WO2020078440A1 - 采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法 - Google Patents

采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法 Download PDF

Info

Publication number
WO2020078440A1
WO2020078440A1 PCT/CN2019/111758 CN2019111758W WO2020078440A1 WO 2020078440 A1 WO2020078440 A1 WO 2020078440A1 CN 2019111758 W CN2019111758 W CN 2019111758W WO 2020078440 A1 WO2020078440 A1 WO 2020078440A1
Authority
WO
WIPO (PCT)
Prior art keywords
camera
iris
distance
face
gimbal
Prior art date
Application number
PCT/CN2019/111758
Other languages
English (en)
French (fr)
Inventor
何召锋
邱显超
马力
李星光
张慧
刘京
李言
李志林
Original Assignee
北京中科虹霸科技有限公司
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
Priority claimed from CN201811212252.5A external-priority patent/CN109451233B/zh
Priority claimed from CN201811472517.5A external-priority patent/CN109614909B/zh
Priority claimed from CN201910201929.3A external-priority patent/CN109977828B/zh
Application filed by 北京中科虹霸科技有限公司 filed Critical 北京中科虹霸科技有限公司
Publication of WO2020078440A1 publication Critical patent/WO2020078440A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N23/00—Cameras or camera modules comprising electronic image sensors; Control thereof
    • H04N23/60—Control of cameras or camera modules

Definitions

  • the present disclosure relates to the field of biometrics collection and recognition, and in particular, to a device for collecting high-definition facial images and a method for automatic tilt adjustment of a camera head.
  • the present disclosure provides a device for collecting high-definition facial images, including a first smart camera, a lighting unit, a distance sensor, and a gimbal.
  • the first smart camera includes a face optical unit, a first iris optical unit, and an algorithm module , And the control module, where,
  • the distance sensor is used to measure the object distance from the user to the first smart camera
  • the first iris optical unit performs automatic focusing according to the object distance
  • the first smart camera and the lighting unit are fixed on the gimbal;
  • the gimbal includes at least two motors that support horizontal rotation and vertical rotation, and are used to adjust the angle of the first smart camera and the lighting unit;
  • the face optical unit, the first iris optical unit and the first smart camera are connected through a MIPI interface;
  • the algorithm module performs face detection on the collected facial image, adjusts the rotation angle of the gimbal according to the face detection result, so that the face is located in the center of the image;
  • the algorithm module performs image quality judgment on the collected facial images, and transmits the facial images that meet the quality requirements to the back end for processing;
  • the first smart camera setting control module performs logic control on other peripheral devices included in the device, and the peripheral device is an optional device.
  • the present disclosure also provides a method for automatic tilt adjustment of a camera head, which is applicable to the device.
  • the method includes:
  • the first adjusted estimated angle is a rough estimate of the angle at which the camera should rotate to collect the face image
  • the first adjusted estimated angle and the control system calculate the second adjusted estimated angle of the gimbal.
  • the second adjusted estimated angle is an accurate estimate of the angle at which the camera should rotate to collect the face image;
  • step S4 Adjust the tilt angle of the gimbal according to the actual rotation angle based on the current gimbal angle, and then return to step S1).
  • FIG. 1 schematically shows a schematic diagram of an apparatus for collecting high-definition facial images of the present disclosure
  • FIG. 2 schematically shows a flow chart of collecting high-definition facial images using the device for collecting high-definition facial images in the present disclosure
  • FIG. 3 schematically shows a structural diagram of an iris collection device with a camera array of the present disclosure
  • FIG. 4 schematically shows the field of view of the iris cameras 101 and 102 of the present disclosure
  • FIG. 5 schematically shows the field of view area of the iris camera 101 and the face camera 104 of the present disclosure
  • FIG. 6A shows the first process flow of the method for expanding the collection distance of the present disclosure
  • FIG. 6B shows the second process flow of the method for expanding the collection distance of the present disclosure
  • FIG. 6C shows the third process flow of the method for expanding the collection distance of the present disclosure
  • FIG. 7 shows a selection method of the iris camera of the present disclosure
  • FIG. 8 is a flowchart of a method for automatically tilting and adjusting the camera head of the present disclosure
  • 9A-9B are schematic diagrams of the position and angle between the user's eyes and the face camera of the present disclosure.
  • 10A to 10D show state diagrams of a pan / tilt and a face camera that calculate and adjust the estimated angle of the present disclosure
  • FIG. 11 shows a method for calculating the pitch angle limit of the gimbal of the present disclosure
  • FIG. 12 is a schematic structural diagram of a device for automatically tilting and adjusting the camera head of the present disclosure
  • FIG. 13 shows the pan / tilt angle adjustment range adapted to the maximum height to the minimum height change of the present disclosure.
  • the optical module used for general face image acquisition has a larger field of view and a larger acquisition range, but the image resolution is lower and the recognition accuracy is lower.
  • the optical module used for general iris image acquisition has a small field of view, a small acquisition range, and a large magnification of the image. It can capture detailed texture information of the eye area and has high recognition accuracy.
  • Recognition algorithms generally use video streaming to determine image quality. If the captured image is a high-definition image, the bandwidth requirements for information transmission are relatively high, and the requirements for back-end computing resources are also strict.
  • the present disclosure provides a device for collecting high-definition facial images, which can simultaneously collect high-definition facial images and iris images to provide basic data for a back-end recognition algorithm.
  • FIG. 1 is a schematic diagram of a device for collecting high-definition facial images of the present disclosure.
  • the device for collecting high-definition facial images in this embodiment includes a first smart camera 101, a distance sensor 103, and a pan / tilt 104.
  • This embodiment also includes a second smart camera 102.
  • the second smart camera 102 is an optional device. When the first smart camera cannot meet the needs of expanding the recognition distance and the recognition angle, the second smart camera 102 is added.
  • the second smart camera 102 is correspondingly matched to the second iris optical unit 108, where the second smart camera 102, as a slave smart camera of the device, works in coordination with the first smart camera according to a certain logic for the need to expand the range of iris collection.
  • the distance sensor 103 is used to measure the object distance from the user to the first smart camera, and may be a TOF laser ranging sensor, a 3D structured light ranging sensor, an infrared ranging sensor, an ultrasonic ranging sensor, or a lidar distance sensor.
  • the first iris optical unit 107 and the second iris optical unit 108 perform auto focusing according to the object distance.
  • the first smart camera 101, the second smart camera 102, and the lighting unit 105 are fixed on the gimbal 104, and the gimbal 104 includes at least two motors that support horizontal rotation and vertical rotation to adjust the first intelligent The angles of the camera 101, the second smart camera 102, and the lighting unit 105.
  • the lighting unit 105 includes one or more light-emitting components.
  • the light-emitting components mix a variety of spectra, and can adjust the light intensity according to the distance information provided by the distance sensor 103 to the first smart camera 101.
  • the light-emitting components LED, VCSEL laser lighting module are LED, VCSEL laser lighting module.
  • the device of this embodiment adjusts the gimbal 104 so that the human face is finally in the center of the entire image.
  • the face optical unit 106, the first iris optical unit 107 are respectively connected to the first smart camera 101 through the MIPI interface, and the second iris optical unit 108 is respectively connected to the second smart camera 102 through the MIPI interface.
  • the face optical unit 106 includes a sensor with a resolution of not less than 2M, an optical lens with a FOV of not less than 65 °, and the MTF value of the face optical unit in the acquisition range is not less than 50%@0.5lp/mm.
  • the first iris optical unit 107 and the second iris optical unit 108 include a sensor with a resolution of not less than 8M, and a lens, and the MTF value of the first iris optical unit 107 and the second iris optical unit 108 within the acquisition range Not less than 50% @ 3lp / mm, the spatial resolution of pixels is greater than 16pixel / mm.
  • the first iris optical unit 107 and the second iris optical unit 108 may adopt a fixed focus structure or an auto focus motor.
  • the first smart camera 101 serves as the main smart camera, and has an internal algorithm module, which has image processing capabilities, can judge the quality of the collected iris image, and set a threshold.
  • Threshold the preset value
  • the corresponding facial image is transferred to the background for iris registration and iris recognition; otherwise, the collected iris and face images are automatically discarded and the collection continues.
  • the first smart camera 101 includes a control module, which can logically control the second smart camera 102.
  • FIG. 2 schematically shows a flow chart of using the device to collect a high-definition facial image in the present disclosure. As shown in FIG. 2, it specifically includes the following steps:
  • S2 The face optical unit of the first smart camera collects facial images
  • step S3 determine whether the face is in the target area; if so, proceed to step S5, if not, proceed to step S4;
  • the first smart camera detects the position and size of the face from the image, controls the gimbal according to the position of the face, and adjusts the face to near the center of the entire image;
  • the distance sensor provides the first smart camera with the object distance from the user to the first smart camera
  • step S8 determine whether the iris image meets the quality requirements, if so, proceed to step S9, if not, return to step S5, and re-read the distance information;
  • the first smart camera sets an algorithm module to judge the quality of the collected image, set a threshold, and judge whether it meets the quality requirements;
  • the first smart camera continuously performs auto focus and image quality judgment, extracts clear images (faces and eye patterns) and pushes them to the back-end image processing platform;
  • the present disclosure provides a device for collecting high-definition facial images.
  • the image determined by the main smart camera through quality judgment is transmitted to the background processing, thereby reducing the pressure on the transmission bandwidth; Collect high-resolution facial images while meeting the ease of use and accuracy of biometrics.
  • Iris recognition is a biometrics technology based on the detailed features of the iris area of the human eye. Because of its high accuracy and uniqueness, it has attracted more and more attention in the field of information security.
  • the traditional iris recognition device uses a single camera to capture the iris images of the user's eyes one by one, which cannot meet multiple distances. Multi-view iris acquisition or dual cameras are used to capture the iris images of the user's eyes at the same time, but when capturing The iris image does not meet the requirements. For example, due to the problem of the illumination angle, the quality of the collected iris image is not high, so you have to re-acquire. In this case, the quality of the iris image is also uncertain in the next acquisition.
  • the present disclosure provides an iris collection device and method for expanding the collection distance, which can realize the collection of clearly focused iris images in a wider collection distance range, and provide high-quality images for iris recognition.
  • Fig. 3 is a structural block diagram of an iris collection device of the present disclosure.
  • 201 is a master camera for collecting iris images
  • 202 is a slave camera for collecting iris images
  • 203 is a slave camera for collecting Face image
  • 204 is a depth camera
  • 2041 is a face camera
  • 2042 is an infrared camera
  • 2043 is an infrared transmitter
  • 2042 receives the infrared scattered point emitted by 2043, perceives the target depth information, and thus outputs the distance between the target and the collection device
  • 2051 and 2052 are left and right infrared lamp boards, which are used to fill light when the 201 and 202 cameras collect iris images
  • 206 are visible light lamp boards, which are used to supplement visible light when the ambient light is dark.
  • the iris camera 1 (which may be the first iris optical unit 107 in FIG. 1) is the main camera, and its lens image focal length is f ′ 1 , the image sensor resolution is m 1 , the object side focal length is f 1 , and the effective collection distance is
  • the iris cameras 201 and 202 and the face camera 203 are closely aligned laterally, and the field of view of the iris cameras 201 and 202 mostly overlap.
  • 4 shows the field of view areas A1 and A2 of the iris cameras 201 and 202.
  • the iris camera 201 can collect clearly focused iris images when the collection distance is within the range of D1, and the iris camera 202 when the collection distance is within the range of D2 All the clearly focused iris images can be collected. It can be seen that the human eye moves forward and backward in the areas E1 and E2. By switching between the iris cameras 201 and 202, the iris images with clear focus can be obtained.
  • the formed camera array can effectively expand the effective collection distance.
  • the iris camera 201 and the face camera 2041 are vertically and tightly arranged vertically, and their optical axis directions are consistent.
  • FIG. 5 shows the field of view of the iris camera 201 and the face camera 2041.
  • the face camera 2041 is located above the iris camera 201.
  • the field of view A1 of the iris camera 201 and the field of view A4 of the face camera 2041 coincide within the effective acquisition distance range D1 of the iris camera 201.
  • the relative horizontal position of the human eye in the images collected by the two cameras is consistent, and the user can adjust the left and right according to the position of the eyes in the visible light image collected by the face camera 2041 to achieve the iris camera 201 ’s The effect of the human eye in the horizontal center of the image during imaging.
  • the camera array is connected with the control chip, receives the instruction of the control module through the communication interface, and starts the corresponding camera to collect images.
  • FIG. 6A is the first method flow of the present disclosure for expanding the collection distance, which can be a long-distance iris collection method flow.
  • the device includes an iris camera 1, and an iris camera 2 and / or an iris camera 3 Other peripheral devices), including the following specific steps:
  • the control module performs distance detection through the human eye detection algorithm and the corresponding distance lookup table
  • the control module selects the image collected by the corresponding iris camera according to the detected distance, and performs feature extraction, iris registration, and iris recognition processing.
  • FIG. 6B is the second method flow of the present disclosure for expanding the collection distance, which can be a long-distance iris collection method flow.
  • the device includes an iris camera 1, a depth camera, and an iris camera 2 and / or an iris camera 3, including the following specific steps:
  • control module selects the corresponding iris camera to start collecting images, transmits the images to the control module, and performs feature extraction, iris registration, and iris recognition processing.
  • FIG. 6C is the third process flow of the method for expanding the collection distance of the present disclosure, which may be a long-distance iris collection method flow.
  • the device includes an iris camera 1, a face camera, and an iris camera 2 and / or an iris camera 3, including the following specific steps:
  • the control module performs distance detection through the human eye detection algorithm and the corresponding distance lookup table
  • the control module selects the corresponding iris camera to start collecting images, transmits the images to the control module, and performs feature extraction, iris registration, and iris recognition processing.
  • FIG. 7 shows a method for selecting an iris camera of the present disclosure.
  • D1, D2, D3 represent the effective collection distance range of iris camera 1, 2, 3, Indicates the start value and end value of the effective collection distance of iris camera 1, Indicates the start value and end value of the effective collection distance of iris camera 2, Represents the start value and end value of the effective collection distance of the iris camera 3, Represents the start value and end value of the effective collection distance of the iris camera 1, 3 overlapping, Represents the starting value and ending value of the effective collection distance of the iris camera 1, 2 overlapping
  • the traditional method is that the effective collection distances of multiple cameras are connected from end to end, and there is no overlapping range.
  • the user is at the threshold of the effective collection distance, there may be some errors in the distance detection algorithm, and the detected distance will fluctuate up and down, which may cause the iris camera.
  • the frequent switching phenomenon affects the stability of the device.
  • the effective collection distances of the multiple iris cameras of the present disclosure overlap to a certain extent. Compared with the traditional method of connecting the effective collection distances of the multiple cameras, it can effectively avoid the problem of boundary oscillation at the critical distance, and prevent the iris camera from appearing Regular frequent switching phenomenon can achieve smooth switching of iris camera.
  • the iris acquisition equipment includes iris camera 1, iris camera 2 and depth camera.
  • the depth camera detects the distance in real time and outputs the detection result to the control module To , Select iris camera 2, when the distance is from To , Keep the iris camera 2 when the distance from To , Select iris camera 1.
  • the iris acquisition equipment includes iris camera 1, iris camera 2, and face camera.
  • the face camera collects face images in real time and transmits them to the control module for human eye detection and distance detection.
  • Select iris camera 1 when the distance is from To
  • Keep iris camera 1 when the distance from To
  • the traditional iris recognition device is only equipped with an infrared camera, and its position and angle remain fixed after installation.
  • the infrared camera can only scan objects within a certain range. The scanning range is limited. When the user stands in front of the device at the best collection distance At this time, too high or too short may cause the device to fail to scan the user's eyes, which is not conducive to the smooth progress of iris recognition.
  • the present disclosure proposes a method for adaptively adjusting the tilt angle of the camera gimbal based on the image collected by the face camera, which can control the tilt rotation of the gimbal through the servo to achieve the effect of acquiring iris images of users of different heights.
  • the method for automatically adjusting the pitch of the camera gimbal described in the embodiments of the present disclosure can be used to realize the rotation of the gimbal 104 in the vertical direction in FIG. 1.
  • FIG. 8 is a flow chart of a method for automatically tilting the camera gimbal of the present disclosure. Specifically, a flow chart of a method for automatically acquiring iris images of users of different heights. By adjusting the position of the face camera to the reference point, the iris camera The center of the eyes of the prospective user ensures that the human eye is imaged within the field of view of the iris camera.
  • the specific steps are:
  • FIG. 9A is a schematic diagram of the position and angle between the user ’s eyes and the face camera 203
  • FIG. 9B is a schematic diagram of the image captured by the face camera 203 in the scene shown in FIG. 9A.
  • the vertical height H2 between the optical axes of the face camera 203 the formula is:
  • D2 is the interpupillary distance between the eyes of the user, which ranges from 6 cm to 8 cm
  • D1 is the interpupillary distance between the eyes collected by the face camera 203;
  • H2 calculates the angle required to rotate the face camera 203 from the current position to the position that can be aligned with the center of the user's eyes.
  • the formula is:
  • L is the horizontal distance between the user's eyes and the face camera 203, which is calculated by the algorithm processing module according to the binocular pupil distance D1 in the image collected by the face camera 203.
  • the first adjusted estimated angle and the control system calculate the second adjusted estimated angle of the gimbal.
  • FIG. 10A is a schematic diagram of the initial state of the gimbal and the face camera 203.
  • point O is the rotation center, indicating the rotation point based on the gimbal
  • A0 is the position of the face camera 203
  • dashed arc A is The rotation trajectory of the face camera 203 is an arc with O as the center and the distance between the face camera 203 and the gimbal as the radius.
  • the initial position of the gimbal and face camera 203 corresponds to the horizontal direction Z0, the face
  • the angle of change of the camera 203 from the initial direction Z0 to the direction of aligning with the eye X1 is, and in this process, the angle of change of the gimbal and the face camera 203 using the gimbal as the origin, from the initial direction Z0 to the direction of aligning with the eye Y1 is, There is the following correspondence between the two rotation angles:
  • the adjustment method is: when the absolute value of the angle required to rotate the face camera 203 this time is calculated to be greater than the above When the absolute value of the actual rotation angle of the gimbal is increased, the control coefficient increases; when the absolute value of the angle that the face camera 203 needs to rotate this time is calculated to be smaller than the absolute value of the actual rotation angle of the previous gimbal, the control coefficient decreases; Otherwise, the control coefficient remains unchanged;
  • step S2 An example of the execution method of step S2) (assuming that the user and the face camera 203 remain stationary during this process):
  • the first adjusted estimated angle taking the positional relationship between the human eye and the camera shown in FIG. 10A as the initial state, calculate the rotation required to rotate the face camera 203 from the current position to a position that can be aligned with the center of the user's eyes
  • the camera 203 moves from the position A0 to the position A1, the corresponding direction of the gimbal and the face camera 203 rotates upward from Z0 by ⁇ 1 to reach the direction of Z1, and faces above the position of the human eye;
  • the positional relationship between the human eye and the camera shown in FIG. 10C (the second state of the angle adjustment of the gimbal and the face camera 203) is the initial state, and Z1 ′ represents the gimbal and the person
  • A1 ' represents the position of the face camera 203, and calculates the angle required to rotate the face camera 203 from the current position to the position that can be aligned with the center of the user's eyes, which is recorded as ⁇ 2, ⁇ 2> 0, this
  • the method of updating the control coefficient at the time is that k is reduced by a certain step, then ⁇ 2 ⁇ 2, the gimbal rotates down ⁇ 2, and the state shown in FIG.
  • the face camera 203 moves from the position A1 'to the position A2, and the gimbal and the face camera 203 rotate downward from Z1' by ⁇ 2 to reach the Z2 direction, toward the human eye position.
  • the PID algorithm calculates the actual rotation angle value of the gimbal according to the currently input rotation angle and the stored gimbal rotation angle, and smoothly controls the jump rotation angle data to make the cloud
  • the stage rotates smoothly, and the PID parameters depend on the rotation step of the servo, image resolution, and camera angle of view.
  • step S3 An example of the execution process of step S3):
  • step S2 Taking the positional relationship between the human eye and the camera shown in FIG. 10A as the initial state, the state shown in FIG. 10B is obtained after step S2), and then step S3) is performed, and the ⁇ 1 calculated in step S2) is smoothed Obtain the rotation angle ⁇ 1 ′,
  • Step S3) The calculated actual rotation angle of the gimbal is a relative angle.
  • the actual rotation angle is superimposed on the current absolute angle and recorded as the updated absolute angle. This value does not exceed the rotation angle limit of the gimbal.
  • step S4 Based on the current gimbal angle, adjust the pitch angle of the gimbal according to the actual rotation angle through the servo, and then return to step S1).
  • FIG. 12 is a schematic structural diagram of an apparatus for automatically tilting and adjusting the camera gimbal of the present disclosure, which may be a schematic structural diagram of an apparatus for automatically acquiring iris images of users of different heights.
  • the first acquisition unit is used to collect the iris image and transmit it to the main board, which is fixed in the middle of the gimbal and is an infrared camera;
  • the second acquisition unit is used to collect the face image and transmit it to the main board. It is closely arranged with the first acquisition unit and the distance does not exceed 5cm. It is a visible light camera;
  • the lighting unit is used to turn on the lighting when the first acquisition unit collects images, and is a set of two infrared lamp boards, which are symmetrically arranged on the left and right ends of the gimbal with the first acquisition unit as the center;
  • the angle adjustment unit is used to receive the angle adjustment command sent by the main board, control the gimbal to rotate and pitch, fixed on the gimbal, and the distance between the lighting unit and the lighting unit is not less than 0.1cm, to prevent the servo from blocking the light emitted by the infrared lamp, 360 ° rotating servo.
  • the first acquisition unit may be used as the first iris optical unit 107 in FIG. 1
  • the second acquisition unit may be used as the face optical unit 106 in FIG. 1.
  • the front of the device is covered with glass lenses, which are touch screens in the display area, and filters that transmit infrared light in front of the infrared light board.
  • the tilt rotation angle of the gimbal ranges from -20 ° to 20 °, and the horizontal direction is 0 °.
  • the calculation method of the pan / tilt rotation angle limit of the present disclosure is shown in FIG. 11, including the pan / tilt and camera, other units are not shown, f1, f2 indicate the best shortest and farthest collection distance, h indicates the pan / tilt pitch angle adjustment process
  • the adjustment distance in the vertical direction (up or down) d represents the range between the shortest distance and the longest distance of the image that can be collected at the current vertical distance h
  • ⁇ represents the pitch angle of the gimbal, according to the relationship between the above variables
  • the rotation angle of the camera should be between pitch 15 ° -25 °.
  • the maximum and minimum height H is related to the installation height PH of the equipment, the pitch rotation angle ⁇ of the servo, and the working distance L.
  • the algorithm will adjust the ⁇ value according to different heights, so the height that can be covered is:
  • H is the maximum value H max ; when ⁇ is a negative minimum value, H is the minimum value H min , and the maximum and minimum value of ⁇ is defined by the structural design of the device.
  • H ⁇ [H min , H max ] the adaptable height is between H min and H max , and the height difference range is H max -H min .
  • the device is turned on, the servo is initialized, and the initial position of the gimbal is recorded as an absolute angle.
  • the face camera captures the image in real time and transmits it to the motherboard.
  • the algorithm processing module detects the face camera.
  • the user's binocular position in the image calculate the longitudinal distance difference between the binocular position and the reference point, calculate the angle that the gimbal needs to rotate according to the control coefficient and the distance difference, input it to the PID module, and set the gimbal according to the preset parameters
  • the rotation angle needs to be smoothed, and ensure that the absolute rotation angle of the gimbal does not exceed the range of the gimbal rotation angle.
  • the gimbal rotation is achieved by controlling the servo, and then the algorithm processing module performs human eye detection on the next frame of image collected by the face camera , Cyclically execute the aforementioned process until the device is turned off.
  • the present disclosure is suitable for obtaining eye images of users of different heights.
  • the user does not need to bend down or step on the feet.
  • the device can automatically adjust the camera's pitch angle to realize the camera and the user's eyes Alignment between, obtain images containing human eye information, iris recognition.
  • the embodiments of the present disclosure provide a device for collecting high-definition facial images, including a first smart camera, a lighting unit, a distance sensor, and a gimbal.
  • the first smart camera includes a face optical unit, and An iris optical unit, algorithm module, and control module, in which:
  • the distance sensor is used to measure the object distance from the user to the first smart camera
  • the first iris optical unit performs automatic focusing according to the object distance
  • the first smart camera and the lighting unit are fixed on the gimbal;
  • the gimbal includes at least two motors that support horizontal rotation and vertical rotation, and are used to adjust the angle of the first smart camera and the lighting unit;
  • the face optical unit, the first iris optical unit and the first smart camera are connected through a MIPI interface;
  • the algorithm module performs face detection on the collected facial image, adjusts the rotation angle of the gimbal according to the face detection result, so that the face is located in the center of the image;
  • the algorithm module performs image quality judgment on the collected facial images, and transmits the facial images that meet the quality requirements to the back end for processing;
  • the first smart camera setting control module performs logic control on other peripheral devices included in the device.
  • the peripheral device is an optional device and is added according to actual needs. When the first smart camera cannot meet the recognition distance and recognition When angle is required, add peripheral device.
  • the face optical unit includes a sensor with a resolution of not less than 2M, an optical lens with a FOV of not less than 65 °, and the MTF value of the face optical unit in the acquisition range is not less than 50%@0.5lp / mm.
  • the first iris optical unit includes a sensor and a lens with a resolution of not less than 8M, the MTF value of the first iris optical unit in the acquisition range is not less than 50% @ 3lp / mm, and the pixel space The resolution is greater than 16pixel / mm.
  • the first iris optical unit may adopt a fixed focus structure or an auto focus motor.
  • the device further includes a camera array, the camera array includes an iris camera 1 and an iris camera 2, the first iris optical unit is an iris camera 1, and the second iris optical unit is an iris camera 2;
  • iris Camera 1 is the main camera, its lens image focal length is f ′ 1 , image sensor resolution is m 1 , object focal length is f 1 , and effective collection distance is
  • the device further includes a camera array, the camera array includes an iris camera 1 and an iris camera 3, the first iris optical unit is an iris camera 1, and the other peripheral device is an iris camera 3; an iris camera 1 is the main camera, the lens image focal length is f ′ 1 , the image sensor resolution is m 1 , the object side focal length is f 1 , and the effective collection distance is
  • the device further includes a camera array including an iris camera 1, an iris camera 2 and an iris camera 3, the first iris optical unit is an iris camera 1, and the second iris optical unit is an iris Camera 2, the other peripheral device is an iris camera 3;
  • iris camera 1 is the main camera, its lens image focal length is f ′ 1 , the image sensor resolution is m 1 , the object side focal length is f 1 , and the effective collection distance is
  • the iris camera 2 is a slave camera, its lens image focal length f ′ 2 , image sensor resolution m 2 , object focal length f 2 , effective collection distance is
  • control module is used to perform distance detection through the human eye detection algorithm and the corresponding distance lookup table after all iris cameras are initialized at the same time and start collecting iris images, and used to select the image collected by the corresponding iris camera according to the detected distance
  • the specific steps for selecting the corresponding iris camera according to the detected distance are:
  • control module is used to initialize the depth camera in the camera array and start collecting infrared scattergrams. After obtaining the distance detection result, select the corresponding iris camera according to the detected distance to start collecting images, and perform feature extraction and iris Registration, iris recognition processing;
  • the specific steps to select the corresponding iris camera according to the detected distance are:
  • control module is used to perform distance detection through the human eye detection algorithm and the corresponding distance lookup table after the face camera in the camera array starts to collect the face image, and select the corresponding iris camera to start collection according to the detected distance Image, feature extraction, iris registration, iris recognition processing;
  • the specific steps to select the corresponding iris camera according to the detected distance are:
  • the lighting unit includes one or more light-emitting components.
  • the light-emitting components mix multiple spectra, and adjust light intensity and angle according to distance information.
  • the light-emitting components are configured as LED or VCSEL laser lighting modules.
  • the distance sensor is set as a TOF laser ranging sensor, a 3D structured light ranging sensor, an infrared ranging sensor, an ultrasonic ranging sensor, or a lidar distance sensor.
  • the peripheral device includes a second smart camera
  • the second smart camera includes a second iris optical unit, which is controlled by the control module of the first smart camera, and works in coordination with the first smart camera, depending on the object Autofocus.
  • the second smart camera is fixed on the gimbal, and is connected to the second iris optical unit through a MIPI interface.
  • the second iris optical unit includes a sensor with a resolution of not less than 8M, and a lens, and the MTF value of the second iris optical unit in the acquisition range is not less than 50% @ 3lp / mm, of the pixel
  • the spatial resolution is greater than 16pixel / mm.
  • the second iris optical unit adopts a fixed-focus structure or an auto-focus motor.
  • the peripheral device may also configure more smart cameras according to actual application scenario requirements.
  • a method for collecting high-definition facial images includes:
  • the algorithm module detects the position and size of the face through the face detection algorithm, adjusts the pan / tilt angle according to the face position, and adjusts the face to the center of the entire image;
  • the algorithm module judges the quality of the iris image through the quality judgment algorithm, selects the iris image that meets the quality requirements, and transmits the corresponding facial image to the back end for image processing.
  • the device for collecting high-definition facial images of the present disclosure captures facial images through the facial optical unit, and at the same time enhances the processing capability of the first smart camera, can collect high-definition facial images to meet the needs of biological Recognition is easy to use and accurate; only high-definition facial images are transmitted, which effectively reduces the bandwidth requirements for data transmission; the device has a relatively flexible architecture, is easy to install and carry, and can add optical units and smart camera units as needed.
  • An embodiment of the present disclosure also provides an iris collection device that extends the collection distance, including a camera array and an infrared lighting unit, and is connected to a control module through a communication interface.
  • the camera array is configured with a master camera and at least one slave camera, both It is an iris camera, which is used to collect iris images.
  • the iris camera is equipped with lenses with different image square focal lengths or image sensors with different resolutions to achieve different object side focal lengths. There is a certain overlap in the collection distance range; the iris cameras are closely arranged The field of view is mostly coincident; the camera array is equipped with a control chip to communicate with the control module.
  • the control module independently controls the opening and closing of each iris camera; the infrared lighting units are distributed on the left and right sides or around the camera array;
  • the control module receives the image data collected by the camera array, performs distance detection, selects the image collected by the corresponding iris camera according to the distance detection result, or selects the corresponding iris camera, sends an image collection instruction to the control chip, and starts the corresponding iris camera Collect iris images to improve the focus of the device on targets at different distances and improve image quality.
  • the camera array further includes a depth camera, which is used to collect depth images, perform distance detection, and output the distance detection result.
  • a depth camera which is used to collect depth images, perform distance detection, and output the distance detection result.
  • the camera array further includes a face camera, which is used to collect face images and transmit them to the control module for distance detection.
  • the camera array includes two iris cameras, iris camera 1 is the main camera, the lens image focal length is f ′ 1 , the image sensor resolution is m 1 , the object side focal length is f 1 , and the effective collection distance is
  • the camera array includes two iris cameras, iris camera 1 is the main camera, the lens image focal length is f ′ 1 , the image sensor resolution is m 1 , the object side focal length is f 1 , and the effective collection distance is
  • the camera array includes three iris cameras, iris camera 1 is the main camera, the lens image focal length is f ′ 1 , the image sensor resolution is m 1 , the object side focal length is f 1 , and the effective collection distance is
  • the iris camera 2 is a slave camera, its lens image focal length f ′ 2 , image sensor resolution m 2 , object focal length f 2 , effective collection distance is
  • the effective collection distance of the iris camera 1 is 40 cm to 65 cm
  • the effective collection distance of the iris camera 2 is 60 cm to 85 cm
  • the effective collection distance of the iris camera 3 is 25 cm to 45 cm
  • the resolution of the image sensor of the iris camera 1-3 The rate is not less than 2 Mega Pixel.
  • An embodiment of the present disclosure also provides a method for expanding a collection distance according to the device, the device includes an iris camera 1, and an iris camera 2 and / or an iris camera 3, and the method for extending the collection distance includes the following specific steps:
  • the control module performs distance detection through the human eye detection algorithm and the corresponding distance lookup table
  • the control module selects the image collected by the corresponding iris camera according to the detected distance, and performs feature extraction, iris registration, and iris recognition processing.
  • An embodiment of the present disclosure also provides a method for expanding a collection distance according to the device, the device includes an iris camera 1, a depth camera, and an iris camera 2 and / or an iris camera 3, and the method for extending the collection distance includes the following specific steps :
  • control module selects the corresponding iris camera to start collecting images, transmits the images to the control module, and performs feature extraction, iris registration, and iris recognition processing.
  • An embodiment of the present disclosure also provides a method for expanding a collection distance according to the device, the device includes an iris camera 1, a face camera, and an iris camera 2 and / or an iris camera 3.
  • the method for extending the collection distance includes the following specific step:
  • the control module performs distance detection through the human eye detection algorithm and the corresponding distance lookup table
  • the control module selects the corresponding iris camera to start collecting images, transmits the images to the control module, and performs feature extraction, iris registration, and iris recognition processing.
  • a method for selecting a corresponding iris camera according to the detection distance includes the following specific steps:
  • step 3 select a method for collecting images by the corresponding iris camera according to the detection distance, including the following specific steps:
  • step 3 select a method for the corresponding iris camera to collect images according to the detection distance, including the following specific steps:
  • the extended iris collection device of the present disclosure includes at least two iris cameras.
  • the iris cameras are configured with lenses with different image square focal lengths or image sensors with different resolutions to achieve different object side focal lengths ;
  • the control module selects the image collected by each iris camera according to the distance detection result, or controls the iris camera to switch, and starts the corresponding iris camera to collect the image within a certain object distance range, so as to achieve multiple object distance ranges.
  • the technical effect of obtaining a clearly focused iris image can extend the effective collection distance to the range of 25cm-85cm.
  • the camera switching control method of the present disclosure can prevent the iris camera from frequently switching at the critical point of the effective collection range.
  • the embodiment of the present disclosure also provides a method for automatically adjusting the pitch of the camera gimbal, which is suitable for adjusting the gimbal in the device for collecting high-definition facial images, so as to realize the rotation in the vertical direction.
  • the automatic tilt adjustment method of the camera gimbal includes:
  • the first adjusted estimated angle is a rough estimate of the angle at which the camera should rotate to collect the face image
  • the first adjusted estimated angle and the control system calculate the second adjusted estimated angle of the gimbal.
  • the second adjusted estimated angle is an accurate estimate of the angle at which the camera should rotate to collect the face image;
  • step S4 Adjust the tilt angle of the gimbal according to the actual rotation angle based on the current gimbal angle, and then return to step S1).
  • the method for calculating the first adjusted estimated angle of the camera in step S1) is:
  • H1 is the pixel value of the vertical distance between the center of the two eyes in the face image and the preset reference point
  • D1 is the pixel value of the interpupillary distance in the face image
  • D2 is the interpupillary distance of the user's eyes
  • L is the horizontal distance between the user's eyes and the face camera, and is derived from the distance between the eyes of the eyes in the face image D1 and the distance lookup table.
  • step S2 the method for calculating the second adjusted estimated angle in step S2 is:
  • k> 0 represents the control coefficient
  • the initial value is set to 1, with a certain step adaptively increase and decrease adjustment.
  • the method for adaptively increasing or decreasing the control coefficient is that when the absolute value of the angle at which the face camera needs to be rotated this time is greater than the absolute value of the actual angle of rotation of the gimbal last time, the control coefficient k increases; When the absolute value of the angle that the second face camera needs to rotate is smaller than the absolute value of the actual rotation angle of the previous gimbal, the control coefficient k decreases; otherwise, the control coefficient remains unchanged.
  • the smoothing method of the gimbal rotation angle in step S3) is to use a PID control algorithm to calculate the actual gimbal rotation angle value based on the current gimbal rotation angle and the previous gimbal actual rotation angle. The data is smoothly controlled.
  • An embodiment of the present disclosure also provides a device for automatically adjusting the pitch of a camera gimbal, including a first acquisition unit for collecting an iris image and transmitting it to a main board, which is fixed in the middle of the gimbal and is an infrared camera;
  • the second acquisition unit is used to collect the face image and transmit it to the main board. It is closely arranged with the first acquisition unit and the distance does not exceed 5cm. It is a visible light camera;
  • the lighting unit is used to turn on the lighting when the first acquisition unit collects images, and is a set of two infrared lamp boards, which are symmetrically arranged on the left and right ends of the gimbal with the first acquisition unit as the center;
  • the angle adjustment unit is used to receive the angle adjustment command sent by the main board, control the gimbal to rotate and pitch, fixed on the gimbal, and the distance from the lighting unit is not less than 0.1cm.
  • the optical axes of the first acquisition unit and the second acquisition unit are parallel, and the angle of view of the second acquisition unit covers the angle of view of the first acquisition unit.
  • the present disclosure achieves the technical effect of expanding the perceivable range of the device. Compared with the traditional fixed-angle iris recognition device, it effectively expands the range of the iris image that the device can collect, and adaptively adjusts the tilt of the gimbal The angle brings great convenience to users of different heights.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Studio Devices (AREA)
  • Image Input (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

本公开提供了一种采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法，该装置包括第一智能相机、照明单元、用于测量用户到第一智能相机的物距的距离传感器及调节第一智能相机以及照明单元的角度的云台，第一智能相机包括人脸光学单元、根据物距进行自动对焦的第一虹膜光学单元、算法模块及控制模块。第一智能相机、照明单元固定于云台上；云台包括多个电机，支持水平方向旋转和垂直方向旋转；人脸光学单元、第一虹膜光学单元分别与第一智能相机通过MIPI接口连接；算法模块对采集的面部图像进行人脸检测，根据人脸检测结果调节云台旋转角度，使人脸位于图像中心，还对采集的面部图像进行图像质量判断，将符合要求的面部图像传输到后端处理。

Description

采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法
交叉引用
本公开要求申请号为CN 201811212252.5，申请日为2018年10月18日，公开名称为“一种采集高清晰度面部图像的装置”的中国专利申请，申请号为CN 201811472517.5，申请日为2018年12月04日，公开名称为“一种扩展采集距离的虹膜采集设备与方法”的中国专利申请，及申请号为CN 201910201929.3，申请日为2019年03月19日，公开名称为“一种摄像头云台自动俯仰调节的方法与装置”的中国专利申请，此处通过引用将其全部内容并入。
技术领域
本公开涉及生物特征采集识别领域，具体的涉及一种采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法。
背景技术
近年来，各种生物识别技术已经在安防、银行等领域得到广泛应用，尤其是人脸识别、虹膜识别两种技术的市场份额已经稳步提升，大有追赶指纹识别技术的趋势。人脸识别技术由于识别精度的原因，无法用于大规模的数据库；虹膜识别技术受制于用户易用性方面的问题，主要是用户的交互体验差，无法得到大规模的应用。而结合人脸识别和虹膜识别两种面部特征，达到易用性与安全性的完美结合，对推广非接触式生物特征识别技术有深远的影响。
发明内容
本公开提供了一种采集高清晰度面部图像的装置，包括第一智能相机，照明单元，距离传感器，云台，所述第一智能相机包括人脸光学单元，第一虹膜光学单元，算法模块，和控制模块，其中，
所述距离传感器用于测量用户到第一智能相机的物距；
所述第一虹膜光学单元根据物距进行自动对焦；
所述第一智能相机、照明单元固定于所述云台上；
所述云台包括至少两个电机，支持水平方向旋转和垂直方向旋转，用于调节第一智能相机以及照明单元的角度；
所述人脸光学单元，第一虹膜光学单元分别与第一智能相机通过MIPI接口连接；
所述算法模块对采集的面部图像进行人脸检测，根据人脸检测结果调节云台旋转角度，使人脸位于图像中心；
所述算法模块对采集的面部图像进行图像质量判断，将符合质量要求的面部图像传输到后端进行处理；
所述第一智能相机设置控制模块，对装置包含的其他外设装置进行逻辑控制，所述外设装置为可选装置。
本公开还提供了一种摄像头云台自动俯仰调节的方法，适用于所述装置，所述方法包括：
S1)根据人脸摄像头采集的人脸图像和预设参考点位置，计算人脸摄像头的第一调节估计角度，第一调节估计角度为摄像头为采集人脸图像应转动角度的粗略估计；
S2)由第一调节估计角度和控制系统计算云台的第二调节估计角度，第二调节估计角度为摄像头为采集人脸图像应转动角度的准确估计；
S3)对云台的第二调节估计角度进行平滑处理，计算云台实际旋转角度值；
S4)以当前云台角度为基准，按照实际旋转角度调节云台俯仰角度，之后返回到步骤S1)。
应当理解，前述大体的描述和后续详尽的描述均为示例性说明和解释，并不应当用作对本公开所要求保护内容的限制。
附图说明
参考随附的附图，本公开更多的目的、功能和优点将通过本公开实施方式的如下描述得以阐明，其中：
图1示意性示出本公开采集高清晰度面部图像的装置的示意图；
图2示意性示出本公开利用采集高清晰度面部图像装置采集高清晰面部图像的流程图；
图3示意性示出本公开具有摄像头阵列的虹膜采集设备的结构图；
图4示意性示出本公开虹膜摄像头101、102的视场区域；
图5示意性示出本公开虹膜摄像头101、人脸摄像头104的视场区域；
图6A示出本公开扩展采集距离的方法流程一；
图6B示出本公开扩展采集距离的方法流程二；
图6C示出本公开扩展采集距离的方法流程三；
图7示出本公开虹膜摄像头的选择方法；
图8所示为本公开的摄像头云台自动俯仰调节的方法流程图；
图9A～图9B所示为本公开的用户双眼与人脸摄像头之间的位置角度示意图；
图10A～图10D所示为本公开计算调节估计角度的云台与人脸摄像头状态图；
图11所示为本公开的云台俯仰角度限值计算方法；
图12所示为本公开的摄像头云台自动俯仰调节的装置结构示意图；
图13所示为本公开的适应最大身高到最小身高变化的云台角度调节范围。
具体实施方式
通过参考示范性实施例，本公开的目的和功能以及用于实现这些目的和功能的方法将得以阐明。然而，本公开并不受限于以下所公开的示范性实施例；可以通过不同形式来对其加以实现。说明书的实质仅仅是帮助相关领域技术人员综合理解本公开的具体细节。
在下文中，将参考附图描述本公开的实施例。在附图中，相同的附图标记代表相同或类似的部件，或者相同或类似的步骤。
通用的人脸图像采集使用的光学模块视场较大，采集范围也比较大，但是图像分辨率较低，识别精度比较低。通用的虹膜图像采集使用的光学模块视场角较小，采集范围小，图像的放大率较大，能够拍摄到眼睛区域的细致纹理信息，识别精度比较高。识别算法一般都是利用视频流进行图像质量判断，如果采集的图像是高清的图像，那么对信息传输的带宽要求比较高，对后端的计算资源要求也比较严格。
对此，本公开提供一种采集高清晰度面部图像的装置，可以同时采集高清晰度的人脸图像和虹膜图像，为后端的识别算法提供基础数据。
图1所示为本公开采集高清晰度面部图像的装置的示意图，如图1所示，本实施例采集高清晰度面部图像的装置包括第一智能相机101，距离传感器103，云台104，照明单元105，人脸光学单元106，第一虹膜光学单元107，算法模块(图中未显示)，和控制模块(图中未显示)；
本实施例还包括第二智能相机102，所述第二智能相机102为可选装置，当时所述第一智能相机无法满足识别距离扩大、识别角度扩宽的需求时，添加第二智能相机102，第二智能相机102相应的匹配第二虹膜光学单元108，其中第二智能相机102作为装置的从属智能相机，按照一定的逻辑与第一智能相机协调工作，用于需要扩展虹膜采集的范围。
所述距离传感器103用于测量用户到第一智能相机的物距，可以是TOF激光测距传感器、3D结构光测距传感器、红外测距传感器、超声波测距传感器或者激光雷达距离传感器。
所述第一虹膜光学单元107、第二虹膜光学单元108根据物距进行自动对焦。
所述第一智能相机101、第二智能相机102以及照明单元105固定在云台104上，所述云台104包括至少两个电机，支持水平方向旋转和垂直方向旋转，用于调节第一智能相机101、第二智能相机102以及照明单元105的角度。
所述照明单元105包括一个或多个发光元器件，发光元器件混合多种光谱，能够根据距离传感器103向第一智能相机101提供的距离信息进行光强调整。
所述发光元器件LED，VCSEL激光照明模块。
在人脸光学单元106捕捉人脸，第一智能相机101根据捕捉的人脸采集用户面部图像时，本实施例的装置通过调整云台104，使得人脸最终处于整个图像中心。
所述人脸光学单元106，第一虹膜光学单元107分别与第一智能相机101通过MIPI接口连接，所述第二虹膜光学单元108分别与第二智能相机102通过MIPI接口连接。
所述人脸光学单元106包括不低于2M分辨率的传感器，FOV不小于65°的光学镜头，所述人脸光学单元在采集范围内的MTF值不低于50％@0.5lp/mm。
所述第一虹膜光学单元107、第二虹膜光学单元108包括不低于8M分辨率的传感器，和镜头，所述第一虹膜光学单元107、第二虹膜光学单元108在采集范围内的MTF值不低于50％@3lp/mm，像素的空间分辨率大于16pixel/mm。
所述第一虹膜光学单元107、第二虹膜光学单元108可以采用定焦结构或自动对焦马达。
所述第一智能相机101作为主智能相机，内部设置算法模块，具备图像处理能力，能够对采集到的虹膜图像进行质量判断，设置阈值，当第一智能相机检测到采集的图像质量达到预先设置的阈值，则将与之对应的面部图像传输到后台进行虹膜注册、虹膜识别；否则自动丢弃此次采集的虹膜与人脸图像，继续采集。
所述第一智能相机101包括控制模块，能够对第二智能相机102进行逻辑控制。
图2示意性示出本公开采用所述装置采集高清晰面部图像的流程图，如图2所示，具体包括以下步骤：
S1：启动采集过程；
S2：第一智能相机的人脸光学单元采集面部图像；
S3：判断人脸是否在目标区域；若是，进行步骤S5，若否，进行步骤S4；
S4：调整云台位置；
第一智能相机从图像中检测出人脸的位置与大小，根据人脸位置控制云台，将人脸调整到整个图像的中心位置附近；
S5：读取距离信息；
距离传感器向第一智能相机提供用户到第一智能相机的物距；
S6：选择虹膜光学单元、调节焦距；
根据用户人脸大小、或者距离传感器测量的物距来确定使用第一虹膜光学单元或者第二虹膜光学单元，调节焦距，使第一智能相机对准人脸的眉心区域，持续跟踪；同时控制照明单元的光强，保证在不同的距离下都获得比较类似的照明条件；
S7：读取虹膜图像；
S8：判断虹膜图像是否满足质量要求，若是，进行步骤S9，若否，返回步骤S5，重新读取距离信息；
第一智能相机设置算法模块，对采集到的图像进行质量判断，设置阈值，判断是否符合质量要求；
S9：推送高清晰度面部图像；
第一智能相机持续进行自动对焦和图像质量判断，提取清晰的图像(人脸和眼纹)推送给后端图像处理平台；
S10：采集过程结束。
本公开提供了一种采集高清晰度面部图像的装置，在采集设备前端除了进行数据采集之外，通过主智能相机通过质量判断的图像传输到后台处理，从而减轻传输带宽压力；本公开在能够采集高清晰度的面部图像，同时满足生物识别的易用性和准确度。
虹膜识别是一种基于人眼虹膜区域的细节特征的生物识别技术，因其具有高精确度和唯一性越来越受到信息安全领域的关注。传统的虹膜识别装置，采用单一的摄像头，用于逐次捕捉用户双眼的虹膜图像，无法满足多距离，多视角的虹膜采集或是采用双摄 像头，用于同时捕捉用户双眼的虹膜图像，但是当捕捉的虹膜图像不符合要求，例如由于光照角度的问题，采集的虹膜图像质量不高，就得重新进行采集，在这种情况下，其实下一次的采集，虹膜图像质量也是不确定的。
对此，本公开提供一种扩展采集距离的虹膜采集设备与方法，可以实现在更广泛的采集距离范围内采集到对焦清晰的虹膜图像，为虹膜识别提供高质量的图像。
图3所示为本公开虹膜采集设备的结构框图，如图3所示，201为主摄像头，用于采集虹膜图像；202为从摄像头，用于采集虹膜图像；203为从摄像头，用于采集人脸图像；204为深度摄像头，2041为人脸摄像头，2042为红外摄像头，2043为红外发射器，2042接收2043发射的红外散点，感知目标深度信息，从而输出目标与采集设备之间的距离；2051、2052为左、右红外灯板，用于在201、202摄像头采集虹膜图像时进行补光；206为可见光灯板，用于在环境光较暗时补充可见光。
虹膜摄像头1(可以为图1中的第一虹膜光学单元107)为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000001
虹膜摄像头2(可以为图1中的第二虹膜光学单元108)为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
Figure PCTCN2019111758-appb-000002
满足f′ 1＜f′ 2，m 1＝m 2，f 1＜f 2，
Figure PCTCN2019111758-appb-000003
虹膜摄像头201、202、人脸摄像头203横向紧密排列，虹膜摄像头201和202的视场范围大部分重合。图4示出了虹膜摄像头201、202的视场区域A1、A2，虹膜摄像头201在采集距离为D1范围内时均能采集到对焦清晰的虹膜图像，虹膜摄像头202在采集距离为D2范围内时均能采集到对焦清晰的虹膜图像，可以看出人眼在区域E1、E2中前后移动，通过虹膜摄像头201、202切换，均可获取对焦清晰的虹膜图像，由此虹膜摄像头201、202组合而成的摄像头阵列可以有效的扩展有效采集距离。
虹膜摄像头201、人脸摄像头2041纵向垂直紧密排列，其光轴方向保持一致，图5示出了虹膜摄像头201、人脸摄像头2041的视场区域，人脸摄像头2041位于虹膜摄像头201上方，也可以设置于虹膜摄像头201下方，虹膜摄像头201的视场范围A1与人脸摄像头2041的视场范围A4在虹膜摄像头201的有效采集距离范围D1内重合。与横向排列相比，两种摄像头采集的图像中，人眼的相对水平位置保持一致，用户可以根据人脸摄像头2041采集的可见光图像中眼睛的位置来进行左右调节，以达到在虹膜摄像头201的成像中人眼处于图像的水平中心的效果。
摄像头阵列与控制芯片连接，通过通信接口接收控制模块的指令，启动相应摄像头采集图像。
红外灯板2051、2052发射光与虹膜摄像头201光轴之间存在一定夹角，不低于5°。
图6A为本公开扩展采集距离的方法流程一，可为长距离虹膜采集方法流程，所述设备包括虹膜摄像头1、以及虹膜摄像头2和/或虹膜摄像头3(可以为图1所示装置中的其他外设装置)，包括下列具体步骤：
1)同时初始化摄像头阵列中的所有虹膜摄像头，开始采集虹膜图像，通过通讯接口传输到控制模块；
2)控制模块通过人眼检测算法以及相应的距离查找表进行距离检测；
3)控制模块根据检测的距离，选择相应虹膜摄像头采集的图像，进行特征提取、虹膜注册、虹膜识别处理。
图6B为本公开扩展采集距离的方法流程二，可为长距离虹膜采集方法流程，所述设备包括虹膜摄像头1、深度摄像头以及虹膜摄像头2和/或虹膜摄像头3，包括下列具体步骤：
1)初始化摄像头阵列中的深度摄像头，开始采集红外散点图，得到距离检测结果，通过通讯接口传输到控制模块；
2)控制模块根据检测的距离，选择相应的虹膜摄像头开始采集图像，将图像传输到控制模块，进行特征提取、虹膜注册、虹膜识别处理。
图6C为本公开扩展采集距离的方法流程三，可为长距离虹膜采集方法流程，所述设备包括虹膜摄像头1、人脸摄像头以及虹膜摄像头2和/或虹膜摄像头3，包括下列具体步骤：
1)初始化摄像头阵列中的人脸摄像头，开始采集人脸图像，通过通讯接口传输到控制模块；
2)控制模块通过人眼检测算法以及相应的距离查找表进行距离检测；
3)控制模块根据检测的距离，选择相应的虹膜摄像头开始采集图像，将图像传输到控制模块，进行特征提取、虹膜注册、虹膜识别处理。
图7示出本公开虹膜摄像头的选择方法，以设备包含3颗虹膜摄像头为例，包含两颗摄像头的情况以此类推。D1、D2、D3表示虹膜摄像头1、2、3的有效采集距离范围，
Figure PCTCN2019111758-appb-000004
表示虹膜摄像头1有效采集距离的起始值、终止值，
Figure PCTCN2019111758-appb-000005
表示虹膜摄像头2有效采集距离的起始值、终止值，
Figure PCTCN2019111758-appb-000006
表示虹膜摄像头3有效采集距离的起始值、终止 值，
Figure PCTCN2019111758-appb-000007
表示虹膜摄像头1、3重叠的有效采集距离的起始值、终止值，
Figure PCTCN2019111758-appb-000008
表示虹膜摄像头1、2重叠的有效采集距离的起始值、终止值，
Figure PCTCN2019111758-appb-000009
当检测距离为
Figure PCTCN2019111758-appb-000010
时，选择虹膜摄像头3；当检测距离为
Figure PCTCN2019111758-appb-000011
时，选择虹膜摄像头1；当检测距离为
Figure PCTCN2019111758-appb-000012
时，选择虹膜摄像头2；当检测距离为
Figure PCTCN2019111758-appb-000013
时，保持之前选择的虹膜摄像头；当检测距离为
Figure PCTCN2019111758-appb-000014
时，保持之前选择的虹膜摄像头。
传统方法为多颗摄像头有效采集距离首尾衔接，不存在重叠范围，当用户位于有效采集距离的临界处，由于距离检测算法可能存在一定误差，而检测到的距离上下浮动，极有可能导致虹膜摄像头频繁切换现象，影响设备稳定。而本公开的多颗虹膜摄像头有效采集距离存在一定重叠，与传统的多颗摄像头有效采集距离首尾衔接的方法相比，能够有效的避免在临界距离处发生的边界震荡问题，防止虹膜摄像头出现无规律的频繁切换现象，可以实现虹膜摄像头平稳切换。
实施例1
虹膜采集设备包括虹膜摄像头1、虹膜摄像头2以及深度摄像头，当用户由远及近走近设备，深度摄像头实时检测距离，向控制模块输出检测结果，当距离从
Figure PCTCN2019111758-appb-000015
到
Figure PCTCN2019111758-appb-000016
时，选择虹膜摄像头2，当距离从
Figure PCTCN2019111758-appb-000017
到
Figure PCTCN2019111758-appb-000018
时，保持虹膜摄像头2，当距离从
Figure PCTCN2019111758-appb-000019
到
Figure PCTCN2019111758-appb-000020
时，选择虹膜摄像头1。
实施例2
虹膜采集设备包括虹膜摄像头1、虹膜摄像头2以及人脸摄像头，当用户由近及远离开设备，人脸摄像头实时采集人脸图像，传输到控制模块进行人眼检测并检测距离，当距离从
Figure PCTCN2019111758-appb-000021
到
Figure PCTCN2019111758-appb-000022
时，选择虹膜摄像头1，当距离从
Figure PCTCN2019111758-appb-000023
到
Figure PCTCN2019111758-appb-000024
时，保持虹膜摄像头1，当距离从
Figure PCTCN2019111758-appb-000025
到
Figure PCTCN2019111758-appb-000026
时，选择虹膜摄像头2。
传统虹膜识别设备只设置一颗红外摄像头，在安装后其位置、角度保持固定，红外摄像头只能扫描到一定范围之内的物体，扫描范围受限，当用户站在设备前方最佳采集距离处时，身高过高或过矮都可能造成设备无法扫描到用户双眼，不利于虹膜识别顺利进行。
对此，本公开提出根据人脸摄像头采集图像自适应调节摄像头云台俯仰角度的方法，可以通过舵机控制云台俯仰转动，实现获取不同身高用户的虹膜图像的效果。本公开各实施例所述的摄像头云台自动俯仰调节的方法可以用于实现图1中云台104在垂直方向上旋转。
一.摄像头云台自动俯仰调节的方法
图8所示为本公开的摄像头云台自动俯仰调节的方法流程图，具体地，为自动获取不同身高用户虹膜图像的方法流程图，通过调节人脸摄像头对准参考点位置，使得虹膜摄像头对准用户双眼中心，保证人眼在虹膜摄像头视场范围内成像，具体步骤为：
S1)根据人脸摄像头采集的人脸图像和预设参考点位置，计算人脸摄像头的第一调节估计角度。
图9A为用户双眼与人脸摄像头203之间的位置角度示意图，图9B为图9A所示场景人脸摄像头203采集图像示意图，如图9A～图9B所示，首先计算用户双眼中心位置与人脸摄像头203光轴之间的垂直高度H2，公式为：
Figure PCTCN2019111758-appb-000027
其中，D2为用户双眼瞳距，取值6cm-8cm，D1为人脸摄像头203采集图像中的双眼瞳距；
之后由H2计算出人脸摄像头203从当前位置旋转到能够对准用户双眼中心的位置所需要转动的角度，公式为：
α＝tan -1(H2/L)
L为用户双眼与人脸摄像头203之间的水平距离，由算法处理模块根据人脸摄像头203采集图像中的双眼瞳距D1计算得出。
S2)由第一调节估计角度和控制系统计算云台的第二调节估计角度。
图10A为云台与人脸摄像头203初始状态示意图，如图10A所示，O点为旋转中心，表示以云台为基准的旋转点，A0表示人脸摄像头203的位置，虚弧线A表示人脸摄像头203的旋转轨迹，是一段以O为圆心、人脸摄像头203与云台之间的距离为半径的圆弧，云台、人脸摄像头203初始位置对应方向为水平方向Z0，人脸摄像头203从初始方向Z0到对准眼睛X1方向变化角度为，而这一过程中云台与人脸摄像头203以云台为原点、从初始方向Z0旋转到对准眼睛Y1方向的变化角度为，两个旋转角度存在以下的对应关系：
θ＝kα
其中k＞0，代表控制系数，初始值设置为1，以一定的步长自适应的增减调整，其调整方法为：当计算出本次人脸摄像头203需要旋转的角度的绝对值大于上一次云台实际旋转角度的绝对值时，控制系数增大；当计算出本次人脸摄像头203需要旋转的角度的绝对值小于上一次云台实际旋转角度的绝对值时，控制系数减小；否则控制系数不变；
举例说明步骤S2)的执行方法(假设这一过程中用户与人脸摄像头203保持静止)：
在计算第一调节估计角度过程中，以图10A所示的人眼与摄像头之间的位置关系为初始状态，计算人脸摄像头203从当前位置旋转到能够对准用户双眼中心的位置所需要转动的角度，记为α1，α1＜0，在k＝1时，θ1＝α1，云台向上旋转θ1，得到图10B(为云台与人脸摄像头203角度调节状态一)所示状态，人脸摄像头203从位置A0移到位置A1，云台和人脸摄像头203对应方向从Z0向上旋转θ1到达Z1方向，朝向人眼位置上方；
在计算第二调节估计角度过程中，以图10C(为云台与人脸摄像头203角度调节状态二)所示的人眼与摄像头之间的位置关系为初始状态，Z1’表示云台、人脸摄像头203的对应方向，A1’表示人脸摄像头203的位置，计算人脸摄像头203从当前位置旋转到能够对准用户双眼中心的位置所需要转动的角度，记为α2，α2＞0，此时更新控制系数的方法为k以一定的步长减小，则θ2＜α2，云台向下旋转θ2，得到图10D(为云台与人脸摄像头203角度调节状态三)所示状态，人脸摄像头203从位置A1’移到位置A2，云台和人脸摄像头203对应方向从Z1’向下旋转θ2到达Z2方向，朝向人眼位置。
S3)对云台的第二调节估计角度进行平滑处理，计算云台实际旋转角度值。
将第二调节估计角度输入PID算法，PID算法根据当前输入的旋转角度与存储的上一次的云台旋转角度，计算云台实际旋转角度值，对跃变的旋转角度数据进行平滑控制，使云台平滑旋转，PID参数根据舵机旋转步长、图像分辨率、摄像头视场角而定。
举例说明步骤S3)的执行过程：
以图10A所示的人眼与摄像头之间的位置关系为初始状态，经过步骤S2)处理后得到图10B所示状态，之后执行步骤S3)，对步骤S2)计算得出的θ1进行平滑处理得到旋转角度θ1′，|θ1′|＜|θ1|，人脸摄像头203从位置A1移到位置A1’，云台和人脸摄像头203对应方向从Z1向上旋转θ1′到达Z1’方向，防止跃变现象，处理后得到图10C所示状态；
步骤S3)计算得出的云台实际旋转角度值为相对角度，将实际旋转角度值与当前绝对角度叠加，记为更新的绝对角度，该值不超过云台的旋转角度限值。
S4)以当前云台角度为基准，通过舵机按照实际旋转角度调节云台俯仰角度，之后返回到步骤S1)。
二.摄像头云台自动俯仰调节的装置结构
图12所示为本公开摄像头云台自动俯仰调节的装置结构示意图，可以为自动获取不同身高用户虹膜图像的设备结构示意图。
第一获取单元，用于采集虹膜图像并传输到主板，固定在云台中部，为红外摄像头；
第二获取单元，用于采集人脸图像并传输到主板，与第一获取单元紧密排列，间距不超过5cm，为可见光摄像头；
照明单元，用于在第一获取单元采集图像时开启照明，为两组红外灯板，以第一获取单元为中心对称设置在云台左、右两端；
角度调节单元，用于接收主板发出的角度调节指令，控制云台进行俯仰旋转，固定在云台上，与照明单元之间的距离不小于0.1cm，避免舵机遮挡红外灯发射的光线，为360°旋转舵机。
其中，第一获取单元可以用作图1中的第一虹膜光学单元107，第二获取单元可以用作图1中的人脸光学单元106。
上述各个单元与设备的前后面板之间留有一定空间，设备正面覆盖玻璃镜片，所述玻璃镜片在显示屏区域为触摸屏，在红外灯板前方为透射红外光的滤光片。云台俯仰旋转角度范围覆盖-20°～20°，以水平方向为0°基准。
本公开云台俯仰旋转角度限值的计算方法如图11，包括云台与摄像头，其他单元未显示，f1、f2表示最佳的最近、最远采集距离，h表示云台俯仰角度调节过程中垂直方向(向上或者向下)的调节距离，d表示当前垂直距离h处可采集图像的最近距离到最远距离之间的范围，α表示云台的俯仰角度，根据上述变量之间的关系，在采集距离为45cm-65cm范围内，为满足15-25cm的俯仰高度，和前后5-10cm的可采集范围，摄像头的旋转角度应处于俯仰15°-25°之间。
三.适应最大身高到最小身高变化的云台角度调节范围
如图13所示，最大最小身高H和设备的安装高度PH、舵机的俯仰旋转角度θ、以及工作距离L相关，算法会根据不同身高调整θ值，因此能覆盖到的高度为：
H＝PH+L*tan(θ)；
当θ为正的最大值时，H为最大值H max；当θ为负的最小值时，H值为最小值H min，且θ的最大最小值由设备的结构设计限定。
因此H∈[H min,H max]，可适应的身高在H在H min和H max之间，身高差范围为H max-H min。
例如设备θ∈[-20，20]，当设备安装高度为PH为160cm，工作距离在55cm-60cm处时，H min可达到140cm，H max可达到180cm，可以覆盖身高落差40cm的范围，满足绝大部分身高的用户。
本公开虹膜识别设备的工作原理：
设备处于开启状态，初始化舵机，记录云台初始位置为绝对角度，人脸摄像头实时拍摄图像并传输到主板，当用户走入设备有效采集距离范围之内，算法处理模块检测到人脸摄像头拍摄的图像中的用户双眼位置，计算双眼位置与参考点之间的纵向距离差值，根据控制系数以及距离差值计算出云台需要旋转的角度，输入到PID模块，根据预设参数对云台需要旋转的角度进行平滑处理，并确保云台绝对旋转角度不超过云台旋转角度范围，通过控制舵机实现云台旋转，之后算法处理模块对人脸摄像头采集的下一帧图像进行人眼检测，循环执行前述过程，直到设备关闭。
本公开适用于获取不同身高用户的眼部图像，用户无需弯腰或者垫脚，只要与设备之间的距离处于有效采集范围之内，设备即可自动调节摄像头的俯仰角度，实现摄像头与用户双眼之间的对准，获取包含人眼信息的图像，进行虹膜识别。
综上所述，本公开实施例提供了一种采集高清晰度面部图像的装置，包括第一智能相机，照明单元，距离传感器，云台，所述第一智能相机包括人脸光学单元，第一虹膜光学单元，算法模块，和控制模块，其中：
所述距离传感器用于测量用户到第一智能相机的物距；
所述第一虹膜光学单元根据物距进行自动对焦；
所述第一智能相机、照明单元固定于所述云台上；
所述云台包括至少两个电机，支持水平方向旋转和垂直方向旋转，用于调节第一智能相机以及照明单元的角度；
所述人脸光学单元，第一虹膜光学单元分别与第一智能相机通过MIPI接口连接；
所述算法模块对采集的面部图像进行人脸检测，根据人脸检测结果调节云台旋转角度，使人脸位于图像中心；
所述算法模块对采集的面部图像进行图像质量判断，将符合质量要求的面部图像传输到后端进行处理；
所述第一智能相机设置控制模块，对装置包含的其他外设装置进行逻辑控制，所述外设装置为可选装置，根据实际需求添加，当所述第一智能相机不能满足识别距离、识别角度需求时，添加外设装置。
可选的，所述人脸光学单元包括不低于2M分辨率的传感器，FOV不小于65°的光学镜头，所述人脸光学单元在采集范围内的MTF值不低于50％@0.5lp/mm。
可选的，所述第一虹膜光学单元包括不低于8M分辨率的传感器和镜头，所述第一虹膜光学单元在采集范围内的MTF值不低于50％@3lp/mm，像素的空间分辨率大于16pixel/mm。
可选的，所述第一虹膜光学单元可采用定焦结构或自动对焦马达。
可选的，所述装置还包括摄像头阵列，所述摄像头阵列包括虹膜摄像头1和虹膜摄像头2，所述第一虹膜光学单元为虹膜摄像头1，所述第二虹膜光学单元为虹膜摄像头2；虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000028
虹膜摄像头2为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
Figure PCTCN2019111758-appb-000029
满足f′ 1＜f′ 2，m 1＝m 2，f 1＜f 2，
Figure PCTCN2019111758-appb-000030
可选的，所述装置还包括摄像头阵列，所述摄像头阵列包括虹膜摄像头1和虹膜摄像头3，所述第一虹膜光学单元为虹膜摄像头1，所述其他外设装置为虹膜摄像头3；虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000031
虹膜摄像头3为从摄像头，其镜头像方焦距为f′ 3、图像传感器分辨率为m 3、物方焦距为f 3、有效采集距离为
Figure PCTCN2019111758-appb-000032
f′ 1＝f′ 3，m 1＞m 3，f 1＞f 3，
Figure PCTCN2019111758-appb-000033
可选的，所述装置还包括摄像头阵列，所述摄像头阵列包括虹膜摄像头1、虹膜摄像头2及虹膜摄像头3，所述第一虹膜光学单元为虹膜摄像头1，所述第二虹膜光学单元为虹膜摄像头2，所述其他外设装置为虹膜摄像头3；虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000034
虹膜摄像头2为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
Figure PCTCN2019111758-appb-000035
虹膜摄像头3为从摄像头，其镜头像方焦距为f′ 3、图像传感器分辨率为m 3、物方焦距为f 3、有效采集距离为
Figure PCTCN2019111758-appb-000036
f′ 1＝f′ 3＜f′ 2，m 1＝m 2＞m 3，f 3＜f 1＜f 2，
Figure PCTCN2019111758-appb-000037
可选的，控制模块用于在所有虹膜摄像头被同时初始化并开始采集虹膜图像之后，通过人眼检测算法以及相应的距离查找表进行距离检测，并用于根据检测的距离选择相应虹膜摄像头采集的图像，进行特征提取、虹膜注册、虹膜识别处理；其中，根据检测的距离选择相应虹膜摄像头的具体步骤为：
1)当检测距离为
Figure PCTCN2019111758-appb-000038
时，选择虹膜摄像头1；
2)当检测距离为
Figure PCTCN2019111758-appb-000039
时，选择虹膜摄像头2；
3)当检测距离为
Figure PCTCN2019111758-appb-000040
时，保持之前选择的虹膜摄像头；
其中
Figure PCTCN2019111758-appb-000041
表示虹膜摄像头1有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000042
表示虹膜摄像头2有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000043
可选的，控制模块用于在摄像头阵列中的深度摄像头被初始化并开始采集红外散点图，得到距离检测结果之后，根据检测的距离选择相应的虹膜摄像头开始采集图像，并进行特征提取、虹膜注册、虹膜识别处理；
其中，根据检测的距离选择相应虹膜摄像头的具体步骤为：
1)当检测距离为
Figure PCTCN2019111758-appb-000044
时，选择虹膜摄像头3；
2)当检测距离为
Figure PCTCN2019111758-appb-000045
时，选择虹膜摄像头1；
3)当检测距离为
Figure PCTCN2019111758-appb-000046
时，保持之前选择的虹膜摄像头；
其中
Figure PCTCN2019111758-appb-000047
表示虹膜摄像头1有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000048
表示虹膜摄像头3有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000049
可选的，控制模块用于在摄像头阵列中的人脸摄像头开始采集人脸图像之后，通过人眼检测算法以及相应的距离查找表进行距离检测，并根据检测的距离选择相应的虹膜摄像头开始采集图像，进行特征提取、虹膜注册、虹膜识别处理；
其中，根据检测的距离选择相应虹膜摄像头的具体步骤为：
1)当检测距离为
Figure PCTCN2019111758-appb-000050
时，选择虹膜摄像头3；
2)当检测距离为
Figure PCTCN2019111758-appb-000051
时，选择虹膜摄像头1；
3)当检测距离为
Figure PCTCN2019111758-appb-000052
时，选择虹膜摄像头2；
4)当检测距离为
Figure PCTCN2019111758-appb-000053
时，保持之前选择的虹膜摄像头；
5)当检测距离为
Figure PCTCN2019111758-appb-000054
时，保持之前选择的虹膜摄像头；
其中
Figure PCTCN2019111758-appb-000055
表示虹膜摄像头1有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000056
表示虹膜摄像头2有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000057
表示虹膜摄像头3有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000058
可选的，所述照明单元包括一个或多个发光元器件，所述发光元器件混合多种光谱，根据距离信息进行光强和角度的调整。
可选的，所述发光元器件设置为LED或VCSEL激光照明模块。
可选的，所述距离传感器设置为TOF激光测距传感器、3D结构光测距传感器、红外测距传感器、超声波测距传感器或者激光雷达距离传感器。
可选的，所述外设装置包括第二智能相机，所述第二智能相机包括第二虹膜光学单元，由所述第一智能相机的控制模块控制，与第一智能相机协调工作，根据物距进行自动对焦。
可选的，所述第二智能相机固定于所述云台上，与第二虹膜光学单元通过MIPI接口连接。
可选的，所述第二虹膜光学单元包括不低于8M分辨率的传感器，和镜头，所述第二虹膜光学单元在采集范围内的MTF值不低于50％@3lp/mm，像素的空间分辨率大于16pixel/mm。
可选的，所述第二虹膜光学单元采用定焦结构或自动对焦马达。
可选的，所述外设装置还可以根据实际应用场景需求配置更多的智能相机。
可选的，一种采集高清晰度面部图像的方法，包括：
1)使用人脸光学单元采集面部图像，算法模块通过人脸检测算法检测出人脸的位置与大小，根据人脸位置调节云台角度，将人脸调整到整个图像的中心位置；
2)根据人脸大小、或者距离传感器测量的物距选择触发第一虹膜光学单元或者第二虹膜光学单元，并调节焦距，同时控制照明单元光强；
3)算法模块通过质量判断算法对虹膜图像进行质量判断，选择符合质量要求的虹膜图像，将与其对应的面部图像传输到后端进行图像处理。
本公开的有益效果：本公开的采集高清晰度面部图像的装置，通过人脸光学单元捕捉人脸图像，同时加强了第一智能相机的处理能力，能够采集高清晰度的面部图像，满足生物识别的易用性和准确度；只传输高清晰度的面部图像，有效降低了数据传输的带宽要求；本装置架构比较灵活，安装和携带方便，可以根据需要增加光学单元和智能相机单元。
本公开实施例还提供了一种扩展采集距离的虹膜采集设备，包括摄像头阵列、红外照明单元，通过通讯接口与控制模块连接，所述摄像头阵列配置一颗主摄像头与至少一颗从摄像头，均为虹膜摄像头，用于采集虹膜图像，所述虹膜摄像头配置不同像方焦距的镜头、或者不同分辨率的图像传感器，以达到不同的物方焦距，其采集距离范围存在一定重叠；虹膜摄像头紧密排列，视场范围大部分重合；摄像头阵列配置控制芯片，与控制模块进行数据通讯，控制模块独立控制每颗虹膜摄像头的开启与关闭；所述红外照 明单元分布在摄像头阵列的左右两侧或者四周；所述控制模块接收摄像头阵列采集的图像数据，进行距离检测，根据距离检测结果选择相应虹膜摄像头采集的图像，或者选择相应虹膜摄像头，向控制芯片发送采集图像指令，启动相应虹膜摄像头采集虹膜图像，提升设备对不同距离目标的对焦程度，提高图像质量。
可选的，所述摄像头阵列还包括深度摄像头，用于采集深度图像，进行距离检测，输出距离检测结果。
可选的，所述摄像头阵列还包括人脸摄像头，用于采集人脸图像，传输到控制模块进行距离检测。
可选的，所述摄像头阵列包括两颗虹膜摄像头，虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000059
虹膜摄像头2为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
Figure PCTCN2019111758-appb-000060
满足f′ 1＜f′ 2，m 1＝m 2，f 1＜f 2，
Figure PCTCN2019111758-appb-000061
可选的，所述摄像头阵列包括两颗虹膜摄像头，虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000062
虹膜摄像头3为从摄像头，其镜头像方焦距为f′ 3、图像传感器分辨率为m 3、物方焦距为f 3、有效采集距离为
Figure PCTCN2019111758-appb-000063
f′ 1＝f′ 3，m 1＞m 3，f 1＞f 3，
Figure PCTCN2019111758-appb-000064
可选的，所述摄像头阵列包括三颗虹膜摄像头，虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
Figure PCTCN2019111758-appb-000065
虹膜摄像头2为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
Figure PCTCN2019111758-appb-000066
虹膜摄像头3为从摄像头，其镜头像方焦距为f′ 3、图像传感器分辨率为m 3、物方焦距为f 3、有效采集距离为
Figure PCTCN2019111758-appb-000067
f′ 1＝f′ 3＜f′ 2，m 1＝m 2＞m 3，f 3＜f 1＜f 2，
Figure PCTCN2019111758-appb-000068
可选的，虹膜摄像头1的有效采集距离为40cm～65cm，虹膜摄像头2的有效采集距离为60cm～85cm，虹膜摄像头3的有效采集距离为25cm～45cm，虹膜摄像头1-3的图像传感器的分辨率不低于2Mega Pixel。
本公开实施例还提供了一种根据所述设备的扩展采集距离的方法，所述设备包括虹膜摄像头1、以及虹膜摄像头2和/或虹膜摄像头3，扩展采集距离的方法包括下列具体步骤：
1)同时初始化摄像头阵列中的所有虹膜摄像头，开始采集虹膜图像，通过通讯接口传输到控制模块；
2)控制模块通过人眼检测算法以及相应的距离查找表进行距离检测；
3)控制模块根据检测的距离，选择相应虹膜摄像头采集的图像，进行特征提取、虹膜注册、虹膜识别处理。
本公开实施例还提供了一种根据所述设备的扩展采集距离的方法，所述设备包括虹膜摄像头1、深度摄像头以及虹膜摄像头2和/或虹膜摄像头3，扩展采集距离的方法包括下列具体步骤：
1)初始化摄像头阵列中的深度摄像头，开始采集红外散点图，得到距离检测结果，通过通讯接口传输到控制模块；
2)控制模块根据检测的距离，选择相应的虹膜摄像头开始采集图像，将图像传输到控制模块，进行特征提取、虹膜注册、虹膜识别处理。
本公开实施例还提供了一种根据所述设备的扩展采集距离的方法，所述设备包括虹膜摄像头1、人脸摄像头以及虹膜摄像头2和/或虹膜摄像头3，扩展采集距离的方法包括下列具体步骤：
1)初始化摄像头阵列中的人脸摄像头，开始采集人脸图像，通过通讯接口传输到控制模块；
2)控制模块通过人眼检测算法以及相应的距离查找表进行距离检测；
3)控制模块根据检测的距离，选择相应的虹膜摄像头开始采集图像，将图像传输到控制模块，进行特征提取、虹膜注册、虹膜识别处理。
可选的，所述设备包括虹膜摄像头1、虹膜摄像头2时，步骤3)根据检测距离选择相应虹膜摄像头的方法，包括下列具体步骤：
1)当检测距离为
Figure PCTCN2019111758-appb-000069
时，选择虹膜摄像头1；
2)当检测距离为
Figure PCTCN2019111758-appb-000070
时，选择虹膜摄像头2；
3)当检测距离为
Figure PCTCN2019111758-appb-000071
时，保持之前选择的虹膜摄像头；
其中
Figure PCTCN2019111758-appb-000072
表示虹膜摄像头1有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000073
表示虹膜摄像头2有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000074
可选的，所述设备包括虹膜摄像头1、虹膜摄像头3时，步骤3)根据检测距离选择相应虹膜摄像头采集图像的方法，包括下列具体步骤：
1)当检测距离为
Figure PCTCN2019111758-appb-000075
时，选择虹膜摄像头3；
2)当检测距离为
Figure PCTCN2019111758-appb-000076
时，选择虹膜摄像头1；
3)当检测距离为
Figure PCTCN2019111758-appb-000077
时，保持之前选择的虹膜摄像头；
其中
Figure PCTCN2019111758-appb-000078
表示虹膜摄像头1有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000079
表示虹膜摄像头3有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000080
可选的，所述设备包括虹膜摄像头1、虹膜摄像头2以及虹膜摄像头3时，步骤3)根据检测距离选择相应虹膜摄像头采集图像的方法，包括下列具体步骤：
1)当检测距离为
Figure PCTCN2019111758-appb-000081
时，选择虹膜摄像头3；
2)当检测距离为
Figure PCTCN2019111758-appb-000082
时，选择虹膜摄像头1；
3)当检测距离为
Figure PCTCN2019111758-appb-000083
时，选择虹膜摄像头2；
4)当检测距离为
Figure PCTCN2019111758-appb-000084
时，保持之前选择的虹膜摄像头；
5)当检测距离为
Figure PCTCN2019111758-appb-000085
时，保持之前选择的虹膜摄像头；
其中
Figure PCTCN2019111758-appb-000086
表示虹膜摄像头1有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000087
表示虹膜摄像头2有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000088
表示虹膜摄像头3有效采集距离范围的起始值、终止值，
Figure PCTCN2019111758-appb-000089
本公开的有益效果：本公开的扩展采集距离的虹膜采集设备，包括至少两颗虹膜摄像头，虹膜摄像头配置不同像方焦距的镜头、或者不同分辨率的图像传感器，用来达到不同的物方焦距；控制模块根据距离检测结果对每颗虹膜摄像头采集的图像进行选择，或者控制虹膜摄像头切换，在一定的物距范围内启动相应的虹膜摄像头采集图像，以达到在多个物距范围内均能获取到对焦清晰的虹膜图像的技术效果，可将有效采集距离扩展到25cm-85cm范围内。而且本公开的摄像头切换控制方法，能够防止虹膜摄像头在有效采集范围的临界处发生频繁切换的现象。
本公开实施例还提供了一种摄像头云台自动俯仰调节的方法，适用于调节采集高清晰度面部图像的装置中的云台，以实现在垂直方向上旋转。该摄像头云台自动俯仰调节的方法包括：
S1)根据人脸摄像头采集的人脸图像和预设参考点位置，计算人脸摄像头的第一调节估计角度，第一调节估计角度为摄像头为采集人脸图像应转动角度的粗略估计；
S2)由第一调节估计角度和控制系统计算云台的第二调节估计角度，第二调节估计角度为摄像头为采集人脸图像应转动角度的准确估计；
S3)对云台旋转角度进行平滑处理，计算云台实际旋转角度值；
S4)以当前云台角度为基准，按照实际旋转角度调节云台俯仰角度，之后返回到步骤S1)。
可选的，步骤S1)中计算摄像头的第一调节估计角度的方法为：
1)计算用户双眼中心位置与人脸摄像头光轴之间的垂直高度H2，公式为：
Figure PCTCN2019111758-appb-000090
其中，H1为人脸图像中双眼中心与预设参考点之间垂直距离的像素值，D1为人脸图像中双眼瞳距的像素值，D2为用户双眼瞳距；
2)由H2计算出人脸摄像头从当前位置旋转到能够对准用户双眼中心的位置所需要转动的角度，公式为：
α＝tan -1(H2/L)
L为用户双眼与人脸摄像头之间的水平距离，由人脸图像中的双眼瞳距D1和距离查找表得出。
可选的，步骤S2)中计算第二调节估计角度的方法为：
θ＝kα
其中k＞0，代表控制系数，初始值设置为1，以一定的步长自适应增减调节。
可选的，所述控制系数自适应增减调节方法为，当本次人脸摄像头需要旋转的角度的绝对值大于上一次云台实际旋转角度的绝对值时，控制系数k增大；当本次人脸摄像头需要旋转的角度的绝对值小于上一次云台实际旋转角度的绝对值时，控制系数k减小；否则控制系数不变。
可选的，步骤S3)中的云台旋转角度平滑处理方法为，采用PID控制算法根据当前云台旋转角度与上一次云台实际旋转角度计算云台实际旋转角度值，对跃变的旋转角度数据进行平滑控制。
可选的，根据所需的最近采集距离、最远采集距离以及覆盖的高度范围，计算云台俯仰调节的限值。
本公开实施例还提供了一种摄像头云台自动俯仰调节的装置，包括第一获取单元，用于采集虹膜图像并传输到主板，固定在云台中部，为红外摄像头；
第二获取单元，用于采集人脸图像并传输到主板，与第一获取单元紧密排列，间距不超过5cm，为可见光摄像头；
照明单元，用于在第一获取单元采集图像时开启照明，为两组红外灯板，以第一获取单元为中心对称设置在云台左、右两端；
角度调节单元，用于接收主板发出的角度调节指令，控制云台进行俯仰旋转，固定在云台上，与照明单元之间的距离不小于0.1cm。
可选的，第一获取单元和第二获取单元的光轴平行，第二获取单元的视场角涵盖第一获取单元的视场角。
本公开的有益效果：本公开实现了扩展设备可感知范围的技术效果，与传统的固定角度的虹膜识别设备相比，有效的扩展了设备可采集虹膜图像的范围，通过自适应调节云台俯仰角度，为不同身高用户带来极大的便利。
结合这里披露的本公开的说明和实践，本公开的其他实施例对于本领域技术人员都是易于想到和理解的。说明和实施例仅被认为是示例性的，本公开的真正范围和主旨均由权利要求所限定。

Claims (20)

  1. 一种采集高清晰度面部图像的装置，包括第一智能相机，照明单元，距离传感器，云台，所述第一智能相机包括人脸光学单元，第一虹膜光学单元，算法模块，和控制模块，其特征在于，
    所述距离传感器用于测量用户到第一智能相机的物距；
    所述第一虹膜光学单元根据物距进行自动对焦；
    所述第一智能相机、照明单元固定于所述云台上；
    所述云台包括至少两个电机，支持水平方向旋转和垂直方向旋转，用于调节第一智能相机以及照明单元的角度；
    所述人脸光学单元，第一虹膜光学单元分别与第一智能相机通过MIPI接口连接；
    所述算法模块对采集的面部图像进行人脸检测，根据人脸检测结果调节云台旋转角度，使人脸位于图像中心；
    所述算法模块对采集的面部图像进行图像质量判断，将符合质量要求的面部图像传输到后端进行处理；
    所述第一智能相机设置控制模块，对装置包含的其他外设装置进行逻辑控制，所述外设装置为可选装置。
  2. 根据权利要求1所述的装置，其特征在于，所述人脸光学单元包括不低于2M分辨率的传感器，FOV不小于65°的光学镜头，所述人脸光学单元在采集范围内的MTF值不低于50％@0.5lp/mm。
  3. 根据权利要求1所述的装置，其特征在于，所述第一虹膜光学单元包括不低于8M分辨率的传感器和镜头，所述第一虹膜光学单元在采集范围内的MTF值不低于50％@3lp/mm，像素的空间分辨率大于16pixel/mm。
  4. 根据权利要求3所述的装置，其特征在于，所述第一虹膜光学单元可采用定焦结构或自动对焦马达。
  5. 根据权利要求1所述的装置，其特征在于，所述照明单元包括一个或多个发光元器件，所述发光元器件混合多种光谱，根据距离信息进行光强和角度的调整。
  6. 根据权利要求5所述的装置，其特征在于，所述发光元器件设置为LED或VCSEL激光照明模块。
  7. 根据权利要求1所述的装置，其特征在于，所述外设装置包括第二智能相机，所述第二智能相机包括第二虹膜光学单元，由所述第一智能相机的控制模块控制，与第一智能相机协调工作，根据物距进行自动对焦。
  8. 根据权利要求7所述的装置，其特征在于，所述第二智能相机固定于所述云台上，与第二虹膜光学单元通过MIPI接口连接。
  9. 根据权利要求7所述的装置，其特征在于，所述第二虹膜光学单元包括不低于8M分辨率的传感器和镜头，所述第二虹膜光学单元在采集范围内的MTF值不低于50％@3lp/mm，像素的空间分辨率大于16pixel/mm。
  10. 根据权利要求7所述的装置，其特征在于，所述第二虹膜光学单元采用定焦结构或自动对焦马达。
  11. 根据权利要求7所述的装置，其特征在于，所述装置还包括摄像头阵列，所述摄像头阵列包括虹膜摄像头1和虹膜摄像头2，所述第一虹膜光学单元为虹膜摄像头1，所述第二虹膜光学单元为虹膜摄像头2；虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
    Figure PCTCN2019111758-appb-100001
    虹膜摄像头2为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
    Figure PCTCN2019111758-appb-100002
    满足f′ 1＜f′ 2，m 1＝m 2，f 1＜f 2，
    Figure PCTCN2019111758-appb-100003
  12. 根据权利要求1所述的装置，其特征在于，所述装置还包括摄像头阵列，所述摄像头阵列包括虹膜摄像头1和虹膜摄像头3，所述第一虹膜光学单元为虹膜摄像头1，所述其他外设装置为虹膜摄像头3；虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
    Figure PCTCN2019111758-appb-100004
    虹膜摄像头3为从摄像头，其镜头像方焦距为f′ 3、图像传感器分辨率为m 3、物方焦距为f 3、有效采集距离为
    Figure PCTCN2019111758-appb-100005
    f′ 1＝f′ 3，m 1＞m 3，f 1＞f 3，
    Figure PCTCN2019111758-appb-100006
  13. 根据权利要求7所述的装置，其特征在于，所述装置还包括摄像头阵列，所述摄像头阵列包括虹膜摄像头1、虹膜摄像头2及虹膜摄像头3，所述第一虹膜光学单元为虹膜摄像头1，所述第二虹膜光学单元为虹膜摄像头2，所述其他外设装置为虹膜摄像头3；虹膜摄像头1为主摄像头，其镜头像方焦距为f′ 1、图像传感器分辨率为m 1、物方焦距为f 1、有效采集距离为
    Figure PCTCN2019111758-appb-100007
    虹膜摄像头2为从摄像头，其镜头像方焦距f′ 2、图像传感器为分辨率m 2、物方焦距为f 2、有效采集距离为
    Figure PCTCN2019111758-appb-100008
    虹膜摄像头3为从摄像头，其镜头像方焦距为f′ 3、图像传感器分辨率为m 3、物方焦距为f 3、有效采集距离为
    Figure PCTCN2019111758-appb-100009
    f′ 1＝f′ 3＜f′ 2，m 1＝m 2＞m 3，f 3＜f 1＜f 2，
    Figure PCTCN2019111758-appb-100010
  14. 根据权利要求11所述的装置，其特征在于，控制模块用于在所有虹膜摄像头被同时初始化并开始采集虹膜图像之后，通过人眼检测算法以及相应的距离查找表进行距离检测，并用于根据检测的距离选择相应虹膜摄像头采集的图像，进行特征提取、虹膜注册、虹膜识别处理；其中，根据检测的距离选择相应虹膜摄像头的具体步骤为：
    1)当检测距离为
    Figure PCTCN2019111758-appb-100011
    时，选择虹膜摄像头1；
    2)当检测距离为
    Figure PCTCN2019111758-appb-100012
    时，选择虹膜摄像头2；
    3)当检测距离为
    Figure PCTCN2019111758-appb-100013
    时，保持之前选择的虹膜摄像头；
    其中
    Figure PCTCN2019111758-appb-100014
    表示虹膜摄像头1有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100015
    表示虹膜摄像头2有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100016
  15. 根据权利要求12所述的装置，其特征在于，控制模块用于在摄像头阵列中的深度摄像头被初始化并开始采集红外散点图，得到距离检测结果之后，根据检测的距离选择相应的虹膜摄像头开始采集图像，并进行特征提取、虹膜注册、虹膜识别处理；
    其中，根据检测的距离选择相应虹膜摄像头的具体步骤为：
    1)当检测距离为
    Figure PCTCN2019111758-appb-100017
    时，选择虹膜摄像头3；
    2)当检测距离为
    Figure PCTCN2019111758-appb-100018
    时，选择虹膜摄像头1；
    3)当检测距离为
    Figure PCTCN2019111758-appb-100019
    时，保持之前选择的虹膜摄像头；
    其中
    Figure PCTCN2019111758-appb-100020
    表示虹膜摄像头1有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100021
    表示虹膜摄像头3有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100022
  16. 根据权利要求13所述的装置，其特征在于，控制模块用于在摄像头阵列中的人脸摄像头开始采集人脸图像之后，通过人眼检测算法以及相应的距离查找表进行距离检测，并根据检测的距离选择相应的虹膜摄像头开始采集图像，进行特征提取、虹膜注册、虹膜识别处理；
    其中，根据检测的距离选择相应虹膜摄像头的具体步骤为：
    1)当检测距离为
    Figure PCTCN2019111758-appb-100023
    时，选择虹膜摄像头3；
    2)当检测距离为
    Figure PCTCN2019111758-appb-100024
    时，选择虹膜摄像头1；
    3)当检测距离为
    Figure PCTCN2019111758-appb-100025
    时，选择虹膜摄像头2；
    4)当检测距离为
    Figure PCTCN2019111758-appb-100026
    时，保持之前选择的虹膜摄像头；
    5)当检测距离为
    Figure PCTCN2019111758-appb-100027
    时，保持之前选择的虹膜摄像头；
    其中
    Figure PCTCN2019111758-appb-100028
    表示虹膜摄像头1有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100029
    表示虹膜摄像头2有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100030
    表示虹膜摄像头3有效采集距离范围的起始值、终止值，
    Figure PCTCN2019111758-appb-100031
  17. 一种摄像头云台自动俯仰调节的方法，其特征在于，适用于如权利要求1所述的装置，所述方法包括：
    S1)根据人脸摄像头采集的人脸图像和预设参考点位置，计算人脸摄像头的第一调节估计角度，第一调节估计角度为摄像头为采集人脸图像应转动角度的粗略估计；
    S2)由第一调节估计角度和控制系统计算云台的第二调节估计角度，第二调节估计角度为摄像头为采集人脸图像应转动角度的准确估计；
    S3)对云台的第二调节估计角度进行平滑处理，计算云台实际旋转角度值；
    S4)以当前云台角度为基准，按照实际旋转角度调节云台俯仰角度，之后返回到步骤S1)。
  18. 根据权利要求17所述的方法，其特征在于，步骤S1)中计算摄像头的第一调节估计角度的方法为：
    1)计算用户双眼中心位置与人脸摄像头光轴之间的垂直高度H2，公式为：
    Figure PCTCN2019111758-appb-100032
    其中，H1为人脸图像中双眼中心与预设参考点之间垂直距离的像素值，D1为人脸图像中双眼瞳距的像素值，D2为用户双眼瞳距；
    2)由H2计算出人脸摄像头从当前位置旋转到能够对准用户双眼中心的位置所需要转动的角度，公式为：
    α＝tan -1(H2/D)
    D为用户双眼与人脸摄像头之间的水平距离，由人脸图像中的双眼瞳距D1和距离查找表得出。
  19. 根据权利要求17所述的方法，其特征在于，步骤S1)中计算第二调节估计角度的方法为：
    θ＝kα
    其中k＞0，代表控制系数，初始值设置为1，以一定的步长自适应增减调节，当本次人脸摄像头需要旋转角度的绝对值大于上一次云台实际旋转角度的绝对值时，控制系数k增大；当本次人脸摄像头需要旋转的角度的绝对值小于上一次云台实际旋转角度的绝对值时，控制系数k减小；否则控制系数不变。
  20. 根据权利要求17所述的方法，其特征在于，步骤S1)中的第二调节估计角度平滑处理方法为，采用PID控制算法根据当前云台需要旋转的角度与上一次云台实际旋转角度计算云台实际旋转角度值，对跃变的旋转角度数据进行平滑控制。
PCT/CN2019/111758 2018-10-18 2019-10-17 采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法 WO2020078440A1 (zh)

Applications Claiming Priority (6)

Application Number Priority Date Filing Date Title
CN201811212252.5 2018-10-18
CN201811212252.5A CN109451233B (zh) 2018-10-18 2018-10-18 一种采集高清晰度面部图像的装置
CN201811472517.5A CN109614909B (zh) 2018-12-04 2018-12-04 一种扩展采集距离的虹膜采集设备与方法
CN201811472517.5 2018-12-04
CN201910201929.3A CN109977828B (zh) 2019-03-18 2019-03-18 一种摄像头云台自动俯仰调节的方法与装置
CN201910201929.3 2019-03-18

Publications (1)

Publication Number Publication Date
WO2020078440A1 true WO2020078440A1 (zh) 2020-04-23

Family

ID=70283362

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/111758 WO2020078440A1 (zh) 2018-10-18 2019-10-17 采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法

Country Status (1)

Country Link
WO (1) WO2020078440A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112036257A (zh) * 2020-08-07 2020-12-04 华中师范大学 一种无感知人脸图像采集方法和系统
CN112417997A (zh) * 2020-11-03 2021-02-26 北京中科虹霸科技有限公司 虹膜图像采集装置及方法
CN114612998A (zh) * 2022-02-22 2022-06-10 北京无线电计量测试研究所 一种远距离虹膜采集系统及采集方法
CN114743255A (zh) * 2020-12-23 2022-07-12 深圳爱酷智能科技有限公司 多模态生物特征识别方法、装置、可读存储介质及设备
CN115356261A (zh) * 2022-07-29 2022-11-18 燕山大学 一种汽车球笼防尘罩的缺陷检测系统及方法
US11539873B2 (en) * 2019-01-22 2022-12-27 Lg Innotek Co., Ltd. Camera apparatus and autofocusing method therefor
WO2023028242A1 (en) * 2021-08-25 2023-03-02 Tools for Humanity Corporation Controlling a two-dimensional mirror gimbal for purposes of iris scanning
CN115938018A (zh) * 2022-09-28 2023-04-07 杭州指安科技股份有限公司 一种生物识别集成设备及方法
CN116170564A (zh) * 2022-12-26 2023-05-26 中建三局第一建设工程有限责任公司 顶模施工核心筒楼板钢筋图像采集系统及方法

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101520838A (zh) * 2008-02-27 2009-09-02 中国科学院自动化研究所 自动跟踪和自动变焦的虹膜图像获取方法
US20150245767A1 (en) * 2014-02-28 2015-09-03 Lrs Identity, Inc. Dual iris and color camera in a mobile computing device
CN204667421U (zh) * 2015-05-04 2015-09-23 黄榕 自适应定位虹膜图像获取系统
CN108446648A (zh) * 2018-03-26 2018-08-24 北京上古视觉科技有限公司 一种虹膜采集系统及虹膜识别系统
CN109451233A (zh) * 2018-10-18 2019-03-08 北京中科虹霸科技有限公司 一种采集高清晰度面部图像的装置
CN109614909A (zh) * 2018-12-04 2019-04-12 北京中科虹霸科技有限公司 一种扩展采集距离的虹膜采集设备与方法
CN109977828A (zh) * 2019-03-18 2019-07-05 北京中科虹霸科技有限公司 一种摄像头云台自动俯仰调节的方法与装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101520838A (zh) * 2008-02-27 2009-09-02 中国科学院自动化研究所 自动跟踪和自动变焦的虹膜图像获取方法
US20150245767A1 (en) * 2014-02-28 2015-09-03 Lrs Identity, Inc. Dual iris and color camera in a mobile computing device
CN204667421U (zh) * 2015-05-04 2015-09-23 黄榕 自适应定位虹膜图像获取系统
CN108446648A (zh) * 2018-03-26 2018-08-24 北京上古视觉科技有限公司 一种虹膜采集系统及虹膜识别系统
CN109451233A (zh) * 2018-10-18 2019-03-08 北京中科虹霸科技有限公司 一种采集高清晰度面部图像的装置
CN109614909A (zh) * 2018-12-04 2019-04-12 北京中科虹霸科技有限公司 一种扩展采集距离的虹膜采集设备与方法
CN109977828A (zh) * 2019-03-18 2019-07-05 北京中科虹霸科技有限公司 一种摄像头云台自动俯仰调节的方法与装置

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US11539873B2 (en) * 2019-01-22 2022-12-27 Lg Innotek Co., Ltd. Camera apparatus and autofocusing method therefor
CN112036257A (zh) * 2020-08-07 2020-12-04 华中师范大学 一种无感知人脸图像采集方法和系统
CN112417997A (zh) * 2020-11-03 2021-02-26 北京中科虹霸科技有限公司 虹膜图像采集装置及方法
CN114743255B (zh) * 2020-12-23 2024-03-26 深圳爱酷智能科技有限公司 多模态生物特征识别方法、装置、可读存储介质及设备
CN114743255A (zh) * 2020-12-23 2022-07-12 深圳爱酷智能科技有限公司 多模态生物特征识别方法、装置、可读存储介质及设备
WO2023028242A1 (en) * 2021-08-25 2023-03-02 Tools for Humanity Corporation Controlling a two-dimensional mirror gimbal for purposes of iris scanning
US20230062777A1 (en) * 2021-08-25 2023-03-02 Tools for Humanity Corporation Controlling a two-dimensional mirror gimbal for purposes of iris scanning
US11895404B2 (en) 2021-08-25 2024-02-06 Worldcoin Foundation Controlling a two-dimensional mirror gimbal for purposes of iris scanning
CN114612998A (zh) * 2022-02-22 2022-06-10 北京无线电计量测试研究所 一种远距离虹膜采集系统及采集方法
CN114612998B (zh) * 2022-02-22 2024-04-12 北京无线电计量测试研究所 一种远距离虹膜采集系统及采集方法
CN115356261A (zh) * 2022-07-29 2022-11-18 燕山大学 一种汽车球笼防尘罩的缺陷检测系统及方法
CN115938018A (zh) * 2022-09-28 2023-04-07 杭州指安科技股份有限公司 一种生物识别集成设备及方法
CN116170564A (zh) * 2022-12-26 2023-05-26 中建三局第一建设工程有限责任公司 顶模施工核心筒楼板钢筋图像采集系统及方法
CN116170564B (zh) * 2022-12-26 2023-10-27 中建三局第一建设工程有限责任公司 顶模施工核心筒楼板钢筋图像采集方法

Similar Documents

Publication Publication Date Title
WO2020078440A1 (zh) 采集高清晰度面部图像的装置和摄像头云台自动俯仰调节的方法
CN105554385B (zh) 一种远距离多模态生物特征识别方法及其系统
US20170061210A1 (en) Infrared lamp control for use with iris recognition authentication
CN109451233B (zh) 一种采集高清晰度面部图像的装置
CN101051349B (zh) 采用主动视觉反馈的双目虹膜采集设备
WO2015176657A1 (zh) 虹膜识别装置及其制造方法和应用
CN102855471B (zh) 远距离虹膜智能成像装置及方法
US9300858B2 (en) Control device and storage medium for controlling capture of images
KR100977499B1 (ko) 거울의 패닝과 틸팅을 이용한 원거리 홍채 영상 획득시스템
US20050084179A1 (en) Method and apparatus for performing iris recognition from an image
US20120275648A1 (en) Imaging device and imaging method and program
EP1387314A1 (en) Method and apparatus for picking up image of object being authenticated
WO2019218303A1 (zh) 智能售货柜、物品识别方法、装置、服务器和存储介质
KR100869998B1 (ko) 원거리 홍채 영상 획득 시스템
WO2013056001A1 (en) System and method for the long range acquisition of iris images for stationary and mobile subjects
CN105279490A (zh) 一种人机交互式虹膜图像自动采集装置
WO2021204211A1 (zh) 人脸和虹膜图像采集方法、装置、可读存储介质及设备
CN108985291B (zh) 一种基于单摄像头的双眼追踪系统
JP5625443B2 (ja) 撮像システムおよび撮像装置
CN109614909B (zh) 一种扩展采集距离的虹膜采集设备与方法
CN106934349B (zh) 双摄像头成像及虹膜采集识别一体化设备
CN202196428U (zh) 单图像传感器自适应双眼虹膜同步采集系统
US11882354B2 (en) System for acquisiting iris image for enlarging iris acquisition range
JP2024003037A (ja) 電子機器、電子機器の制御方法、プログラムおよび記憶媒体
CN216352422U (zh) 多模态图像采集装置

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: 19872708

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: 19872708

Country of ref document: EP

Kind code of ref document: A1