WO2019157989A1

WO2019157989A1 - Biological feature 3d data acquisition method and biological feature 3d data recognition method

Info

Publication number: WO2019157989A1
Application number: PCT/CN2019/074455
Authority: WO
Inventors: 左忠斌; 左达宇
Original assignee: 左忠斌
Priority date: 2018-02-14
Filing date: 2019-02-01
Publication date: 2019-08-22

Abstract

The present invention provides a biological feature 3D data acquisition method and a biological feature 3D data recognition method. According to the acquisition method, multiple biological feature images of an organism are acquired by means of a camera, and according to the multiple biological feature images, a biological feature 3D model is built, so as to implement biological feature 3D data acquisition of the organism; the identify information of the organism is used as an identifier to form a database comprising multiple pieces of biological feature 3D data; and biological feature 3D data stored in the database is found using the identify information of a target organism, and corresponding point cloud comparison is performed to identify the target organism. Also provided is a biological feature 3D data recognition system based on visible light photography. The present invention improves the efficiency of biological feature information acquisition and recognition, the biological features in space are completely recovered using the acquired biological feature information in space, and great possibility is provided for applications such as identification.

Description

Biometric 3D data acquisition method and biometric 3D data recognition method

Technical field

The invention relates to the field of biometric identification technology, in particular to a biometric 3D data acquisition method and a recognition method.

Background technique

A biological characteristic is an inherent physiological or behavioral characteristic of a living being, such as a fingerprint, a palm print, an iris, or a human face. Biometrics have certain uniqueness and stability, that is, the difference between certain biological characteristics of any two organisms is relatively large, and biological characteristics generally do not change greatly with time, which makes biometrics suitable for application. In scenarios such as authentication information in an authentication or identification system.

The current biometric data is 2D data of the spatial plane. Taking the biometrics of the head and face as an example, the data application of the head and face is stuck in a simple image application, that is, only from a certain angle. The head and face data are processed, identified and applied in other aspects. Taking the biological characteristics of the hand as an example, the 2D method is mainly used to identify the characteristics of one or several hands, and some illegal molecules are collected according to the hand. 2D pictures, imitation of 2D hand features, deceived part of the identification system, bringing a great security risk to personal information security.

Therefore, there is an urgent need for 3D data recognition for biometrics to improve safety and support for subsequent applications.

Summary of the invention

In view of the above problems, the present invention has been made in order to provide a visible light photographing based biometric 3D data recognizing method that overcomes the above problems or at least partially solves the above problems.

The invention provides a biometric 3D data acquisition method, comprising:

A. Start device: After turning on the power switch, start the power management module to supply power to each module of the system, and simultaneously start the camera matrix, central control module, shadowless lighting system and display module;

B. Human hand placement: Place the human hand on the transparent glass cover. By adjusting the position of the hand, the information of the hand is all within the orientation of the information collection. Due to the use of the shadowless lighting system, the angles are collected. The hand information is not shaded; the device includes a virtual position of the hand, providing a description of the placement position of the human hand, ensuring that the entire human hand falls within the range of camera matrix information collection;

C. Parameter setting: Through the display interface, various parameters of the camera matrix can be set;

D. Information collection: After the parameters are set, the camera matrix is started to start the information of the opponent's part, and the collected information will be transmitted to the central control module for analysis and processing in the form of pictures; the camera matrix composed of multiple visible light cameras will be used for biometrics. Information is collected to obtain multiple biometric images;

Processing the plurality of biometric images to extract respective feature points in the plurality of biometric images;

E. Information processing: the signal collected by the camera matrix is transmitted to the central control module for signal processing, and the feature point cloud data of the biometric feature is generated based on the extracted feature points of the plurality of biometric images, including: according to the extracted Characterizing the respective feature points in the plurality of biometric images, performing feature point matching, establishing a matching feature point data set; calculating the respective cameras relative to the living body by using the beam adjustment method according to the optical information of the plurality of visible light cameras Calculating the relative position of the feature in space, and calculating spatial depth information of the feature points in the plurality of biometric images according to the relative position; generating biometrics according to the matched feature point data set and the spatial depth information of the feature points Feature point cloud data;

include:

E.1 Filtering the acquired image

Since the main feature collection point of the human hand is concentrated on the top of the hand finger, the fingertip fingerprint feature also has the unique biometric feature. Therefore, after collecting the feature points of the hand finger, the non-finger end information needs to be adopted first. The algorithm's method is filtered out. The overall idea of the whole algorithm is as follows:

E.1.1 establishing a library file of the joint pattern of the finger end and the second finger and a feature library of the finger joint pattern;

E.1.2 Importing the feature library to perform feature recognition on the information collected by the finger;

E.1.3 After feature recognition, calculate the area of the feature area and calculate the range of the feature area of the fingertip;

E.1.4 Image segmentation of the feature area and the non-finger feature area;

E.1.5 The information of the non-finger feature area is removed from the original image;

E.1.6 The information of the new feature area is further filtered;

E.2 feature point extraction of the acquired image;

E.3 acquisition of image matching and calculation of spatial depth information;

E.4 generation of feature point cloud data;

Constructing a 3D model of the biometric according to the feature point cloud data to implement acquisition of the biometric 3D data; comprising: setting a reference size of the 3D model to be constructed; and calculating a space according to the reference size and the feature point cloud data Position information, determining a spatial size of each feature point in the feature point cloud data, thereby constructing a 3D model of the biometric feature;

The time data of biometric information collected by multiple visible light cameras is recorded, and a 3D model of biometrics with time dimension is constructed according to the feature point cloud data and time data to realize the collection of biometric four-dimensional data.

Optionally, the features of the respective feature points in the plurality of biometric images are described by using a scale invariant feature transform SIFT feature descriptor.

Optionally, the spatial depth information of the feature points in the plurality of biometric images includes: spatial location information and color information.

Optionally, the 3D model of the biometric includes at least one of the following 3D data:

Describe the spatial shape feature data of the 3D model;

Describe the surface texture feature data of the 3D model;

Describe the surface material and lighting feature data of the 3D model.

Optionally, before acquiring the biometric information by using a camera matrix composed of multiple visible light cameras, the method further comprises: arranging the multiple visible light cameras by:

Constructing a support structure, and setting an arc-shaped load-bearing structure on the support structure;

A plurality of visible light cameras are disposed on the curved load bearing structure.

Optionally, the support structure is a cabinet, and the arc-shaped load-bearing structure is disposed in the cabinet, and the method further includes:

Providing a transparent glass cover on one side of the lens facing the lens of the plurality of visible light cameras;

When the hand of the creature is placed on the transparent glass cover, the camera matrix hand-held information composed of a plurality of visible light cameras disposed on the curved load-bearing structure is used for acquisition.

Optionally, it also includes:

Before taking advantage of the camera matrix hand-held information composed of multiple visible light cameras, the camera parameters of each camera are set through the display interface.

Optionally, the multi-vision visual depth calculation method is adopted, which specifically includes:

The biometric information is collected by using a camera matrix composed of multiple visible light cameras to obtain a plurality of biometric images;

Transmitting the plurality of biometric images to a processing unit having an image processor GPU and a central processing unit CPU;

The image information of the plurality of biometric images is allocated to a block block of the GPU for calculation, and combined with the centralized scheduling and allocation function of the CPU, the feature points of the plurality of biometric images are calculated.

Optionally, the GPU is a dual GPU, and each GPU has multiple blocks.

The invention also provides a biometric 3D data identification method, comprising the following steps:

S01. Collecting biometric information,

Collecting a plurality of biometric images of the living body by using a visible light camera, and constructing a 3D model of the biometrics according to the plurality of biometric images to implement biometric 3D data collection of the living body;

S02. storing biometric 3D data,

Collecting the collected biometric 3D data by using the identity information (I1, I2...In) of the living body as an identification mark to form a database including a plurality of biometric 3D data (D1, D2...Dn);

S03. Identification of the target organism,

Collecting biometric 3D data (T1, T2...Tn) of the target organism, and using the identity information (I1, I2...In) of the target organism to find biometric 3D data stored in the database (D1, D2...Dn) Correlating the biometric 3D data (T1, T2...Tn) of the target organism with the biometric 3D data (D1, D2...Dn) stored in the corresponding database to identify the target organism identity of;

The comparison method comprises the following specific steps: performing feature point fitting based on spatial direct matching method, and selecting three or more feature points as matching key points in the corresponding rigid regions of the two point clouds, through coordinate transformation Directly perform feature point corresponding matching; given initial coarse alignment conditions of two point clouds, seek rigid transformation between the two to minimize alignment error;

After the feature points are matched, the data of the point cloud after the best fit of the overall surface is aligned;

The least squares method is used to calculate the similarity.

Optionally, step S01 further includes:

Collecting multiple biometric images of the organism through multiple visible light cameras,

Generating feature point cloud data of the biometric based on the extracted feature points in the plurality of biometric images;

A 3D model of the biometric is constructed according to the feature point cloud data to implement biometric 3D data acquisition.

Optionally, the step of generating feature point cloud data of the biometric based on the extracted feature points in the plurality of biometric images further includes:

Performing matching of feature points according to the extracted features of the feature points in the plurality of biometric images to establish a matching feature point data set;

Calculating, according to optical information of the visible light camera, a relative position of each visible light camera relative to the biological feature in space, and calculating spatial depth information of the feature points in the plurality of biometric images according to the relative position;

The feature point cloud data of the biometric is generated according to the matched feature point data set and the spatial depth information of the feature point.

Optionally, the feature of each feature point in the plurality of biometric images is described by using a scale invariant feature transform SIFT feature descriptor;

According to the optical information of multiple visible light cameras, the relative position of each visible light camera relative to the biological features is calculated by the beam adjustment method.

Optionally, the step of constructing a 3D model of the biometric according to the feature point cloud data further includes:

Setting a reference size of the 3D model to be constructed;

And determining a spatial size of each feature point in the feature point cloud data according to the reference size and spatial location information of the feature point cloud data, thereby constructing a 3D model of the biometric.

Describe the spatial shape feature data of the 3D model;

Describe the surface texture feature data of the 3D model;

Describe the surface material and lighting feature data of the 3D model.

Optionally, a plurality of visible light cameras are used to form a camera matrix to collect biometric information of the living body, and the camera matrix is arranged by:

Optionally, the living body is a human body, and the identity information includes one or more of a name, a gender, an age, and a document number.

Optionally, the document number includes one or more of an ID number, a passport number, a driver's license number, a social security number, or a military officer number.

Optionally, the biometric information is head information, facial information, and/or iris information, and the method further includes:

Constructing a base connected to the supporting structure, and setting a seat for taking a photographing position of the human body on the base;

When the human body is positioned on the seat, the head information, facial information, and/or iris information of the human body is collected using a camera matrix composed of a plurality of visible light cameras disposed on the curved load bearing structure.

Optionally, a display is disposed on the curved carrying structure;

After constructing a 3D model of the head, face and/or iris, visually displaying the 3D data on the display;

Before the head information, the face information, and/or the iris information are collected by using a camera matrix composed of a plurality of visible light cameras, the photographing parameters of each visible light camera are set through the display interface.

The embodiment of the invention provides a biometric 3D data recognition method and system based on visible light photographing. In the method, the biometric information is collected by using a camera matrix composed of multiple visible light cameras to obtain a plurality of biometric images; Processing a plurality of biometric images to extract respective feature points in the plurality of biometric images; and subsequently generating feature point cloud data of the biometrics based on the respective feature points in the extracted plurality of biometric images; Point cloud data constructs a 3D model of biometrics to enable acquisition of biometric 3D data. It can be seen that the embodiment of the present invention uses multiple visible light camera control technologies to collect biometric information, which can significantly improve the collection efficiency of biometric information. Moreover, the embodiment of the present invention utilizes the feature information of the biometrics collected in space. The complete restoration of the spatial characteristics of biometrics provides unlimited possibilities for the subsequent application of biometric data. The 3D data is identified by identifying the target identity information, and the target person's data does not need to be compared with the massive data in the database one by one, thereby improving the efficiency of the comparison recognition, greatly improving the speed of the identification, and adopting the sky-based direct matching-based Tianmu. The point cloud fitting method is used to fit the feature points, and the fast fitting comparison of the biometric points is realized, thereby realizing the identity identification and identification.

DRAWINGS

Various other advantages and benefits will become apparent to those skilled in the art from a The drawings are only for the purpose of illustrating the preferred embodiments and are not to be construed as limiting. Throughout the drawings, the same reference numerals are used to refer to the same parts. In the drawing:

1 is a flow chart showing a biometric 3D data recognition method based on visible light photographing according to an embodiment of the invention;

2 is a flow chart showing a biometric 3D data acquisition method based on visible light photographing according to an embodiment of the invention;

3 shows a schematic diagram of a 3D data recognition system for head information, face information, and/or iris information, in accordance with an embodiment of the present invention;

4 is a schematic diagram showing the connection of an internal module and an external connection of a bearer structure in the 3D data identification system shown in FIG. 3;

5 is a schematic diagram showing the connection of a serial port integration module, a camera matrix, and a central processing module in the 3D data identification system shown in FIG. 3;

6 shows a schematic diagram of a 3D data identification system device in accordance with another embodiment of the present invention;

FIG. 7 is a block diagram showing the structure of a 3D data collection device according to an embodiment of the present invention;

FIG. 8 is a block diagram showing the structure of a 3D data collection device according to another embodiment of the present invention.

Detailed ways

Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While the embodiments of the present invention have been shown in the drawings, the embodiments Rather, these embodiments are provided so that this disclosure will be more fully understood and the scope of the disclosure will be fully disclosed.

To solve the above technical problem, an embodiment of the present invention provides a biometric 3D data recognition method based on visible light photographing. FIG. 1 is a flow chart showing a biometric 3D data recognition method based on visible light photographing according to an embodiment of the present invention:

S01. Collecting biometric information,

Collecting a plurality of biometric images of the living body through a visible light camera, and constructing a 3D model of the biometrics according to the plurality of biometric images to realize biometric 3D data collection of the living body;

S02. storing biometric 3D data,

S03. Identification of the target organism,

Collect biometric 3D data (T1, T2...Tn) of the target organism, and use the identity information (I1, I2...In) of the target organism to find the biometric 3D data (D1, D2...Dn) stored in the database, and target the target The biometric 3D data (T1, T2...Tn) of the organism are respectively compared with the biometric 3D data (D1, D2...Dn) stored in the corresponding database to identify the identity of the target organism.

Preferably, as shown in FIG. 2, the collecting of the biometric information in step S01 may further include the following steps S102 to S108.

Step S102, using multiple visible light cameras to collect biometric information to obtain a plurality of biometric images. Preferably, multiple visible light cameras form a camera matrix to collect the living body;

Step S104, processing a plurality of biometric images to extract respective feature points in the plurality of biometric images;

Step S106, generating feature point cloud data of the biometric feature based on the respective feature points in the extracted plurality of biometric images;

Step S108: Construct a 3D model of the biometric feature according to the feature point cloud data to implement the collection of the biometric 3D data of the living body.

In this embodiment, the collection of biometric information is performed by using multiple visible light camera control technologies, and the collection efficiency of the biometric information can be significantly improved. Moreover, the embodiment of the present invention utilizes the collected feature information of the biometrics in space to completely restore the biometric features. The various features in space provide unlimited possibilities for the subsequent application of biometric data.

In another embodiment of the present invention, a camera can be used for biometric information collection. At this time, the camera can be rotated one turn along a predetermined track, thereby realizing multi-angle shooting of biometric information. Biometric image.

In an optional embodiment of the present invention, the feature point cloud data of the biometric feature is generated based on the respective feature points in the extracted plurality of biometric images in the step S106, which may include the following steps S1061 to S1063.

Step S1061: Perform matching of feature points according to characteristics of respective feature points in the extracted plurality of biometric images, and establish a matching feature point data set.

Step S1062, calculating spatial relative positions of the cameras relative to the biometrics according to the optical information of the plurality of visible light cameras, and calculating spatial depth information of the feature points in the plurality of biometric images according to the relative positions.

Step S1063: Generate feature point cloud data of the biometric feature according to the matched feature point data set and the spatial depth information of the feature point.

In the above step S1061, the features of the respective feature points in the plurality of biometric images may be described by using a SIFT (Scale-Invariant Feature Transform) feature descriptor. The SIFT feature descriptor has 128 feature description vectors, which can describe the 128 aspects of any feature point in direction and scale, significantly improve the accuracy of feature description, and the feature descriptor has spatial independence.

In the step S1062, the spatial relative position of each camera relative to the biometric feature is calculated according to the optical information of the plurality of visible light cameras. The embodiment of the present invention provides an optional solution, in which, according to the solution, The optical information of the visible light camera is calculated by the beam adjustment method to determine the relative position of each camera relative to the biological features in space.

In the definition of the beam adjustment method, assuming that there is a point in the 3D space, which is seen by multiple cameras located at different positions, the beam adjustment method is able to extract the coordinates of the 3D point from the multi-view information and The relative position of each camera and the process of optical information.

Further, the spatial depth information of the feature points in the plurality of biometric images mentioned in step S1062 may include: spatial position information and color information, that is, may be an X-axis coordinate of the feature point in the spatial position, and the feature point is in the space The Y-axis coordinate of the position, the Z-axis coordinate of the feature point at the spatial position, the value of the R channel of the color information of the feature point, the value of the G channel of the color information of the feature point, the value of the B channel of the color information of the feature point, and the feature The value of the alpha channel of the color information of the point, and so on. In this way, the generated feature point cloud data includes spatial location information and color information of the feature points, and the format of the feature point cloud data can be as follows:

X1 Y1 Z1 R1 G1 B1 A1

X2 Y2 Z2 R2 G2 B2 A2

......

Xn Yn Zn Rn Gn Bn An

Where Xn represents the X-axis coordinate of the feature point at the spatial position; Yn represents the Y-axis coordinate of the feature point at the spatial position; Zn represents the Z-axis coordinate of the feature point at the spatial position; and Rn represents the value of the R-channel of the color information of the feature point. ; Gn represents the value of the G channel of the color information of the feature point; Bn represents the value of the B channel of the color information of the feature point; and An represents the value of the alpha channel of the color information of the feature point.

In the embodiment of the present invention, the biological feature of the plane 2D plus the time dimension constitutes a 3D biological feature, and completely restores various features of the biological feature in space, providing unlimited possibilities for subsequent application of biometric data. .

In an optional embodiment of the present invention, the 3D model of the biometric feature is constructed according to the feature point cloud data in step S108 above, specifically, the reference size of the 3D model to be constructed is set; and then the reference size and feature point cloud data are further determined according to the reference size and the feature point cloud data. The spatial position information determines the spatial size of each feature point in the feature point cloud data, thereby constructing a 3D model of the biometric.

In the 3D model of the constructed biometrics, 3D data describing spatial shape feature data of the 3D model, surface texture feature data describing the 3D model, surface material describing the 3D model, and lighting feature data may be included in the embodiment of the present invention. limit.

In an optional embodiment of the present invention, time data of collecting biometric information by multiple visible light cameras may also be recorded, thereby constructing a 3D model of biometrics with time dimension according to feature point cloud data and time data to realize biometrics. Collection of four-dimensional data. The four-dimensional data here may be a plurality of 3D data sets of the same time interval or different time intervals, different angles, different orientations, different expression forms, and the like.

In an optional embodiment of the present invention, before collecting biometric information by using a camera matrix composed of multiple visible light cameras in step S102, a plurality of visible light cameras may be disposed, and a method of arranging multiple visible light cameras may include The following steps S202 to S204.

Step S202, constructing a support structure, and setting an arc-shaped load-bearing structure on the support structure;

Step S204, placing a plurality of visible light cameras on the curved load bearing structure.

It can be seen that the embodiment of the present invention uses multiple visible light camera control technologies to collect biometric information, which can significantly improve the collection efficiency of biometric information. Also, a plurality of cameras are arranged to form a camera matrix on the curved load bearing structure.

Further, when the biometrics to be collected are different, the specific collection manner of step S102 is also different, which will be described in detail below.

Case 1, if the biometric information is the head information, the facial information, and/or the iris information of the human body, the base connected to the support structure may be built, and the seat for fixing the photographed position of the human body is set on the base; when the person is seated On the chair, the head, face and/or iris information is acquired using a camera matrix consisting of multiple visible light cameras arranged on the curved load bearing structure.

In an alternative embodiment, the display can also be placed on the curved load bearing structure; after the 3D model of the head face is constructed, the head face 3D data is visually displayed on the display.

In an optional embodiment, before the head and face information is collected by using a camera matrix composed of multiple visible light cameras, camera parameters such as sensitivity, shutter speed, and zoom can be set through the display interface. The embodiment of the present invention is not limited to the multiple, the aperture, and the like.

Specific embodiment 1

(1-1) Head and face visible light camera 3D data acquisition device design ideas are as follows.

A. Head and Face Visible Light Camera 3D Data Acquisition Device As shown in FIG. 3, the device may include:

The base 31 serves as a main bottom support structure for the entire inventive device;

The seat 32 fixes the position of the human body and adjusts the height of the human body;

Support structure 33, connecting the bottom of the device and other body mechanisms;

Display 34, an operation interface for the operation of the device system;

Carrying structure 35, fixed structure of camera, central processing unit and light;

Camera matrix 36, facial information collection of the human head;

Strip fill light 37, ambient light is used in addition.

B. Description of the connection relationship of the device

The base 31 is connected to the seat 32 through a connecting structure;

The base 31 is connected to the support structure 33 through a mechanism connection structure;

The support structure 33 is connected to the support structure 35 by a mechanical connection structure;

The display 34 is mechanically fixed to the carrying structure 35;

The camera matrix 36 is fixed on the carrying structure 35 by structural fixing;

The band fill lamp 37 is fixed to the carrier structure 35 by means of a structural fixing.

(1-2) The internal modules of the load-bearing structure 35 are composed as follows.

A. As shown in FIG. 4, the internal module of the load bearing structure 35 can be composed of the following parts:

A power management module that is responsible for providing the various power supplies required for the entire system;

The light management module can adjust the brightness of the light through the central processing module;

The serial port integration module is responsible for two-way communication between the central processing module and the camera matrix;

Central processing module, responsible for system information processing, display, lighting, seat control;

Seat lift management module, responsible for seat height adjustment;

The display driver management module is responsible for the display driver of the display.

B. The internal modules of the bearer structure 35 and the external connections are as follows:

1) The power management module provides power to the camera matrix, the serial port integration module, the light management module, the central processing module, the display drive management module, and the seat lift management module;

2) The serial port integration module connects the camera matrix and the central processing module to realize two-way communication between them, as shown in FIG. 5;

2.1) The camera is connected to the serial port integration module in a single serial manner.

2.2) Serial port integration module is connected to the central processing module via USB interface

2.3) The central processing module realizes the visualization operation of the camera matrix through the customized development software interface

2.4) The camera interface parameters can be set on the operation interface.

Sensitivity ISO (range 50 ~ 6400)

Shutter speed (1/4000 to 1/2) (seconds)

Zoom factor (1 to 3.8x)

Aperture (large/small)

2.5) The operation interface can realize the initialization operation of turning on the camera.

2.6) Operation interface can realize the command of camera image acquisition

2.7) The operation interface can realize the setting of camera image storage path

2.8) Operation interface can realize real-time camera browsing and camera switching

3) The light management module is connected to the power management module, the central processing module, and the external band fill light;

4) The seat lift management module is connected to the power management module, the central processing module and the external seat, and the central processing module realizes the up and down adjustment of the seat height through the visual interface;

5) The display driver management module is connected to the power management module, the central processing module, and an external display;

6) The central processing module is connected to the power management module, the light management module, the seat lift management module, the serial port integration module, and the display drive management module.

(1-3) The device is used as follows

A. Start the device: After turning on the power switch, the central processing unit, camera matrix, and band fill light are activated separately.

B. Parameter setting: Through the display interface, various parameters of the camera matrix can be set.

C. Information collection: After the parameters are set, the startup matrix camera starts to collect information on the human head and face. The information collection time is completed within 0.8 seconds, and the collected signals are finally transmitted to the central processing in the format of digital image (.jpg). The module is processed, and the core of the central processing module consists of the following parts:

C.1 CPU (Central Processing Unit): responsible for the transmission scheduling of the entire digital signal, task allocation, memory management, and some single calculation processing;

C.2 GPU (Graphics Processing Unit): Selects a special model GPU with excellent image processing capabilities and efficient computing capabilities.

C.3 DRAM (Dynamic Random Access Memory): As a temporary storage center for digital signal processing, it needs to match the computing power of CPU and GPU to get the best processing and computing performance.

D. Information Processing: The signal collected by the matrix camera is transmitted to the central processing module for signal processing.

D.1 information processing process is as follows

D.1.1 Filtering the acquired image

Using the characteristics of the GPU, combined with the characteristics of the matrix operator of image filtering, image filtering can be completed quickly with the support of certain algorithms.

D.1.2 Feature point extraction of acquired images

Using CPU and GPU matching the overall performance, because the format of various information of this device is image format, combined with GPU with excellent image processing capability, the information content of jpg can be evenly distributed to the block of GPU. Since the device uses dual GPUs, each GPU has 56 blocks, so the 18 jpg images captured by the acquisition information are evenly distributed to 112 blocks for calculation, combined with the centralized scheduling and allocation functions of the CPU. It can quickly calculate the feature points of each photo. Compared with the operation of a separate CPU or CPU with other common models of GPUs, the overall operation speed time is 1/10 or less of the latter.

D.1.3 Acquisition of image matching and calculation of spatial depth information

Image feature points are extracted using the hierarchical structure of the pyramid and the special algorithm of spatial scale invariance. These two special algorithms are combined with the special structure of the GPU selected by the device to maximize the computing performance of the system and achieve fast extraction. Feature points in image information.

The feature descriptor of this process uses SIFT feature descriptor. The SIFT feature descriptor has 128 feature description vectors, which can describe the 128 features of any feature point in direction and scale, and significantly improve the accuracy of feature description. The descriptor has spatial independence.

The special image processing GPU used in this device has excellent calculation and processing capabilities of individual vectors. For SIFT feature vectors with 128 special descriptors, it is most suitable for processing under the conditions of such special GPUs. To take advantage of the special computing power of the GPU, compare the normal CPU or CPU with other common GPUs, and the matching time of feature points will be reduced by 70%.

After the feature points are matched, the system uses the algorithm of the beam adjustment method to calculate the relative position of the camera relative to the head face. According to the spatial coordinates of the relative position, the GPU can quickly calculate the depth of the feature points of the head face. information.

D.1.4 Generation of feature point cloud data

According to D.1.3, the depth information of the head and face feature points in space is calculated. Due to the vector computing capability of the GPU, the spatial position and color information of the head face feature point cloud can be quickly matched to form a standard model establishment requirement. Point cloud information.

E. Feature Size Calibration: The initial reference size is set for the size of the entire model by the criteria of the feature point cloud size.

Through special calibration on the information collection, the special calibration has a spatially determined size, and since the head facial feature point cloud has spatially uniform dimensionality, the size between any feature points of the head face is determined by the special calibration of the size. It can be calculated from the spatial position coordinates of the point cloud.

F. Subsequent processing of data: Based on the calibrated size in E, 3D data of the face and face of the face can be obtained by further processing the point cloud data.

The format of 3D data has the following files:

.obj - describes the spatial shape characteristics of 3D models

.jpg - Describe the surface texture features of 3D models

.mtl - describes the surface material and lighting characteristics of the 3D model

G. Head face 3D data is displayed on the display by a visual method.

Specific embodiment 2

(2-1) The design idea of the human hand visible light camera 3D data acquisition device is as follows.

A. Human hand visible light camera 3D data acquisition device can include:

The cabinet 61 serves as the main body support structure of the entire device;

The camera matrix 62 collects human hand features, which may specifically be finger features and/or palm features;

Transparent glass cover plate 63, a placement device for the human hand;

Central control module 64, system information processing, analysis, display module;

Hand virtual position 65, description of the placement position of the human hand;

The shadowless lighting system 66 provides a lighting environment for hand 3D modeling.

B. Description of the connection relationship of the device

The cabinet 61 is connected to the camera matrix 62 by mechanical fixing;

The cabinet 61 is connected to the transparent glass cover 63 by means of mechanical structure fixing;

The central control module 64 is connected to the cabinet 61 by mechanical fixing;

The shadowless lighting system 66 is connected to the cabinet 61 by mechanical fixing;

The hand virtual position 65 ensures that the human hand as a whole falls within the range of information acquisition of the camera matrix 62.

(2-2) Central control module 64 and external connection.

A. The entire central control module consists of the following components:

a power management module that provides power to the entire device;

Serial port integration module, responsible for command and data transfer of camera matrix and central processing module;

The lighting management module is responsible for managing the external shadowless lighting system;

Display driver module responsible for management of the display module;

Central processing module, analysis, calculation and processing of collected data of the entire system;

Display module, system operation interface.

B. The connection relationship of the entire central control module is as follows:

B.1 power management module provides power to the camera matrix, serial port integration module, central processing module, lighting management module, display driver module, display module;

The B.2 serial port integration module realizes two-way communication between the camera matrix and the central processing module;

B.3 The lighting management module provides power to the shadowless lighting system and is responsible for adjusting the parameters of the lighting system;

B.4 central processing module is connected to the serial port integration module, the light management module, the display driver module and the power management module;

B.5 The display driver module is connected to the central processing module and the display module.

(2-3) The device is used as follows

A. Start device: After turning on the power switch, start the power management module to supply power to each module of the system, and simultaneously start the camera matrix, central control module, shadowless lighting system and display module.

B. Human hand placement: Place the human hand on the transparent glass cover. By adjusting the position of the hand, all the information of the hand falls within the orientation of the information collection. Because the lighting system of the device uses shadowless lighting. In the system, the hand information collected at various angles has no shadow, which can significantly improve the efficiency and accuracy of feature point collection.

C. Parameter setting: Through the display interface, various parameters of the camera matrix can be set.

D. Information collection: After the parameters are set, the camera matrix is started to start the information of the opponent, and the collected information will be transmitted to the central control module for analysis and processing in the format of the picture.

E. Information processing: The signal collected by the camera matrix is transmitted to the central control module for signal processing. The process of information processing is as follows.

D.1 Filtering the acquired image

D.1.1 establish a library file of the joint pattern of the finger end and the second finger and a feature library of the finger joint pattern;

D.1.2 Importing the feature library to perform feature recognition on the information collected by the finger;

D.1.3 After feature recognition, calculate the area of the feature area and calculate the range of the feature area of the fingertip;

D.1.4 Image segmentation of the feature area and the non-finger feature area;

D.1.5 The information of the non-finger feature area is removed from the original image;

D.1.6 The information of the new feature area is further filtered;

D.2 feature point extraction of captured images

D.3 Acquisition of image matching and calculation of spatial depth information

D.4 feature point cloud data generation

F. Data Subsequent Processing: By further processing the point cloud data, the texture structure of the hand can be obtained.

G. Hand 3D data display: The 3D data of the hand is displayed on the display by a visual method.

Here, D.2, D.3, and D.4 can be referred to the foregoing introduction, and will not be described again here.

Preferably, in step S02, the biometric 3D data collected in step S01 is stored, and the collected biometric 3D data is stored by using the identity information (I1, I2...In) of the living body as an identification mark to form the included A database of a plurality of biometric 3D data (D1, D2, . . . Dn), for example, 3D data D1 is stored with the identity information I1 of the organism as a file name, and 3D data D2 of another organism is identified by the identity information of the organism. It is stored as a file name, and so on, forming a database including n biometric 3D data.

Wherein, when the object to be collected, that is, the living body is a human body, the identity information I includes but is not limited to one or more of a person: name, gender, age, and document number, and the document number may include a person who often uses in life. For example, one or more of an ID number, a passport number, a driver's license number, a social security number, or a military officer number.

Preferably, in the identification of the target organism in step S03, the biometric 3D data (T1, T2...Tn) and the database of the target organism (ie, the organism to be identified) are identified by the Tianmu point cloud alignment method. The biometric 3D data (D1, D2...Dn) stored in the comparison is compared to identify the identity of the target organism. First, by inputting the identity information of the target organism, such as the ID number of the human body, it is possible to quickly find the 3D data (D1, D2...Dn) already stored in the database with the ID number as the file name without having to target The human data is compared with the massive data in the database one by one, which improves the efficiency of the comparison recognition, greatly improves the speed of the identification, and then the currently collected 3D data (T1, T2...Tn) of the human body and The 3D data retrieved from the data is compared, and finally the identity of the human body is recognized to be in conformity, thereby implementing identity authentication. Specifically, the method for identifying the eye-point cloud comparison method includes the following steps:

S301. Feature point fitting;

S302. The overall best fit of the surface;

S303. Similarity calculation.

Preferably, the Tianmu point cloud comparison identification method further comprises the following specific steps:

The feature point fitting is performed by the method based on direct matching of spatial domain. Three or more feature points are selected as the matching key points in the corresponding rigid regions of the two point clouds, and the feature points are directly matched by coordinate transformation;

The least squares method is used to calculate the similarity.

The recognition process and working principle of the Yare Eyes point cloud match recognition method are as follows: First, the point cloud is a basic element constituting a 3D model, which includes spatial coordinate information (XYZ) and color information (RGB). Point cloud attributes include spatial resolution, point precision, surface normal vectors, and more. Its characteristics are not affected by external conditions and will not change for translation and rotation. Reverse software enables the editing and processing of point clouds such as imageware, geomagic, catia, copycad and rapidform. The method of direct matching based on spatial domain specific to the celestial point cloud comparison method includes: Iterative closest point method (ICP), the ICP method is usually divided into two steps, the first step is the feature point fitting, and the second step is the overall surface. Good fit. The purpose of fitting the feature points first is to find and align the two point clouds to fit the fit in the shortest time. But it is not limited to this. For example, it can be:

In the first step, three or more feature points are selected as the matching key points in the corresponding rigid regions of the two point clouds, and the feature points are directly matched by coordinate transformation.

ICP is used for registration of curves or surface segments, which is a very effective tool in 3D data reconstruction. Given the rough initial alignment conditions of two 3D models, ICP iteratively seeks rigid transformation between the two to minimize Alignment errors to achieve registration of spatial geometric relationships between the two.

Given set

with

The set elements represent the coordinate points of the surface of the two models. The ICP registration technique iteratively solves the nearest points of the distance, establishes the transformation matrix, and implements transformation on one of them until the convergence condition is reached, and the iteration stops. The coding is as follows:

1.1 ICP algorithm

Enter .P ₁ , P ₂ .

Output. Transformed P ₂

P ₂ (0)=P ₂ , l=0;

Do

Every point in For P ₂ (l)

Find a nearest point y _i in P ₁ ;

End For

Calculation

Registration error E;

If E is greater than a certain threshold

Calculating a transformation matrix T(l) between P ₂ (l) and Y(l);

P ₂ (l+1)=T(l)·P ₂ (l), l=l+1;

Else

stop;

End If

While||P ₂ (l+l)-P ₂ (l)||>threshold;

Registration error

1.2 Matching based on local feature points:

Taking human face information recognition as an example, the face model is mainly divided into a rigid model part and a plastic model part, and the plastic deformation affects the accuracy of the alignment, thereby affecting the similarity. The first time the plastic model acquires data, there will be local differences. One solution is to select feature points only in the rigid region. The feature points are attributes extracted from an object and remain stable under certain conditions. The commonly used method iterates the nearest point ICP feature points for fitting alignment.

First, extract the area whose face is less affected by the expression, such as the tip of the nose, the outer corner of the eye, the forehead, the tibia, and the ear. The human hand knuckles are rigid areas, the palm part is a plastic area, and the feature points are selected optimally in the finger area. The iris is a rigid model.

Requirements for feature points:

1) Completeness. Contains as much object information as possible to distinguish it from other categories of objects; 2) Compactness: Minimize the amount of data required for expression; 3) Also require features to be rotated and translated in the model. Under the mirror transformation, it remains unchanged.

In 3D biometric recognition, the similarity of the input model is calculated by aligning two 3D biometric model point clouds, wherein the registration error is used as a difference metric.

Step 2: After the feature points are best fitted, the data of the point cloud after the best fit of the overall surface is aligned.

The third step is the similarity calculation. Least squares

The least squares method (also known as the least squares method) is a mathematical optimization technique. It finds the best function match for the data by minimizing the sum of the squares of the errors. The least squares method can be used to easily obtain unknown data and minimize the sum of the squares of the errors between the obtained data and the actual data. The least squares method can also be used for curve fitting. Other optimization problems can also be expressed by least squares by minimizing energy or maximizing entropy. It is often used to solve the curve fitting problem and solve the complete fitting of the surface. The iterative algorithm can speed up data convergence and quickly find the optimal solution.

If the 3D data model is input in the STL file format, the deviation is determined by calculating the distance between the point cloud and the triangle. Therefore, the method requires a plane equation for each triangular patch whose deviation is the distance from the point to the plane. For the 3D data model is IGES or STEP model, since the free-form surface expression is NURBS surface, the point-to-surface distance calculation needs to be calculated by numerical optimization method. The deviation is expressed by iteratively calculating the minimum distance from each point in the point cloud to the NURBS surface, or the NURBS surface is discretized at a specified scale, and the point deviation is approximated by the point-to-point distance, or converted into an STL format for deviation calculation. Different coordinate alignment and deviation calculation methods have different detection results. The size of the alignment error will directly affect the accuracy of the detection and the credibility of the evaluation report.

The best fit alignment is to detect the deviation average to the whole, to ensure that the overall deviation is the minimum condition to terminate the alignment process of the iterative calculation, perform 3D analysis on the registration result, and generate the result object in the form of the root mean square of the error between the two figures. The output, the larger the root mean square, the greater the difference between the two models. The opposite is also true. According to the ratio of the coincidence degree, it is judged whether it is the target of the comparison.

The invention also provides a biometric 3D data recognition system based on visible light photographing, which comprises the following devices:

The biometric information collecting device is configured to collect a plurality of biometric images of the living body, and construct a 3D model of the biometrics according to the plurality of biometric images to realize biometric 3D data collection of the living body;

The biometric 3D data storage device is configured to store the collected biometric 3D data by using the identity information (I1, I2...In) of the living body as an identification mark to form a plurality of biometric 3D data (D1, D2...Dn). Database)

An identification device for the target organism for collecting biometric 3D data (T1, T2...Tn) of the target organism, and using the identity information (I1, I2...In) of the target organism to find the biometric 3D stored in the database Data (D1, D2...Dn) for comparing the biometric 3D data (T1, T2...Tn) of the target organism with the biometric 3D data (D1, D2...Dn) stored in the corresponding database, To identify the identity of the target organism.

Preferably, the biometric information collecting device comprises:

An image acquisition unit is configured to collect biometric information by using a camera matrix composed of a plurality of cameras to obtain a plurality of biometric images;

a feature point extracting unit, configured to process a plurality of biometric images to extract respective feature points in the plurality of biometric images;

a point cloud generating unit, configured to generate feature point cloud data of the biometric based on the respective feature points in the extracted plurality of biometric images;

The 3D model building unit is configured to construct a 3D model of the biometric feature according to the feature point cloud data to implement the collection of the biometric 3D data.

It should be noted that the biometric features in the embodiments of the present invention are not limited to the above-mentioned head, face, and/or iris and hand, and may include other biological features, such as the foot, etc. limit.

The biometric 3D data recognition method and system based on visible light photographing provided by the embodiments of the present invention are further described below by using specific embodiments.

The biometric 3D data acquisition and recognition system of the present invention can adopt a system with the acquisition system, or can adopt two sets of systems respectively.

In one embodiment of the present invention, a 3D data identification system for header information, face information, and/or iris information is shown in FIG. 3, and the system may include:

Display 34, an operation interface for the operation of the device system;

Camera matrix 36, 3D data acquisition of human head information, facial information and/or iris information;

Strip fill light 37, ambient light is used in addition.

Device connection description

The base 31 is connected to the seat 32 through a connecting structure;

The display 34 is mechanically fixed to the carrying structure 35;

As shown in FIG. 4, the internal module of the load bearing structure 35 can be composed of the following parts:

The central processing module further includes an identification module for identifying the target organism, firstly selecting the biometric 3D data (T1, T2...Tn) of the target organism and the biometric 3D data stored in the corresponding database (D1). , D2...Dn) to perform the comparison, and then use the Tianmu point cloud comparison method to identify the identity of the target organism;

Seat lift management module, responsible for seat height adjustment;

The internal modules of the load bearing structure 35 and the external connection relationships are as follows:

2.4) The camera interface parameters can be set on the operation interface.

Sensitivity ISO (range 50 ~ 6400)

Shutter speed (1/4000 to 1/2) (seconds)

Zoom factor (1 to 3.8x)

Aperture (large/small)

2.6) Operation interface can realize the command of camera image acquisition

The device is used as follows

D.1 information processing process is as follows

D.1.1 Filtering the acquired image

D.1.2 Feature point extraction of acquired images

D.1.4 Generation of feature point cloud data

F. Subsequent processing of data: Based on the calibrated size in E, 3D data of the face and face of the face or iris can be obtained by further processing the point cloud data.

The format of 3D data has the following files:

.obj - describes the spatial shape characteristics of 3D models

.jpg - Describe the surface texture features of 3D models

G. Head face 3D data is displayed on the display by a visual method.

The identification module in the central processing module finds the biometric 3D data (D1, D2...Dn) stored in the database according to the identity information (I1, I2...In) of the target organism, and biometrics of the target organism The 3D data (T1, T2...Tn) are respectively compared with the biometric 3D data (D1, D2...Dn) stored in the corresponding database to identify the identity of the target organism, and the recognition result output is displayed on the display. on.

Based on the visible light photographing based biometric 3D data identification method provided by the above embodiments, the embodiment of the present invention further provides a visible light camera based biometric 3D data collecting device.

Of course, in another embodiment of the present invention, the seat 32 may not be included. As shown in FIG. 6, it includes 61 support bases, 62 identification devices, 63 control and display devices, 64 camera matrices, and 65 arc-shaped load-bearing mechanisms. , 66 arc-shaped fill light, when collecting and identifying, the human body stands in the U-shaped area enclosed by the device.

FIG. 7 is a block diagram showing the structure of a biometric 3D data acquisition device based on visible light photographing according to an embodiment of the invention. As shown in FIG. 7, the apparatus may include an image acquisition unit 910, a feature point extraction unit 920, a point cloud generation unit 930, and a 3D model construction unit 940.

The image acquisition unit 910 is configured to collect biometric information by using a camera matrix composed of multiple visible light cameras to obtain a plurality of biometric images.

The feature point extraction unit 920 is coupled to the image acquisition unit 910 for processing a plurality of biometric images to extract respective feature points in the plurality of biometric images;

The point cloud generating unit 930 is coupled to the feature point extracting unit 920, and configured to generate feature point cloud data of the biometric feature based on the respective feature points in the extracted plurality of biometric images;

The 3D model building unit 940 is coupled to the point cloud generating unit 930 for constructing a 3D model of the biometric according to the feature point cloud data to implement biometric 3D data collection.

In an optional embodiment of the present invention, the point cloud generating unit 930 is further configured to:

Performing matching of feature points according to characteristics of respective feature points in the extracted plurality of biometric images, and establishing a matching feature point data set;

Calculating spatial relative position of each camera relative to the biometric according to optical information of the plurality of visible light cameras, and calculating spatial depth information of the feature points in the plurality of biometric images according to the relative positions;

In an alternative embodiment of the invention, the features of the respective feature points in the plurality of biometric images are described using scale invariant feature transform SIFT feature descriptors.

According to the optical information of multiple visible light cameras, the relative position of each camera relative to the biological features in space is calculated by the beam adjustment method.

In an optional embodiment of the invention, the spatial depth information of the feature points in the plurality of biometric images comprises: spatial location information and color information.

In an optional embodiment of the present invention, the foregoing 3D model building unit 940 is further configured to:

Setting a reference size of the 3D model to be constructed;

According to the spatial size information of the reference size and the feature point cloud data, the spatial size of each feature point in the feature point cloud data is determined, thereby constructing a 3D model of the biometric.

In an alternative embodiment of the invention, the 3D model of the biometric includes at least one of the following 3D data:

Describe the spatial shape feature data of the 3D model;

Describe the surface texture feature data of the 3D model;

Describe the surface material and lighting feature data of the 3D model.

In an alternative embodiment of the present invention, as shown in FIG. 8, the apparatus shown in FIG. 7 above may further include:

The camera matrix layout unit 1010 is coupled to the image acquisition unit 910 for arranging multiple visible light cameras by the image acquisition unit 910 before acquiring the biometric information using a camera matrix composed of multiple visible light cameras:

Constructing a support structure and providing an arc-shaped load-bearing structure on the support structure;

A plurality of visible light cameras are arranged on the curved load bearing structure.

In this embodiment, a plurality of cameras are arranged to form a camera matrix on the curved load bearing structure.

In an optional embodiment of the present invention, if the biometric information is head and face information, the image capturing unit 910 is further configured to:

Build a base connected to the support structure, and set a seat on the base for fixing the biological photographing position;

When the creature is on the seat, the head face information is acquired using a camera matrix composed of a plurality of visible light cameras disposed on the curved load bearing structure.

The first display unit 1020 is coupled to the 3D model building unit 940 for setting a display on the curved carrying structure; after constructing the 3D model of the head face, visually displaying the head face 3D data on the display.

In an optional embodiment of the present invention, the image collecting unit 910 is further configured to:

Before the head and face information is collected by using a camera matrix composed of a plurality of visible light cameras, the camera parameters of each camera are set through the display interface.

The embodiment of the invention provides a biometric 3D data recognition method and system based on visible light photographing. In the method, the biometric information is collected by using a camera matrix composed of multiple visible light cameras to obtain a plurality of biometric images; Processing a plurality of biometric images to extract respective feature points in the plurality of biometric images; and subsequently generating feature point cloud data of the biometrics based on the respective feature points in the extracted plurality of biometric images; Point cloud data constructs a 3D model of biometrics to enable acquisition of biometric 3D data. It can be seen that the embodiment of the present invention uses multiple visible light camera control technologies to collect biometric information, which can significantly improve the collection efficiency of biometric information. Moreover, the embodiment of the present invention utilizes the feature information of the biometrics collected in space. The complete restoration of the spatial characteristics of biometrics provides unlimited possibilities for the subsequent application of biometric data.

Further, the embodiment of the present invention can realize the processing of the feature information and the generation of the point cloud quickly and efficiently based on the parallel computing of the central processing unit and the graphics processor. Moreover, using the scale-invariant feature transform SIFT feature descriptor combined with the parallel computing power of the special graphics processor, the matching of feature points and the generation of spatial feature point clouds can be quickly realized. In addition, the unique size calibration method can accurately and quickly extract the spatial size of any feature points of biometrics, and generate 3D models of biometrics to achieve 3D data acquisition.

In the description provided herein, numerous specific details are set forth. However, it is understood that the embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of the description.

Similarly, the various features of the invention are sometimes grouped together into a single embodiment, in the above description of the exemplary embodiments of the invention, Figure, or a description of it. However, the method disclosed is not to be interpreted as reflecting the intention that the claimed invention requires more features than those recited in the claims. Rather, as the following claims reflect, inventive aspects reside in less than all features of the single embodiments disclosed herein. Therefore, the claims following the specific embodiments are hereby explicitly incorporated into the embodiments, and each of the claims as a separate embodiment of the invention.

Those skilled in the art will appreciate that the modules in the devices of the embodiments can be adaptively changed and placed in one or more devices different from the embodiment. The modules or units or components of the embodiments may be combined into one module or unit or component, and further they may be divided into a plurality of sub-modules or sub-units or sub-components. In addition to such features and/or at least some of the processes or units being mutually exclusive, any combination of the features disclosed in the specification, including the accompanying claims, the abstract and the drawings, and any methods so disclosed, or All processes or units of the device are combined. Each feature disclosed in this specification (including the accompanying claims, the abstract and the drawings) may be replaced by alternative features that provide the same, equivalent or similar purpose.

In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features that are included in other embodiments and not in other features, combinations of features of different embodiments are intended to be within the scope of the present invention. Different embodiments are formed and formed. For example, in the claims, any one of the claimed embodiments can be used in any combination.

The various component embodiments of the present invention may be implemented in hardware, or in a software module running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that a microprocessor or digital signal processor (DSP) may be used in practice to implement some of some or all of the components of a visible light camera based biometric 3D data acquisition device in accordance with embodiments of the present invention. Or all features. The invention can also be implemented as a device or device program (e.g., a computer program and a computer program product) for performing some or all of the methods described herein. Such a program implementing the invention may be stored on a computer readable medium or may be in the form of one or more signals. Such signals may be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

It is to be noted that the above-described embodiments are illustrative of the invention and are not intended to be limiting, and that the invention may be devised without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as a limitation. The word "comprising" does not exclude the presence of the elements or steps that are not recited in the claims. The word "a" or "an" The invention can be implemented by means of hardware comprising several distinct elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means can be embodied by the same hardware item. The use of the words first, second, and third does not indicate any order. These words can be interpreted as names.

In this regard, it will be appreciated by those skilled in the <RTIgt;the</RTI> <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; </ RTI> <RTIgt; The content directly determines or derives many other variations or modifications consistent with the principles of the invention. Therefore, the scope of the invention should be understood and construed as covering all such other modifications or modifications.

Claims

A biometric 3D data acquisition method, comprising:

A. Start device: After turning on the power switch, start the power management module to supply power to each module of the system, and simultaneously start the camera matrix, central control module, shadowless lighting system and display module;

B. Human hand placement: Place the human hand on the transparent glass cover. By adjusting the position of the hand, the information of the hand is all within the orientation of the information collection. Due to the use of the shadowless lighting system, the angles are collected. The hand information is not shaded; the device includes a virtual position of the hand, providing a description of the placement position of the human hand, ensuring that the entire human hand falls within the range of camera matrix information collection;

C. Parameter setting: Through the display interface, various parameters of the camera matrix can be set;

D. Information collection: After the parameters are set, the camera matrix is started to start the information of the opponent's part, and the collected information will be transmitted to the central control module for analysis and processing in the form of pictures; the camera matrix composed of multiple visible light cameras will be used for biometrics. Information is collected to obtain multiple biometric images;

Processing the plurality of biometric images to extract respective feature points in the plurality of biometric images;

E. Information processing: the signal collected by the camera matrix is transmitted to the central control module for signal processing, and the feature point cloud data of the biometric feature is generated based on the extracted feature points of the plurality of biometric images, including: according to the extracted Characterizing the respective feature points in the plurality of biometric images, performing feature point matching, establishing a matching feature point data set; calculating the respective cameras relative to the living body by using the beam adjustment method according to the optical information of the plurality of visible light cameras Calculating the relative position of the feature in space, and calculating spatial depth information of the feature points in the plurality of biometric images according to the relative position; generating biometrics according to the matched feature point data set and the spatial depth information of the feature points Feature point cloud data;

include:

E.1 Filtering the acquired image

Since the main feature collection point of the human hand is concentrated on the top of the hand finger, the fingertip fingerprint feature also has the unique biometric feature. Therefore, after collecting the feature points of the hand finger, the non-finger end information needs to be adopted first. The algorithm's method is filtered out. The overall idea of the whole algorithm is as follows:

E.1.1 establishing a library file of the joint pattern of the finger end and the second finger and a feature library of the finger joint pattern;

E.1.2 Importing the feature library to perform feature recognition on the information collected by the finger;

E.1.3 After feature recognition, calculate the area of the feature area and calculate the range of the feature area of the fingertip;

E.1.4 Image segmentation of the feature area and the non-finger feature area;

E.1.5 The information of the non-finger feature area is removed from the original image;

E.1.6 The information of the new feature area is further filtered;

E.2 feature point extraction of the acquired image;

E.3 acquisition of image matching and calculation of spatial depth information;

E.4 generation of feature point cloud data;

Constructing a 3D model of the biometric according to the feature point cloud data to implement acquisition of the biometric 3D data; comprising: setting a reference size of the 3D model to be constructed; and calculating a space according to the reference size and the feature point cloud data Position information, determining a spatial size of each feature point in the feature point cloud data, thereby constructing a 3D model of the biometric feature;

The time data of biometric information collected by multiple visible light cameras is recorded, and a 3D model of biometrics with time dimension is constructed according to the feature point cloud data and time data to realize the collection of biometric four-dimensional data.
The method according to claim 1, wherein the features of the respective feature points in the plurality of biometric images are described by using a scale invariant feature transform SIFT feature descriptor.
The method according to claim 1, wherein the spatial depth information of the feature points in the plurality of biometric images comprises: spatial position information and color information.
The method of claim 1 wherein the 3D model of the biometric comprises at least one of the following 3D data:

Describe the spatial shape feature data of the 3D model;

Describe the surface texture feature data of the 3D model;

Describe the surface material and lighting feature data of the 3D model.
The method of claim 1 wherein prior to acquiring biometric information using a camera matrix comprised of a plurality of visible light cameras, the method further comprising arranging the plurality of visible light cameras in the following manner:

Constructing a support structure, and setting an arc-shaped load-bearing structure on the support structure;

A plurality of visible light cameras are disposed on the curved load bearing structure.
The method according to claim 5, wherein the support structure is a cabinet, and the arc-shaped load-bearing structure is disposed in the cabinet, the method further comprising:

Providing a transparent glass cover on one side of the lens facing the lens of the plurality of visible light cameras;

When the hand of the creature is placed on the transparent glass cover, the camera matrix hand-held information composed of a plurality of visible light cameras disposed on the curved load-bearing structure is used for acquisition.
The method of claim 6 further comprising:

Before taking advantage of the camera matrix hand-held information composed of multiple visible light cameras, the camera parameters of each camera are set through the display interface.
The method according to any one of claims 1 to 7, wherein the method for calculating a multi-vision visual depth comprises:

The biometric information is collected by using a camera matrix composed of multiple visible light cameras to obtain a plurality of biometric images;

Transmitting the plurality of biometric images to a processing unit having an image processor GPU and a central processing unit CPU;

The image information of the plurality of biometric images is allocated to a block block of the GPU for calculation, and combined with the centralized scheduling and allocation function of the CPU, the feature points of the plurality of biometric images are calculated.
The method according to claim 8, wherein the GPU is a dual GPU, and each GPU has a plurality of blocks.
A biometric 3D data identification method, comprising the steps of:

S01. Collecting biometric information,

Collecting a plurality of biometric images of the living body by using a visible light camera, and constructing a 3D model of the biometrics according to the plurality of biometric images to implement biometric 3D data collection of the living body;

S02. storing biometric 3D data,

Collecting the collected biometric 3D data by using the identity information (I1, I2...In) of the living body as an identification mark to form a database including a plurality of biometric 3D data (D1, D2...Dn);

S03. Identification of the target organism,

Collecting biometric 3D data (T1, T2...Tn) of the target organism, and using the identity information (I1, I2...In) of the target organism to find biometric 3D data stored in the database (D1, D2...Dn) Correlating the biometric 3D data (T1, T2...Tn) of the target organism with the biometric 3D data (D1, D2...Dn) stored in the corresponding database to identify the target organism identity of;

The comparison method comprises the following specific steps: performing feature point fitting based on spatial direct matching method, and selecting three or more feature points as matching key points in the corresponding rigid regions of the two point clouds, through coordinate transformation Directly perform feature point corresponding matching; given initial coarse alignment conditions of two point clouds, seek rigid transformation between the two to minimize alignment error;

After the feature points are matched, the data of the point cloud after the best fit of the overall surface is aligned;

The least squares method is used to calculate the similarity;

Step S01 further includes:

The biometric information of the living body is collected by using a plurality of visible light cameras to form a camera matrix, and the camera matrix is laid out by:

Constructing a support structure, and setting an arc-shaped load-bearing structure on the support structure;

Arranging a plurality of visible light cameras on the curved load bearing structure;

Collecting multiple biometric images of the organism through multiple visible light cameras,

Processing the plurality of biometric images to extract respective feature points in the plurality of biometric images;

Generating feature point cloud data of the biometric based on the extracted feature points in the plurality of biometric images;

Constructing a 3D model of biometrics according to the feature point cloud data to implement acquisition of biometric 3D data;

And the step of generating the feature point cloud data of the biometric based on the extracted feature points in the plurality of biometric images further includes:

Performing matching of feature points according to the extracted features of the feature points in the plurality of biometric images to establish a matching feature point data set;

Calculating, according to optical information of the visible light camera, spatial relative positions of the visible light cameras with respect to the biological features, and calculating spatial depth information of the feature points in the plurality of biometric images according to the relative positions;

Generating feature point cloud data of the biometric according to the matched feature point data set and the spatial depth information of the feature point;

The features of the respective feature points in the plurality of biometric images are described by using a scale invariant feature transform SIFT feature descriptor;

According to the optical information of multiple visible light cameras, the relative position of each visible light camera relative to the biological feature is calculated by the beam adjustment method;

The spatial depth information of the feature points in the plurality of biometric images includes: spatial location information and color information;

The step of constructing a 3D model of the biometric according to the feature point cloud data further includes:

Setting a reference size of the 3D model to be constructed;

Determining, according to the reference size and the spatial location information of the feature point cloud data, a spatial size of each feature point in the feature point cloud data, thereby constructing a 3D model of the biometric feature;

The 3D model of the biometric includes at least one of the following 3D data:

Describe the spatial shape feature data of the 3D model;

Describe the surface texture feature data of the 3D model;

Describe the surface material and lighting feature data of the 3D model;

The biometric information is head information, facial information, and/or iris information, and the method further includes:

Constructing a base connected to the supporting structure, and setting a seat for taking a photographing position of the human body on the base;

When the human body is positioned on the seat, the head information, facial information, and/or iris information of the human body is collected using a camera matrix composed of a plurality of visible light cameras disposed on the curved load bearing structure.
The method according to claim 10, wherein the living body is a human body, and the identity information comprises one or more of a name, a gender, an age, and a document number.
The method according to claim 11, wherein the document number comprises one or more of an identity card number, a passport number, a driver's license number, a social security number, or a military officer number.
The method of claim 10 wherein:

Providing a display on the curved carrying structure;

After constructing a 3D model of the head, face and/or iris, visually displaying the 3D data on the display;

Before the head information, the face information, and/or the iris information are collected by using a camera matrix composed of a plurality of visible light cameras, the photographing parameters of each visible light camera are set through the display interface.