images for image quality evaluation are multiple vein images collected when the light array emits light with different brightness combinations at the same acquisition distance of the same human body.

[0081] It can be understood that registration acquisition usually collects the vein images of the persons who need to be identified and verified later and saves them in the sample image library, while identification acquisition is to collect the vein images of the persons to be identified during identification and verification, and compare the vein images to be identified with the images in the sample image library to determine whether the identification and verification are successful. In order to ensure the efficiency of identification, the vein images are often collected at a fixed distance during identification acquisition; and during registration acquisition, in order to ensure the accuracy of subsequent vein image analysis and identification verification, images can be collected at different distances to ensure that the best quality vein images are collected as vein sample images in the sample image library.

[0082] Reference Figure 2 , performing image quality evaluation on the m vein images to obtain the final registration or recognition palm vein image result during registration or recognition includes the following steps, that is, the step S600 specifically includes the following steps:

[0083] Step S610: training a neural network evaluation model for vein image clarity and brightness;

[0084] Step S620: During registration or recognition acquisition, m vein images of the same human body at different brightness acquired by the image sensor during registration or recognition are input into the neural network evaluation model to obtain evaluation scores C1, C2...C j ;

[0085] Step S630: Evaluate the vein feature information contained in the m vein images of the same human body under different brightness combinations collected by the image sensor during registration or recognition, and obtain evaluation scores P1, P2, ..., P j ;

[0086] Step S640: For the m vein images, weighted average the neural network evaluation score and the vein feature information evaluation score of each vein image to obtain the quality evaluation scores S1, S2, ..., S of the m vein images. j ;

[0087] Step S650: Set the quality evaluation scores S1, S2, ..., S of m vein images. j The vein image corresponding to the maximum value in is the final vein image acquisition result;

[0088] Among them, C j represents the score of the neural network model evaluation of the jth image among m vein images, P j Represents the vein feature information evaluation score of the jth image among m vein images, 1≤j≤m.

[0089] In one embodiment, steps S610 to S650 may be executed sequentially. In other embodiments, steps S620 and S630 may be executed simultaneously or step S630 may be executed before step S620. That is, the order of executing steps S620 and S630 is not limited. Figure 2 An example of only one of these situations is given.

[0090] In this embodiment, the neural network evaluation model is trained based on a CNN neural network model, but is not limited thereto.

[0091] In this embodiment, step S610 specifically includes:

[0092] Step S611: obtaining a training vein image sample set;

[0093] Step S612: labeling the training vein image sample set with image clarity and brightness quality labels, including:

[0094] Step S6121: using the Laplacian of Gaussian kernel to respectively count the overall image clarity of each vein image sample in the sample set, and obtaining a clarity quality label of the vein image sample according to the overall image clarity of the vein image sample;

[0095] Step S6122: Calculate the brightness mean of each pixel of each vein image sample, and obtain the brightness quality label of the vein image sample according to the brightness mean of each pixel of the vein image sample.

[0096] Step S613: Utilize the CNN neural network for training to obtain a neural network evaluation model.

[0097] In the above steps, for step S611, in order to make the trained neural network evaluation model have higher robustness and universality, different devices can be used to collect vein images of different qualities, for example, including acquisition devices with poor acquisition quality and devices with good acquisition quality, and the vein images collected by these devices have a combination of different brightness and different clarity conditions.

[0098] In this embodiment, the step S630 specifically includes:

[0099] Step S630: performing image segmentation on the j-th vein image to obtain vein features;

[0100] Step S631: Obtain an evaluation score P by calculating the pixel ratio of the number of pixels occupied by the vein feature after image segmentation to the total number of pixels of the jth vein image. j , P j =p t / p s ;

[0101] Among them, p t is the number of pixels occupied by the vein features after image segmentation, p s The total number of pixels in the jth vein image.

[0102] In one embodiment, the j-th vein image is segmented using the PSPNet segmentation method to obtain the vein feature information of the j-th vein image. In other embodiments, other image segmentation algorithms may also be used to obtain the vein feature information.

[0103] The image quality evaluation method using the method in step S600 not only evaluates the brightness and clarity of the image, but also comprehensively evaluates whether the vein features of the image itself are rich. This evaluation method is comprehensive and reliable.

[0104] In the vein image acquisition method, during the registration acquisition or identification acquisition process, the first polarizer and the second polarizer with different polarization directions are used to eliminate reflections; the appropriate distance during registration or identification acquisition is prompted and the different brightness of the light-emitting array is adjusted according to the distance to obtain a plurality of vein images with a suitable registration or identification acquisition distance and different brightness; finally, the image quality of the plurality of vein images is evaluated to obtain a vein image with appropriate brightness, clarity and rich vein features.

[0105] Embodiment 2

[0106] Reference Figure 3The vein image acquisition system 100 provided in the second embodiment of the present application includes a controller 1, a light emitting array 2 for emitting near-infrared light to illuminate the human body, an image sensor 3, a first polarizer 4, a second polarizer 5, a ranging sensor 7, a three-color display light 8, a speaker 9 and an image processor 10.

[0107] Specifically, the light emitting array 2, the image sensor 3, the distance sensor 7, the three-color display light 8, the speaker 9 and the image sensor 3 are all connected to the controller 1. The light emitting array 2, the image sensor 3, the distance sensor 7, the three-color display light 8, the speaker 9 and the image sensor 3 are all connected to the controller 1, the first polarizer 4 is used to convert the light emitted by the light emitting array 2 into polarized light, and the second polarizer 5 is located on the side of the image sensor 3 that senses light, covers the image sensor 3, and its polarization direction forms a preset angle with the polarization direction of the first polarizer 4.

[0108] Specifically, the controller 1 controls the ranging sensor 7 to detect the distance from the human body whose vein image is to be collected to the image sensor 3, controls the three-color display light 8 to display light and controls the speaker 9 to make voice broadcasts to prompt personnel of the correct collection distance during registration or identification according to the distance detected by the ranging sensor 7, controls the light emitting array 2 to emit near-infrared light with different brightness combinations according to the distance detected by the ranging sensor 7, and controls the image sensor 3 to collect multiple vein images with different brightness combinations; the image processor 10 is used to evaluate the image quality of the multiple vein images collected by the image sensor 3 with respect to brightness, clarity and vein feature information.

[0109] It is worth noting that the light emitting array 2, the first polarizer 4, the second polarizer 5, the image sensor 3, the ranging sensor 7, the three-color display light 8 and the speaker 9 in the second embodiment are the same as the light emitting array, the first polarizer, the second polarizer, the image sensor, the ranging sensor and the three-color display light in the first embodiment.

[0110] In this embodiment, the image processor 10 is used to perform image quality evaluation on brightness, clarity and vein feature information of multiple vein images captured by the image sensor 3, that is, to execute the image quality evaluation method of step S600 in the above-mentioned embodiment 1.

[0111] In the vein image acquisition system 100, the first polarizer 4 and the second polarizer 5 with different polarization directions are used to eliminate reflections during the image acquisition process; the controller 1, the distance sensor 7, the three-color display light 8 and the speaker 9 cooperate to indicate the appropriate acquisition distance and control the light brightness combination of the light-emitting array 2 at different acquisition distances, and the image processor 10 is used to evaluate the image quality of multiple vein images acquired by the image sensor 3 to obtain a vein image with appropriate brightness, clarity and rich vein features.

[0112] In this embodiment, the vein image acquisition system 100 further includes an optical lens 6. The second polarizer 5 is located between the image sensor 3 and the optical lens 6. The optical lens 6 includes a plurality of lenses and an aperture for converging and focusing light.

[0113] In this embodiment, the light emitting array 2 is a plurality of near-infrared LEDs, and further, the plurality of near-infrared LEDs include a near-infrared LED emitting a wavelength of 760nm, a near-infrared LED emitting a wavelength of 850nm, and a near-infrared LED emitting a wavelength of 940nm. Any LED is a SMD or a direct-plug light emitting diode. The three-color display lamp 8 is a red, green, and blue three-color LED lamp group, and the red, green, and blue three-color LED lamp group is a visible light LED lamp group. The image sensor 3 is a near-infrared CMOS image sensor 3, specifically but not limited to a model of Sony IMX335 image sensor 3. The ranging sensor 7 is a CM36781 sensor. The image processor 10 is a RV1126 model chip, on which an image processing program is transplanted, and the image processing algorithm implements the steps of the image quality evaluation method of step S600. The controller 1 may be an embedded chip, specifically a single-chip microcomputer or a DSP processor; the single-chip microcomputer may be but not limited to an ATMEL series single-chip microcomputer, an STM8 or an STM32 single-chip microcomputer; the DSP processor may be but not limited to a DSP processor of the AU3822UC82 signal.

[0114] Specifically, the controller 1 controls the distance sensor 7 to detect the distance from the human body (palm, finger or arm) whose veins are to be collected to the image sensor 3. The controller 1 receives the distance information detected by the distance sensor 7, and controls the brightness of each near-infrared light LED of the light-emitting array 2 according to the distance information. Specifically, the controller 1 outputs a PWM pulse signal according to the distance to control each near-infrared light LED to emit light. For example, when the duty cycle of the PWM pulse signal output by the controller 1 is zero, the near-infrared light LED does not emit light. When the duty cycle of the PWM pulse signal output by the controller 1 is 100%, the brightness of the near-infrared light LED is maximum. The controller 1 transmits the multiple vein images collected by the image sensor 3 to the image processor 10.

[0115] Embodiment 3

[0116] Reference Figure 4 The vein image acquisition device 200 provided in the third embodiment of the present application includes a shell 20; the vein image acquisition system 100 in the second embodiment is located in the shell 20.

[0117] In this embodiment, the light-emitting array 2 is a circular light-emitting array 2, the first polarizer is a circular polarizer and is directly opposite to and covers the light-emitting array 2; the first polarizer (not shown) and the light-emitting array 2 are arranged around the image sensor (not shown), the second polarizer 5 and the ranging sensor 7, and the three-color display light 8 is arranged around the first polarizer and the light-emitting array 2.

[0118] In this embodiment, the vein image acquisition device 200 further includes a display 21 located in the housing 20, and the display 21 is used to display the vein image acquired by the image sensor and the image result processed by the image processor. The display is an LED or LCD display.

[0119] Further reference Figure 4 The vein image acquisition device further comprises an optical lens 6, and the second polarizer 5 is located between the image sensor and the optical lens 6. The optical lens 6 comprises a plurality of lenses and an aperture for converging light and focusing.

[0120] It is worth noting that the light-emitting array, the first polarizer, the second polarizer 5, the image sensor, the ranging sensor 7, the three-color display light 8 and the speaker in the third embodiment are the same as the light-emitting array, the first polarizer, the second polarizer 5, the image sensor, the ranging sensor, the three-color display light 8 in the second embodiment.

[0121] In the vein image acquisition device 200, the vein image acquisition system 100 is used to eliminate reflections during the vein image acquisition process, and the controller 1, the distance sensor 7, the three-color display light 8 and the speaker cooperate to indicate the appropriate acquisition distance and control the light brightness combination of the light array 2 at different acquisition distances. The image processor is used to evaluate the image quality of multiple vein images acquired by the image sensor to obtain a vein image with appropriate brightness, clarity and rich vein features.

[0122] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A vein image acquisition method, which is based on a vein acquisition device to acquire a human vein image during registration or identification, wherein the vein acquisition device comprises a light emitting array and an image sensor, wherein the light emitting array is used to emit near-infrared light to illuminate the human body to be imaged, and the image sensor is used to sense the light emitted by the light emitting array to acquire the vein image, characterized in that: include: A first polarizer is disposed on one side of the light emitting surface of the light emitting array and a second polarizer is disposed on the side of the image sensor sensing light, and the polarization directions of the first polarizer and the second polarizer are arranged to form a preset angle; Using a distance measuring sensor to detect the distance between the human body whose vein image is to be collected and the vein collection device; According to the distance detected by the distance measuring sensor, a three-color display lamp is used for light display and a loudspeaker is used for voice broadcast to prompt the correct collection distance during registration or recognition; Controlling the light-emitting array to emit near-infrared light with different brightness combinations at the distance detected by the distance-measuring sensor; Using the image sensor to sense near infrared light of different brightness combinations emitted by the light emitting array, passing through the first polarizer, reflected by the human body and passing through the second polarizer to collect m vein images, where m is greater than 1; and Performing image quality evaluation on the m vein images in terms of brightness, clarity and vein feature information, and selecting the vein image corresponding to the maximum image quality evaluation score as the acquisition result of registration or recognition; The distance detected by the distance measuring sensor is recorded as S, in cm. The steps of using a three-color display lamp to display light and a loudspeaker to broadcast voice according to the distance detected by the distance measuring sensor to remind personnel of the correct collection distance during registration or identification include: When registration and collection are being performed and the distance range detected by the distance measuring sensor is 5≤S≤10, the three-color display light is used to display green, and the control announcer is controlled to broadcast a voice indicating "the distance is suitable for registration and collection"; When identification and collection are being performed and the distance range detected by the distance measuring sensor is 5≤S≤30, the three-color display light is displayed green and the announcer is used to announce the voice indicating "the distance is suitable for identification and collection"; The light-emitting array is set to include first to nth light-emitting elements that respectively emit 1 to n near-infrared wavelengths of light. When performing registration or identification collection, the step of controlling the light-emitting array to emit near-infrared light of different brightness combinations at the distance detected by the ranging sensor includes: When the distance range detected by the distance measuring sensor is: 5≤S≤8, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 10% and 30% of the maximum brightness of the light emitting element, 10% to 30% of the maximum brightness of the light emitting element is divided into A levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is A n kind; When the distance range detected by the distance measuring sensor is: 8<S≤10, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 30% and 50% of the maximum brightness of the light emitting element, 30% to 50% of the maximum brightness of the light emitting element is divided into B levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is B n kind; When the distance range detected by the distance measuring sensor is: 10<S≤20, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 50% and 80% of the maximum brightness of the light emitting element, 50% to 80% of the maximum brightness of the light emitting element is divided into C levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is C n kind; When the distance range detected by the distance measuring sensor is: 20<S≤30, by controlling the light emitting brightness of each light emitting element in the light emitting array to be between 80% and 100% of the maximum brightness of the light emitting element, 80% to 100% of the maximum brightness of the light emitting element is divided into D levels of brightness and the light emitting brightness combination of the first light emitting element to the nth light emitting element is D n kind.

2. The vein image acquisition method according to claim 1, characterized in that: When the m vein images for image quality evaluation are collected during registration, they are multiple vein images of the same human body collected when the light array emits light with different brightness combinations at different collection distances, or they are multiple vein images collected when the light array emits light with different brightness combinations at the same collection distance; When the recognition acquisition is performed, the m vein images for image quality evaluation are multiple vein images acquired when the light array emits light with different brightness combinations at the same acquisition distance.

3. The vein image acquisition method according to claim 1, characterized in that: Performing image quality evaluation on the m vein images to obtain a final registration or recognition palm vein image result during registration or recognition, including: Train a neural network evaluation model for vein image clarity and brightness; During registration or recognition acquisition, m vein images of the same human body at different brightness acquired by the image sensor during registration or recognition are input into the neural network evaluation model to obtain evaluation scores C1, C2...C j ; The vein feature information contained in the m vein images of the same human body under different brightness combinations collected by the image sensor during registration or recognition is evaluated respectively to obtain evaluation scores P1, P2, ..., P j ; For the m vein images, the neural network evaluation score and the vein feature information evaluation score of each vein image are weighted averaged to obtain the quality evaluation scores S1, S2, ..., S of the m vein images. j ; Set the quality evaluation scores of m vein images S1, S2, ... j The vein image corresponding to the maximum value in is the final vein image acquisition result; Among them, C j represents the score of the neural network model evaluation of the jth image among m vein images, P j Represents the vein feature information evaluation score of the jth image among m vein images, 1≤j≤m.

4. The vein image acquisition method according to claim 3, characterized in that: Evaluate the vein feature information contained in the jth vein image and obtain the evaluation score P j ,include: Perform image segmentation on the j-th vein image to obtain vein features; The evaluation score P is obtained by calculating the ratio of the number of pixels occupied by the vein features after image segmentation to the total number of pixels in the jth vein image. j , P j =p t / p s ; Among them, p t is the number of pixels occupied by the vein features after image segmentation, p s The total number of pixels in the jth vein image.

5. The vein image acquisition method according to claim 3, characterized in that: The neural network evaluation model trained on the clarity and brightness of vein images includes: Obtain a training vein image sample set; The training vein image sample set is annotated with labels about image clarity and brightness quality, including: The overall image clarity of each vein image sample in the sample set is respectively counted using the Laplacian kernel of Gaussian, and a clarity quality label of the vein image sample is obtained according to the overall image clarity of the vein image sample; The brightness mean of each pixel of each vein image sample is calculated, and the brightness quality label of the vein image sample is obtained according to the brightness mean of each pixel of the vein image sample.

6. The vein image acquisition method according to claim 3, characterized in that: The neural network evaluation model is trained based on the CNN neural network model.

7. A vein image acquisition system, based on the vein image acquisition method according to claim 1, characterized in that: The invention comprises a controller, a light-emitting array for emitting near-infrared light for irradiating a human body, an image sensor, a first polarizer, a second polarizer, a distance sensor, a three-color display light, a speaker and an image processor; the light-emitting array, the image sensor, the distance sensor, the three-color display light, the speaker and the image sensor are all connected to the controller, the first polarizer is used to convert the light emitted by the light-emitting array into polarized light, the second polarizer is located on a side of the image sensor sensing light, covers the image sensor and its polarization direction forms a preset angle with the polarization direction of the first polarizer; The controller controls the distance measuring sensor to detect the distance from the human body whose vein image is to be collected to the image sensor, controls the three-color display light to display light and controls the speaker to make voice announcements according to the distance detected by the distance measuring sensor to remind the personnel of the correct collection distance during registration or identification, controls the light emitting array to emit near-infrared light with different brightness combinations according to the distance detected by the distance measuring sensor, and controls the image sensor to collect multiple vein images with different brightness combinations; the image processor is used to evaluate the image quality of the multiple vein images collected by the image sensor in terms of brightness, clarity and vein feature information.

8. A vein image acquisition device, characterized in that: include: case; as well as The vein image acquisition system as claimed in claim 7 is located in the housing.

Images