[go: up one dir, main page]

WO2018137226A1 - Procédé et dispositif d'extraction d'empreinte digitale - Google Patents

Procédé et dispositif d'extraction d'empreinte digitale Download PDF

Info

Publication number
WO2018137226A1
WO2018137226A1 PCT/CN2017/072712 CN2017072712W WO2018137226A1 WO 2018137226 A1 WO2018137226 A1 WO 2018137226A1 CN 2017072712 W CN2017072712 W CN 2017072712W WO 2018137226 A1 WO2018137226 A1 WO 2018137226A1
Authority
WO
WIPO (PCT)
Prior art keywords
pixel
gaussian
pixel value
matching
preset
Prior art date
Application number
PCT/CN2017/072712
Other languages
English (en)
Chinese (zh)
Inventor
杨德培
Original Assignee
深圳市汇顶科技股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳市汇顶科技股份有限公司 filed Critical 深圳市汇顶科技股份有限公司
Priority to PCT/CN2017/072712 priority Critical patent/WO2018137226A1/fr
Priority to CN201780000029.1A priority patent/CN107077617B/zh
Publication of WO2018137226A1 publication Critical patent/WO2018137226A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G06V10/751—Comparing pixel values or logical combinations thereof, or feature values having positional relevance, e.g. template matching
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/12—Fingerprints or palmprints
    • G06V40/1347—Preprocessing; Feature extraction

Definitions

  • Embodiments of the present invention relate to the field of fingerprint identification technologies, and in particular, to a fingerprint extraction method and apparatus.
  • fingerprint recognition technology is widely used in electronic devices such as mobile phones and tablet computers.
  • Fingerprint recognition has become one of the main ways of unlocking mobile phones and mobile payment, which brings great life to people.
  • the quality of fingerprint extraction directly affects the accuracy of fingerprint recognition.
  • fingerprint extraction usually reads the sensing data twice by the fingerprint sensing chip, one value of the pixel corresponding to the chip in the state where the finger is not touched, and the value of the corresponding pixel of the chip in the state of the finger touch.
  • the fingerprint image is obtained by comparing the difference between the two data.
  • the traditional approach is to obtain a fingerprint image by subtracting two sensing images.
  • the sensing data read by the fingerprint sensing chip is not stable. Therefore, the fingerprint image obtained by the conventional method by using the two sensing data to make a difference contains a large amount of noise, and even a fingerprint image is not obtained.
  • An object of the present invention is to provide a fingerprint extraction method and apparatus, which can extract a fingerprint image based on a mixed Gaussian background model and can extract a high quality fingerprint image.
  • an embodiment of the present invention provides a fingerprint extraction method, and a package Include: when detecting a finger touch, acquiring a sensing pixel value of each pixel in the fingerprint sensing area; identifying each pixel according to a sensing Gaussian background model of each pixel point and a preset Gaussian background model of each pixel point A corresponding texture feature of the point; a fingerprint image of the finger is generated according to the texture feature corresponding to each pixel point.
  • the embodiment of the present invention further provides a fingerprint extraction device, comprising: a pixel value acquisition module, configured to acquire a sensing pixel value of each pixel in the fingerprint sensing area when detecting a finger touch; and a texture feature recognition module, And identifying a texture feature corresponding to each pixel point according to the mixed Gaussian background model of each pixel point and a preset pixel image generation module; and the fingerprint image generation module is configured to: according to the texture feature corresponding to each pixel point , generating a fingerprint image of the finger.
  • the sensing pixel value of each pixel in the fingerprint sensing area and the preset mixed Gaussian background model of each pixel point are touched by the acquired finger, and then each pixel is identified.
  • the texture features of the points are arranged according to the pixel points to generate the fingerprint image of the finger; the mixed Gaussian background model has a good description effect on the unstable pixels, and the high-quality fingerprint image is extracted by the mixed Gaussian background model.
  • identifying the texture feature corresponding to each pixel point specifically includes: determining, for each pixel point, the pixel point Whether the sensed pixel value matches the mixed Gaussian background model of the pixel; when the sensed pixel value of the pixel matches the mixed Gaussian background model of the pixel, the texture feature of the pixel is recognized as a concave path.
  • This embodiment provides a specific identification manner; that is, the sensing pixel value of the pixel of the concave road is matched with the mixed Gaussian background model of the preset pixel, and the concave path of the fingerprint image is obtained accordingly.
  • identifying the texture feature corresponding to each pixel point further includes: when the pixel value of the pixel point and the pixel point When the mixed Gaussian background model does not match, it is determined whether the sensed pixel value of the pixel meets the preset ridge matching condition; when the sensed pixel value of the pixel satisfies the ridge matching condition The texture feature of the pixel is recognized as a ridge path; when the sensed pixel value of the pixel does not satisfy the ridge matching condition, the texture feature of the pixel is recognized as a concave path.
  • the texture feature of the pixel is considered to be a ridge, otherwise, the texture of the pixel
  • the feature is a concave road; the embodiment further improves the above specific recognition mode; accordingly, the convex road and the concave road of the fingerprint image can be simultaneously acquired to obtain a complete fingerprint image.
  • the mixed Gaussian background model includes several Gaussian components arranged in sequence; determining whether the sensing pixel value of the pixel point matches the mixed Gaussian background model of the pixel point, specifically: sequentially sensing pixel values of the pixel points and several Gaussian components Aligning, determining whether there is a Gaussian component matching the sensed pixel value of the pixel; wherein, when there is a Gaussian component matching the sensed pixel value of the pixel, determining the mixed pixel value of the pixel point and the mixed Gauss of the pixel
  • the background model matches; when there is no Gaussian component matching the sensed pixel value of the pixel, it is determined that the sensed pixel value of the pixel does not match the mixed Gaussian background model of the pixel.
  • the embodiment provides a specific manner for determining whether the sensing pixel value of the pixel point matches the mixed Gaussian background model of the pixel point; wherein the mixed Gaussian background model includes several Gaussian components, which can effectively describe the multi-peak state of the sensing pixel value.
  • the sensing pixel value of the pixel is sequentially compared with the plurality of Gaussian components, and determining whether there is a Gaussian component matching the sensing pixel value of the pixel point includes: calculating the sensing pixel value of the pixel point and each Gaussian component.
  • This embodiment further refines the specific manner of whether the above determination is matched.
  • the minimum difference between the absolute value of the difference between the sample mean value of the Gaussian component and the sensed pixel value of the pixel point is used as the matching parameter, which further improves the matching method. The accuracy of the decision.
  • the method further includes: updating the mixed Gaussian background model of each pixel when the fingerless touch is detected.
  • the mixed Gaussian background model of each pixel is preset to update the mixed Gaussian background model when the environment changes.
  • the preset manner of the mixed Gaussian background model of each pixel specifically includes: a mixed Gaussian model for creating pixel points; the mixed Gaussian model includes a plurality of Gaussian components arranged in sequence; and a base pixel value according to the plurality of acquired pixel points Multi-learning update of the mixed Gaussian model; wherein the basic pixel value of the pixel is acquired without finger touch; normalizing the weights of multiple Gaussian components in the mixed Gaussian model after multiple learning updates Processing; according to the preset selection rule, selecting a plurality of Gaussian components from the plurality of Gaussian components after the normalization to form a mixed Gaussian background model.
  • a specific implementation manner of the preset mixed Gaussian background model is provided to meet actual design requirements.
  • the manner of learning to update specifically includes: comparing the basic pixel value of the pixel point with the plurality of Gaussian components sequentially arranged, and determining whether there is a Gaussian component matching the basic pixel value of the pixel point; when there is one and the pixel When the base pixel value of the point matches the Gaussian component, the weight of the Gaussian component is updated according to the preset weight increment, and the sample mean and the sample variance of the Gaussian component are updated according to the base pixel value of the pixel; Multiple Gaussian components are reordered.
  • a specific implementation manner of the learning update is provided; that is, when there is a Gaussian component matching the basic pixel value of the pixel, the Gaussian component has a higher weight, and the Gaussian component is increased by a preset increment. Weights, and update their sample mean and sample variance, make the Gaussian component arrangement in the mixed Gaussian background model more reasonable.
  • the method further includes: deleting the Gaussian component in the mixed Gaussian model when there is no Gaussian component matching the basic pixel value of the pixel; According to the basic pixel value of the pixel, a new Gaussian component is added to the mixed Gaussian model; the weight of the Gaussian component other than the newly added Gaussian component in the mixed Gaussian model is updated according to the preset weight reduction.
  • This embodiment is a further improvement of the above learning update method; That is, when there is no Gaussian component matching the base pixel value of the pixel point, it is required to update the Gaussian component in the mixed Gaussian background model, and at this time, a new one created according to the base pixel value of the pixel point is added.
  • the Gaussian component and the deletion of a Gaussian component at the end ensure the accuracy of the Gaussian background model.
  • FIG. 1 is a specific flowchart of a fingerprint extraction method according to a first embodiment of the present invention
  • FIG. 2 is a specific flowchart of identifying a texture feature corresponding to each pixel point according to a mixed Gaussian background model of each pixel point and a preset mixed Gaussian background model according to a second embodiment of the present invention
  • FIG. 3 is a specific flowchart of a preset manner of a mixed Gaussian background model of each pixel in the second embodiment of the present invention.
  • FIG. 4 is a specific flowchart of a fingerprint extraction method according to a third embodiment of the present invention.
  • FIG. 5 is a block diagram showing a fingerprint extracting apparatus according to a fourth embodiment of the present invention.
  • FIG. 6 is a block diagram showing a texture feature recognition module according to a fifth embodiment of the present invention.
  • Figure 7 is a block diagram showing a fingerprint extracting apparatus according to a sixth embodiment of the present invention.
  • FIG. 8 is a block schematic diagram of a model preset module in accordance with a sixth embodiment of the present invention.
  • a first embodiment of the present invention relates to a fingerprint extraction method, which is applied to a terminal device such as a smartphone.
  • the specific process of the fingerprint extraction method is shown in Figure 1.
  • step 101 it is determined whether the finger is touched.
  • the sensor in the terminal device can detect whether there is a finger touch, and the sensor can be a pressure sensor or other sensor that can detect a finger press or touch.
  • Step 102 Acquire a sensing pixel value of each pixel in the fingerprint sensing area.
  • the fingerprint sensor collects the sensing pixel value corresponding to each pixel point in the fingerprint sensing area at this time. More specifically, the data collected by the fingerprint sensor is usually an M ⁇ N matrix, and each element in the matrix corresponds to a corresponding pixel value of each pixel point collected by the fingerprint sensor.
  • Step 103 Identify a texture feature corresponding to each pixel point according to the sensed pixel value of each pixel point and a preset mixed Gaussian background model of each pixel point.
  • the sensing pixel value of each pixel point collected by the fingerprint sensor in the finger touch state and the untouch state is different, and the preset mixed Gaussian background model of each pixel point is each collected by the finger untouched state.
  • the sensing pixel value of each pixel point in the embodiment may be referred to as the basic pixel value corresponding to each pixel point; wherein each pixel point corresponds to In a mixed Gaussian background model.
  • the texture features corresponding to each pixel point can be obtained.
  • Step 104 Generate a fingerprint image of the finger according to the texture feature corresponding to each pixel point.
  • the texture features of each pixel pair are obtained, the texture features are arranged in accordance with the pixel points, so that the fingerprint image of the finger can be obtained.
  • the embossed road is set to the flag bit "1"
  • the point of the darker color is used to set the embossed road as the flag bit.
  • 0 represented by a lighter dot, so that the texture features of each pixel can be represented by "1” and "0”, and then the fingerprint of the finger can be obtained by replacing the dot of the corresponding color.
  • the sensing pixel value of each pixel in the fingerprint sensing area and the preset mixed Gaussian background model of each pixel point are touched by the acquired finger, and then each pixel is identified.
  • the texture features of the points are arranged according to the pixel points to generate the fingerprint image of the finger; the mixed Gaussian background model has a good description effect on the unstable pixels, and the high-quality fingerprint image is extracted by the mixed Gaussian background model.
  • a second embodiment of the present invention relates to a fingerprint extraction method.
  • This embodiment is a refinement of the first embodiment.
  • the main refinement is that in the second embodiment of the present invention, the steps in the first embodiment are 103: According to the sensing pixel value of each pixel and the preset Gaussian background model of each pixel, the texture feature corresponding to each pixel is identified, and is specifically described.
  • Step 1031 Determine whether the sensed pixel value of the pixel point matches the mixed Gaussian background model of the pixel point. If yes, go to step 1032; if no, go to step 1033.
  • Step 201 Create a mixed Gaussian model of pixel points, and the mixed Gaussian model includes a plurality of Gaussian components arranged in sequence.
  • X j can be represented by a Gaussian mixture model consisting of M Gaussian components, that is, The mixed Gaussian model corresponding to the pixel is composed of M Gaussian components, wherein the Gaussian component can be represented by the probability of occurrence of the pixel, and the formula is as follows:
  • ⁇ k represents the weight of the kth Gaussian component in the mixed Gaussian model
  • ⁇ k and ⁇ k represent the mean and standard deviation of the kth Gaussian component, respectively
  • ⁇ (X j , ⁇ k , ⁇ k ) is the Gaussian probability density Function, expressed as:
  • Equation (1) can represent the probability of occurrence of the base pixel value of the pixel corresponding to the pixel j point acquired at a certain moment.
  • ⁇ init the standard deviation ⁇ init and the mean of the different Gaussian distributions are uniformly initialized to the pixel values of the corresponding pixel points in the first frame sensing data.
  • Step 202 Perform a plurality of learning updates on the mixed Gaussian model according to the basic pixel values of the plurality of acquired pixels.
  • the learning phase is started from the basic pixel value of the pixel collected by the second frame sensor, and the parameters of different mixed Gaussian models corresponding to each pixel are continuously learned and updated, and the sensor collects 1000 to 2000 frames during the whole learning process. image.
  • the number of frames of the image collected by the sensor is not limited, and may be fluctuated according to the change of the basic pixel value of the pixel point collected by the sensor.
  • the manner of learning to update specifically includes:
  • the base pixel value of the pixel is sequentially compared with the plurality of Gaussian components arranged in order, and it is determined whether or not there is a Gaussian component matching the base pixel value of the pixel.
  • the weight of the Gaussian component is updated according to the preset weight increment, and the sample mean and the sample variance of the Gaussian component are updated according to the base pixel value of the pixel. It should be noted that the basic pixel value of the pixel is acquired when there is no finger touch.
  • the base pixel value of the pixel in the new input fingerprint sensor it is checked whether each base pixel value matches the M Gaussian distributions in the corresponding Gaussian mixture model, if the base pixel value and one of the Gaussian component samples are averaged The absolute value of the difference is less than the first threshold, and the base pixel value is considered to match the Gaussian distribution.
  • the Gaussian component in the mixed Gaussian model matches, the Gaussian component needs to be updated, the weight is increased by the preset weight increment, and the sample mean and the sample variance of the Gaussian component are updated with the current base pixel value. The rest of the Gaussian composition remains unchanged.
  • the specific manner of updating the sample mean of the Gaussian component by the current base pixel value may be expressed by the following formula: (1-P)U k +PX j ; where U k represents the sample mean before updating, X j represents the sensed pixel value; P represents the degree of matching of the base pixel value and the Gaussian component.
  • the matching degree can be calculated based on the difference between the basic pixel value and the mean value of the Gaussian component sample; the smaller the difference is, the higher the matching degree is, the larger the difference is, and the smaller the matching degree is. It should be noted that updating the sample variance of the Gaussian component with the current pixel value is also based on this principle, and details are not described herein again.
  • the last Gaussian component in the mixed Gaussian model is deleted; and a new Gaussian component is added to the mixed Gaussian model according to the base pixel value of the pixel;
  • the preset weight decrement updates the weight of the Gaussian component other than the newly added Gaussian component in the mixed Gaussian model.
  • the base pixel value of the pixel in the newly input fingerprint sensor if each of the base pixel values does not match the M Gaussian distributions in the corresponding Gaussian mixture model, then the last Gaussian component in the mixed Gaussian model is deleted. , create a new Gaussian component to replace it.
  • the mean value of the newly created Gaussian component takes the base pixel value of the pixel, the standard deviation and the weights are initialized with values ⁇ init and ⁇ init .
  • the mean and variance of the remaining Gaussian components are unchanged, and the weight is reduced by the preset weight reduction.
  • the plurality of Gaussian components are reordered according to a preset sorting rule.
  • the preset sorting rule follows the component row with large weight and small variance, and the row with small weight and large variance.
  • Step 203 normalize the weights of the plurality of Gaussian components in the mixed Gaussian model after the plurality of learning updates.
  • the weights of multiple Gaussian components in the mixed Gaussian model change, and the weights of multiple Gaussian components in the mixed Gaussian model may not be equal to 1 (greater than 1 or less) 1) Therefore, it is necessary to normalize the weights of the Gaussian components after multiple updates.
  • Step 204 Select a plurality of Gaussian components from the normalized plurality of Gaussian components according to a preset selection rule to form a mixed Gaussian background model.
  • the Gaussian component is selected, and the weight of the selected Gaussian component is accumulated, and when the accumulated weighted value reaches the preset value, the selection is stopped.
  • the normalized mixed Gaussian model includes five Gaussian components, and the weights of the five Gaussian components sequentially arranged are 0.35, 0.25, 0.20, 0.15, and 0.05, respectively, and if the preset value is set to 0.8, The weights of the first three Gaussian components are accumulated and reach the preset value, then the first three Gaussian components are selected to form a mixed Gaussian background model.
  • the preset value is not limited in this embodiment, and may be set according to experience or requirement; generally, the preset value is less than 1 (when the preset value is equal to 1, it indicates that the selected Gaussian model is selected. All Gaussian components form a mixed Gaussian background model).
  • the mixed Gaussian background model includes a plurality of Gaussian components arranged in sequence.
  • the mixed Gaussian background model of the sensed pixel values of the pixel points and the pixel points is considered. Match, otherwise, the sensed pixel value of the pixel is considered to not match the mixed Gaussian background model of the pixel.
  • the composition it is determined whether there is a Gauss that matches the sensed pixel value of the pixel.
  • the difference between the sensed pixel value of the pixel point and the sample mean value in each Gaussian component is calculated, and the difference with the smallest absolute value is obtained as the matching parameter.
  • the first threshold is preset.
  • the matching parameter is less than or equal to the preset first threshold, it is determined that there is a Gaussian component matching the sensed pixel value of the pixel, and the sensed pixel value of the pixel is considered to be the pixel point.
  • Mixed Gaussian background model matching when the matching parameter is greater than the first threshold, determining that there is no Gaussian component matching the sensing pixel value of the pixel;
  • step 1032 the texture feature of the pixel is identified as a concave path.
  • the texture features of the fingerprint are divided into a concave road and a convex road, and the sensing values corresponding to the fingerprint concave road and the convex road in the finger touch state are also different, and the sensing pixel value of the pixel corresponding to the concave road and the preset mixed Gauss
  • the sensed pixel values of the pixel points in the background model are matched. Therefore, when it is determined that the sensed pixel value of the pixel point matches the mixed Gaussian background model of the pixel point, the texture feature of the pixel point is recognized as a concave line.
  • Step 1033 Determine whether the sensed pixel value of the pixel point satisfies a preset ridge matching condition. If yes, go to step 1034; if no, go to step 1032.
  • the texture feature of the pixel is identified as a ridge. Otherwise, the texture feature of the pixel is identified as a concave path.
  • the second threshold is preset.
  • the matching parameter is greater than the preset second threshold, the sensed pixel value of the pixel point is considered to satisfy the ridge matching condition.
  • the second threshold is greater than the first threshold.
  • Step 1034 identifying the texture feature of the pixel as a ridge.
  • the texture feature of the pixel is recognized as a ridge.
  • the embodiment provides a preset manner of the mixed Gaussian background model of the pixel points and a specific identification manner of the texture features corresponding to each pixel point, which is a perfect description of the first embodiment, and is full The actual design needs.
  • a third embodiment of the present invention relates to a fingerprint extraction method, which is an improvement of the first embodiment, and is mainly improved in that a mixed Gaussian background model of each pixel is updated when no finger touch is detected.
  • FIG. 4 The specific process of the fingerprint extraction method provided in this embodiment is shown in FIG. 4 .
  • Step 301 and step 101 are substantially the same, and steps 303 to 305 are substantially the same as steps 102 to 104, and are not described here again.
  • step 302 is newly added in this embodiment, and the specific explanation is as follows:
  • Step 302 updating the mixed Gaussian background model of each pixel.
  • the mixed Gaussian background model of each pixel point may be updated in cycles.
  • a mixed Gaussian background model of each pixel point may be updated every four hours during the day.
  • the process of updating the mixed Gaussian background model of the pixel points first, one or more learning updates are performed on the mixed Gaussian background model according to the base pixel values of the currently collected pixel points; and then, one or more learning sessions are performed.
  • the weights of the plurality of Gaussian components in the updated mixed Gaussian background model are normalized; finally, according to the preset selection rules, several Gaussian components are selected from the normalized Gaussian components to form New hybrid Gaussian background model.
  • the present embodiment updates the mixed Gaussian background model of each pixel point when no finger touch is detected, and updates the mixed Gaussian background model of the pixel point in time to ensure that the pixel changes when the environment changes. The accuracy of the mixed Gaussian background model.
  • a fourth embodiment of the present invention relates to a fingerprint extracting apparatus applied to an electronic device such as a mobile phone.
  • the fingerprint extraction device includes a pixel value acquisition module 1, a texture feature recognition module 2, and a fingerprint image generation module 3.
  • the pixel value obtaining module 1 is configured to acquire, when detecting a finger touch, a sensing pixel value of each pixel in the fingerprint sensing area;
  • the texture feature recognition module 2 is configured to identify a texture feature corresponding to each pixel point according to a sensed pixel value of each pixel point and a preset mixed Gaussian background model of each pixel point;
  • the fingerprint image generating module 3 is configured to generate a fingerprint image of the finger according to the texture feature corresponding to each pixel point.
  • the present embodiment is an apparatus embodiment corresponding to the first embodiment, and the present embodiment can be implemented in cooperation with the first embodiment.
  • the related technical details mentioned in the first embodiment are still effective in the present embodiment, and are not described herein again in order to reduce repetition. Accordingly, the related art details mentioned in the present embodiment can also be applied to the first embodiment.
  • the fingerprint extraction device compares the sensing pixel value of each pixel in the fingerprint sensing area and the preset mixed Gaussian background model of each pixel point by the acquired finger, and then identifies The texture features corresponding to each pixel point are arranged according to the pixel points to generate a fingerprint image of the finger; the mixed Gaussian background model has a good description effect on the unstable pixels, and the high-quality fingerprint is extracted by using the mixed Gaussian background model. image.
  • a fifth embodiment of the present invention relates to a fingerprint extracting apparatus.
  • This embodiment is a refinement of the fourth embodiment.
  • the main refinement is that in the fifth embodiment of the present invention, the texture feature identifying module 2 is provided.
  • the specific modules included are further described.
  • the texture feature recognition module 2 of the fingerprint extraction device includes a first matching unit 21 and a second matching unit 22.
  • the first matching unit 21 is configured to determine whether the sensed pixel value of each of the determined pixel points matches the mixed Gaussian background model of the pixel point, and the mixed pixel value of the pixel point and the mixed Gaussian back of the pixel point When the scene models match, the first matching unit 21 recognizes the texture features of the pixel points as concave lines.
  • the second matching unit 22 is configured to determine, when the first matching unit 21 determines that the sensing pixel value of the pixel point does not match the mixed Gaussian background model of the pixel point, whether the sensing pixel value of the pixel point satisfies a preset ridge matching condition. When the sensed pixel value of the pixel meets the ridge matching condition, the second matching unit 22 recognizes the texture feature of the pixel as a ridge; and when the sensed pixel value of the pixel does not satisfy the ridge matching condition, the second matching unit 22 The texture feature of the pixel is identified as a concave path.
  • the mixed Gaussian background model includes a plurality of Gaussian components arranged in sequence, and the first matching unit 21 is specifically configured to sequentially compare the sensing pixel values of the pixel points with the plurality of Gaussian components to determine whether there is a sensing of the pixel points.
  • a Gaussian component matching the pixel values when there is a Gaussian component matching the sensed pixel value of the pixel, the first matching unit 21 determines that the sensed pixel value of the pixel matches the mixed Gaussian background model of the pixel;
  • the first matching unit 21 determines that the sensed pixel value of the pixel does not match the mixed Gaussian background model of the pixel.
  • the first matching unit 21 includes: a calculation subunit and a judgment subunit.
  • the calculation subunit is configured to calculate a difference between the sensed pixel value of the pixel point and the sample mean value in each Gaussian component, and obtain the difference with the smallest absolute value as the matching parameter corresponding to the sensed pixel value of the pixel point.
  • the determining subunit is configured to determine whether the matching parameter is less than or equal to a preset first threshold.
  • the determining subunit determines that there is a Gaussian component matching the sensing pixel value of the pixel, and when the matching parameter is greater than the preset first threshold, determining the subunit It is determined that there is no Gaussian component that matches the sensed pixel value of the pixel.
  • the ridge matching condition includes: the matching parameter is greater than a preset second threshold; wherein the second threshold is greater than the first threshold.
  • the first threshold and the second threshold are not limited, and may be set as needed.
  • the present embodiment can be combined with the second embodiment.
  • the implementation methods are implemented in cooperation with each other.
  • the related technical details mentioned in the second embodiment are still effective in the present embodiment, and the technical effects that can be achieved in the second embodiment can also be implemented in the present embodiment. To reduce the repetition, details are not described herein again. Accordingly, the related art details mentioned in the present embodiment can also be applied to the second embodiment.
  • the embodiment provides a specific composition of the texture feature recognition module, which can complete the preset of the mixed Gaussian background model of the pixel and the identification of the texture feature corresponding to each pixel, which is a perfect description of the fourth embodiment to meet the actual situation. Design requirements.
  • a sixth embodiment of the present invention relates to a fingerprint extracting apparatus.
  • the present embodiment is an improvement of the fourth embodiment.
  • the improvement is that, referring to FIG. 7, the fingerprint extracting apparatus further includes a model preset module 4.
  • the model preset module 4 is configured to update the mixed Gaussian background model of each pixel point when no finger touch is detected.
  • the model preset module 4 includes a creation unit 41, a learning update unit 42, a normalization processing unit 43, and a selection unit 44.
  • the creating unit 41 is configured to create a mixed Gaussian model of the pixel points;
  • the mixed Gaussian model includes a plurality of Gaussian components arranged in sequence;
  • the learning update unit 42 is configured to perform a plurality of learning updates on the mixed Gaussian model according to the basic pixel values of the pixels that are acquired multiple times. It should be noted that the basic pixel values of the pixel points pass through the pixel value acquiring module 1 when there is no finger touch. Obtain;
  • the normalization processing unit 43 is configured to normalize the weights of the plurality of Gaussian components in the mixed Gaussian model after the multiple learning update;
  • the selecting unit 44 is configured to select a plurality of Gaussian components from the plurality of Gaussian components after the normalization according to the preset selection rule to form a mixed Gaussian background model.
  • the learning update unit 42 includes a matching subunit, a first update subunit, and an ordering.
  • the subunit and the second update subunit are included in the learning update unit 42 .
  • the matching sub-unit is configured to sequentially compare the basic pixel value of the pixel with the plurality of Gauss components sequentially arranged, and determine whether there is a Gaussian component matching the basic pixel value of the pixel;
  • the first update subunit is configured to: when the matching subunit determines that there is a Gaussian component matching the base pixel value of the pixel, update the weight of the Gaussian component according to the preset weight increment, and according to the base pixel value of the pixel Update the sample mean and sample variance of the Gaussian component;
  • the sorting subunit is configured to reorder a plurality of Gaussian components according to a preset sorting rule
  • the second update subunit is configured to delete the Gaussian component in the mixed Gaussian model when the matching subunit determines that there is no Gaussian component matching the base pixel value of the pixel; according to the base pixel value of the pixel, in the mixed Gaussian A new Gaussian component is added to the model; the weight of the Gaussian component other than the newly added Gaussian component in the mixed Gaussian model is updated according to the preset weight reduction.
  • the present embodiment can be implemented in cooperation with the third embodiment.
  • the technical details mentioned in the third embodiment are still effective in the present embodiment, and the technical effects that can be achieved in the third embodiment are also implemented in the present embodiment. To reduce the repetition, details are not described herein again. Accordingly, the related art details mentioned in the present embodiment can also be applied to the third embodiment.
  • the fingerprint extraction device updates the mixed Gaussian background model of each pixel point when the fingerless touch is detected, and updates the mixed Gaussian background model of the pixel point in time to ensure the environment. When changing, the accuracy of the mixed Gaussian background model of the pixel points.
  • each module involved in the fourth to sixth embodiments of the present invention is a logic module.
  • a logic unit may be a physical unit or a part of a physical unit. It can also be implemented in a combination of multiple physical units.
  • the present embodiment does not introduce a unit that is not closely related to solving the technical problem proposed by the present invention, but this does not mean that there are no other units in the present embodiment.
  • a program instructing associated hardware the program being stored in a storage medium, including instructions for causing a device (which may be a microcontroller, chip, etc.) or a processor to perform the various embodiments of the present application. All or part of the steps of the method.
  • the foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk, and the like. .

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Artificial Intelligence (AREA)
  • Health & Medical Sciences (AREA)
  • Human Computer Interaction (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Software Systems (AREA)
  • Collating Specific Patterns (AREA)
  • Image Input (AREA)
  • Image Analysis (AREA)

Abstract

La présente invention se rapporte au domaine de la reconnaissance d'empreinte digitale, et concerne un procédé et un dispositif d'extraction d'empreinte digitale. Le procédé d'extraction d'empreinte digitale comprend les étapes consistant : à déterminer si la pression d'un doigt est détectée (101); si tel est le cas, à acquérir une valeur de détection de chaque pixel dans une région de détection d'empreinte digitale (102); à effectuer une reconnaissance, en fonction de la valeur de détection de chaque pixel et d'un modèle d'arrière-plan à mélange gaussien prédéfini pour chaque pixel, afin d'obtenir une caractéristique de forme correspondant à chaque pixel (103); et à générer, selon la caractéristique de forme correspondant à chaque pixel, une image d'empreinte digitale du doigt (104). Le procédé extrait une image d'empreinte digitale sur la base d'un modèle d'arrière-plan à mélange gaussien, permettant ainsi l'extraction d'une image d'empreinte digitale de grande qualité.
PCT/CN2017/072712 2017-01-25 2017-01-25 Procédé et dispositif d'extraction d'empreinte digitale WO2018137226A1 (fr)

Priority Applications (2)

Application Number Priority Date Filing Date Title
PCT/CN2017/072712 WO2018137226A1 (fr) 2017-01-25 2017-01-25 Procédé et dispositif d'extraction d'empreinte digitale
CN201780000029.1A CN107077617B (zh) 2017-01-25 2017-01-25 指纹提取方法及装置

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/CN2017/072712 WO2018137226A1 (fr) 2017-01-25 2017-01-25 Procédé et dispositif d'extraction d'empreinte digitale

Publications (1)

Publication Number Publication Date
WO2018137226A1 true WO2018137226A1 (fr) 2018-08-02

Family

ID=59613826

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/072712 WO2018137226A1 (fr) 2017-01-25 2017-01-25 Procédé et dispositif d'extraction d'empreinte digitale

Country Status (2)

Country Link
CN (1) CN107077617B (fr)
WO (1) WO2018137226A1 (fr)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107831468A (zh) * 2017-10-19 2018-03-23 天津大学 一种基于仿射传播聚类的指纹定位算法
WO2019104719A1 (fr) * 2017-12-01 2019-06-06 深圳市汇顶科技股份有限公司 Procédé d'amélioration d'image d'empreinte digitale et module d'image d'empreinte digitale
CN113537196B (zh) * 2021-07-21 2023-04-07 拉扎斯网络科技（上海）有限公司 图片识别方法、装置、计算机设备及计算机可读存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101686338A (zh) * 2008-09-26 2010-03-31 索尼株式会社 分割视频中的前景和背景的系统和方法
CN104463132A (zh) * 2014-12-17 2015-03-25 浙江维尔科技股份有限公司 一种指纹识别方法及装置
CN105095867A (zh) * 2015-07-21 2015-11-25 哈尔滨多智科技发展有限公司 基于深度学习的快速动态人脸提取、识别方法
CN105844265A (zh) * 2016-06-07 2016-08-10 广东欧珀移动通信有限公司 一种指纹图像处理方法和装置

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8325993B2 (en) * 2009-12-23 2012-12-04 Lockheed Martin Corporation Standoff and mobile fingerprint collection
CN102184404B (zh) * 2011-04-29 2012-11-28 汉王科技股份有限公司 掌纹图像中手掌区域的获取方法及装置
CN102393905B (zh) * 2011-07-13 2013-04-10 哈尔滨工程大学 手背静脉模式纹理提取方法
CN104866616B (zh) * 2015-06-07 2019-01-22 中科院成都信息技术股份有限公司 监控视频目标搜索方法
CN105894701B (zh) * 2016-04-05 2018-06-29 江苏电力信息技术有限公司 输电线路防外破大型施工车辆的识别报警方法

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101686338A (zh) * 2008-09-26 2010-03-31 索尼株式会社 分割视频中的前景和背景的系统和方法
CN104463132A (zh) * 2014-12-17 2015-03-25 浙江维尔科技股份有限公司 一种指纹识别方法及装置
CN105095867A (zh) * 2015-07-21 2015-11-25 哈尔滨多智科技发展有限公司 基于深度学习的快速动态人脸提取、识别方法
CN105844265A (zh) * 2016-06-07 2016-08-10 广东欧珀移动通信有限公司 一种指纹图像处理方法和装置

Also Published As

Publication number Publication date
CN107077617A (zh) 2017-08-18
CN107077617B (zh) 2020-08-28

Similar Documents

Publication Publication Date Title
US10726244B2 (en) Method and apparatus detecting a target
CN107480649B (zh) 一种基于全卷积神经网络的指纹汗孔提取方法
WO2020253127A1 (fr) Procédé et appareil d'apprentissage de modèle d'extraction de caractéristiques faciales, procédé et appareil d'extraction de caractéristiques faciales, dispositif et support d'informations
CN111738120B (zh) 人物识别方法、装置、电子设备及存储介质
CN112381837B (zh) 一种图像处理方法及电子设备
CN111144366A (zh) 一种基于联合人脸质量评估的陌生人脸聚类方法
CN106372624B (zh) 人脸识别方法及系统
CN112581355A (zh) 图像处理方法、装置、电子设备和计算机可读介质
CN112560710B (zh) 一种用于构建指静脉识别系统的方法及指静脉识别系统
CN106339719A (zh) 一种图像识别方法及装置
CN111814570A (zh) 一种基于动态阈值的人脸识别方法、系统及存储介质
CN113807237B (zh) 活体检测模型的训练、活体检测方法、计算机设备及介质
WO2023123923A1 (fr) Procédé d'identification de poids de corps humain, dispositif d'identification de poids de corps humain, dispositif informatique, et support
WO2018137226A1 (fr) Procédé et dispositif d'extraction d'empreinte digitale
CN112766218A (zh) 基于非对称联合教学网络的跨域行人重识别方法和装置
CN111079587B (zh) 人脸的识别方法、装置、计算机设备及可读存储介质
CN116266419A (zh) 一种活体检测方法、装置和计算机设备
Chen et al. Fresh tea sprouts detection via image enhancement and fusion SSD
CN117315031A (zh) 插座位姿确定方法、装置、设备及存储介质
CN110135428B (zh) 图像分割处理方法和装置
CN111382791A (zh) 深度学习任务处理方法、图像识别任务处理方法和装置
CN104718559B (zh) 用于图像处理的方法和设备
CN112487232B (zh) 人脸检索方法及相关产品
Abayomi-Alli et al. Facial image quality assessment using an ensemble of pre-trained deep learning models (EFQnet)
CN117475280A (zh) 一种人脸识别模型的训练方法和装置

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17894322

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 17894322

Country of ref document: EP

Kind code of ref document: A1