Apparatus and method for iris image acquisition

Abstract

The application provides a device for face iris image acquisition, comprising a face optical imaging unit, a human eye detection unit, an optical mapping unit, a local area configuration unit, an iris optical imaging unit, an iris image AF automatic focusing unit and an image acquisition processing unit; the optical mapping unit is used for realizing the transformation of the face optical imaging unit in response to the corresponding object image optical geometric mapping relationship of the iris optical imaging unit according to the human eye center pixel position of the human eye detection unit; the local area configuration unit is used for realizing the synchronous dynamic configuration of the local area of the iris image with the fixed pixel resolution of the iris optical imaging unit according to the corresponding optical geometric mapping relationship of the optical mapping unit, the application can meet the frame rate requirement of real-time image processing under the high-pixel resolution image data acquisition and transmission, realize high-speed accurate transmission, and ensure the high-speed accurate acquisition of high-quality automatic focusing iris images at the same time.

Description

Technical Field

[0001] This invention relates to the field of optical image acquisition technology, and in particular to an apparatus and method for acquiring facial iris images. Background Technology

[0002] Faced with the increasing demand for long-distance, wide-angle iris image acquisition, the current mainstream solutions are either 2-axis rotating iris optical imaging systems or high-pixel-resolution iris optical imaging systems.

[0003] Patent application number 202011131179 discloses a device and method for acquiring iris images of multiple target groups. The device includes a base, a vertical motor module, a horizontal motor module, a reflector, a stage module, motion limiting components, and a control system. The base supports and fixes an overall two-dimensional turntable. The vertical motor module includes a vertical motor and a vertical motion adapter plate. The horizontal motor module includes a horizontal motor, a support plate, a horizontal motion adapter plate, and rolling bearings. The reflector is fixed to the stage module, which includes a carrying plate, a carrying support, and fixing components, and is fixed to the horizontal motor module. Multiple motion limiting components are installed in different positions. The control system includes a driver, a PLC controller, and a PC control card to control the turntable movement. The iris camera is vertically mounted. This device can effectively increase the iris recognition field of view, automatically adapt to targets of different heights at different directions and distances, and is fast and robust.

[0004] Application No. 202110588620 discloses an image information processing method and system for a specific object. The system includes a first module for acquiring image or video data; a second module for determining face range data from the image or video data acquired by the first module; and a third module for extracting feature data from the face range data, comprising the following units: a first unit for determining the detection point positions of facial features from the face range data and extracting feature data from the detection point positions; a second unit for determining the detection area positions of facial skin features from the face range data and extracting feature data from the detection area positions; a third unit for determining the detection area positions of facial iris features from the face range data and extracting feature data from the detection point positions; and a fourth module for processing and analyzing the facial feature data, the facial skin feature data, and the facial iris feature data, comprising the following unit: a fourth unit for comparing and analyzing the facial feature data with pre-stored facial feature data and obtaining results. The fifth unit is used to compare and analyze the facial skin feature data with pre-stored facial skin feature data and obtain the results; the sixth unit is used to compare and analyze the facial iris feature data with pre-stored facial iris feature data and obtain the results; the fifth module is used to integrate the comparison and analysis results of the facial facial feature data, the facial skin feature data, and the facial iris feature data to obtain the face recognition determination result, including the following units: the seventh unit is used to perform normalization operations on the comparison and analysis results of the facial facial feature data, the facial skin feature data, and the facial iris feature data respectively; the eighth unit is used to comprehensively process the normalized results of the comparison and analysis results of the facial facial feature data, the facial skin feature data, and the facial iris feature data to obtain the face recognition determination result; the sixth module is used to make response processing based on the comprehensive determination result of the face recognition.

[0005] Similar to the two patent applications mentioned above, existing technologies, given the future trend towards higher pixel resolutions, such as 64M or higher, require separate acquisition of face and iris images when capturing video and performing face and iris scanning. When the face changes, both images must be acquired again, resulting in a massive amount of image data. This leads to challenges in image data acquisition and transmission bandwidth. At 64M pixel resolution, the frame rate is less than 10fps, clearly insufficient for real-time image processing. Such a low frame rate severely impacts image acquisition speed and subsequent recognition performance. Furthermore, high pixel resolution images pose a significant challenge to image processing speed itself, as the massive amount of image data severely affects image computation and processing speed.

[0006] In addition to high-resolution images, the small pixel scale of images with the same image plane makes it difficult to achieve high-speed and accurate optical imaging of iris images. The known traditional autofocus uses the maximum value search of the focus image, and iterates through 2-3 search cycles using the coarse to fine strategy, which takes at least 1 second. A more serious problem is that the depth of field of the standard iris diameter image is extremely small, which makes it impossible to guarantee the focus accuracy. Therefore, how to acquire high-quality iris images at high speed and accuracy is also a huge challenge.

[0007] To address the problems existing in current technologies—namely, the slow acquisition speed and poor subsequent recognition performance of high-resolution iris images, and the focusing accuracy issues during high-resolution iris image acquisition—there is an urgent need for a device and method for acquiring facial iris images to solve these problems. Summary of the Invention

[0008] The purpose of this invention is to solve the problems of slow acquisition speed and poor subsequent recognition performance of high-pixel resolution iris images in the prior art, and to provide a device and method for acquiring facial iris images.

[0009] In order to achieve the above-mentioned objectives of the present invention, the following technical solution is adopted:

[0010] A device for acquiring facial iris images includes:

[0011] Face optical imaging unit, iris optical imaging unit, optical imaging association unit, image acquisition and processing unit;

[0012] The image acquisition and processing unit is connected to the face optical imaging unit, the iris optical imaging unit, and the optical imaging association unit, respectively, and controls the face optical imaging unit, the iris optical imaging unit, and the optical imaging association unit.

[0013] The face optical imaging unit outputs a global area face image;

[0014] The optical imaging association unit is used to establish a synchronous association relationship between the pre-defined face optical imaging unit and iris optical imaging unit for outputting global-local region images;

[0015] The iris optical imaging unit outputs a corresponding local area iris image according to the synchronization correlation;

[0016] The imaging field of view of the face optical imaging unit is greater than or equal to that of the iris optical imaging unit.

[0017] Furthermore,

[0018] The optical imaging association unit includes: a human eye detection unit, an optical mapping unit, and a local region configuration unit;

[0019] The face optical imaging unit is used to acquire global area face images;

[0020] The human eye detection unit is used to detect the position of the center pixels of the two eyes of the face from the global area face image acquired by the face optical imaging unit;

[0021] The optical mapping unit is used to transform the optical geometric mapping relationship between the face optical imaging unit and the iris optical imaging unit based on the center pixel position of the two eyes detected by the human eye detection unit.

[0022] The local region configuration unit is used to synchronously and dynamically configure the local region of the iris image with a fixed pixel resolution corresponding to the iris optical imaging unit according to the object-image optical geometric mapping relationship obtained by the optical mapping unit.

[0023] The iris optical imaging unit is used to acquire a local iris image based on a local iris image with a fixed pixel resolution synchronized by the local region configuration unit.

[0024] Furthermore, when the optical mapping unit transforms the object-image optical geometric mapping relationship of the face optical imaging unit in response to the iris optical imaging unit according to the center pixel position of the two eyes of the face in the human eye detection unit, the calculation method of the transformation factor OMTF is as follows:

[0025] OMTF = Fi * PSe / (Fe * PSe)

[0026] = (PXi - cPXi / 2) / (PXe - cPXe / 2)

[0027] = (PYi - cPYi / 2) / (PYe - cPYe / 2)

[0028] Wherein, OMTF is the transformation factor with inherent object-image optical geometric mapping relationship; Fi is the imaging optical focal length of the iris optical imaging unit; PSe is the unit pixel resolution of the face optical imaging unit; Fe is the imaging optical focal length of the face optical imaging unit; PSi is the unit pixel resolution of the iris optical imaging unit; (PXi, PYi) are the XY coordinates of the center pixel position of the eyes on the image side of the iris optical imaging unit; (cPXi, cPYi) are the XY pixel resolution of the iris optical imaging unit; (PXe, PYe) are the XY coordinates of the center pixel position of the eyes on the image side of the face optical imaging unit; and (cPXe, cPYe) are the XY pixel resolution of the face optical imaging unit.

[0029] Furthermore, by defining a local region of the iris image, the frequency of dynamic synchronization update of the execution association state between the optical mapping unit and the local region configuration unit is controlled.

[0030] Furthermore, when the iris optical imaging unit acquires a local iris image based on the iris image local region synchronized by the local region configuration unit, the local iris image is the image acquired within the region enclosed by the pixel position (PXi, PYi) coordinates and the pixel range (PXroi, PYroi).

[0031] Where: PXroi and PYroi are the XY pixel ranges of a local region of the iris image;

[0032] cPXi / 16<=PXroi<=cPXi / 2, cPYi / 16<=PYroi<=cPYi / 2

[0033] ROI=RECT(left, top, right, bottom)

[0034] =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXi+PXroi / 2, PYi+PYroi / 2);

[0035] or

[0036] ROI=RECT(left,top,width,height)

[0037] =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXroi, PYroi).

[0038] Furthermore, when acquiring local iris images based on pixel positions (PXi, PYi) and pixel ranges (PXroi, PYroi), the physical simulation local array image output (X_ADD_START, Y_ADD_START, X_ADD_END, Y_ADD_END) or the corresponding digital offset local array image output (X_CROP_OFFSET, Y_CROP_OFFSET, X_CROP_WIDTH, Y_CROP_HEIGHT) corresponding to the imaging pixel array path of the iris optical imaging unit is used.

[0039] in:

[0040] X_ADD_START=left=PXi-PXroi / 2,

[0041] Y_ADD_START=top=PYi-PYroi / 2,

[0042] X_ADD_END=right=PXi+PXroi / 2,

[0043] Y_ADD_END=bottom=PYi+PYroi / 2;

[0044] or

[0045] X_CROP_OFFSET=left=PXi-PXroi / 2,

[0046] Y_CROP_OFFSET=top=PYi-PYroi / 2,

[0047] X_CROP_WIDTH=width=PXroi,

[0048] Y_CROP_HEIGHT=height=PYroi.

[0049] Furthermore, it also includes an iris image AF autofocus unit, wherein the image acquisition and processing unit is connected to the iris image AF autofocus unit and controls the iris image AF autofocus unit;

[0050] The iris image AF autofocus unit is used to automatically focus the local area iris image based on the local area iris image corresponding to the iris optical imaging unit, so as to obtain the focused local area iris image.

[0051] Furthermore, the iris image autofocus AF unit employs the imaging lens drive parameter control of the iris optical imaging unit.

[0052] Furthermore, the AF autofocus parameter control depends on the imaging lens drive type and is achieved by adjusting the imaging optical image distance or adjusting the imaging optical focal length / diopter.

[0053] Furthermore, the control of the imaging lens driving parameters of the iris optical imaging unit includes:

[0054] The step count is 2*K+1 and the step size is the STEP parameter, where K is the step count;

[0055] Achieve continuous autofocus in one direction within the predetermined focusing range FR = (2 * K + 1) * STEP;

[0056] in:

[0057] STEP = 2 * FNO * SOC

[0058] FNO is the aperture parameter of the iris optical imaging unit;

[0059] SOC is the minimum physical spot resolution parameter of the iris optical imaging unit.

[0060] Furthermore, the method for controlling the imaging lens driving parameters of the iris optical imaging unit includes:

[0061] Step S1: Calculate PD, where PD is the phase difference corresponding to a local region of the iris image;

[0062] Step S2: Calculate the relative distance FP of the corresponding focusing position based on the phase difference obtained in step S1. The calculation formula is as follows:

[0063] FP = PD * CC;

[0064] Where CC is the conversion coefficient between the phase difference and the corresponding relative distance FP (i.e., the defocus distance, the relative distance from the focus position), which is the defocus rate. The defocus rate is achieved by fitting the actual corresponding linear or nonlinear relationship.

[0065] Step S3: Based on the relative distance FP of the focal position obtained in step S2, drive the PDAF type driver to execute the imaging lens driving parameters of the iris optical imaging unit, and drive the corresponding relative distance FP of the focal position, that is, directly drive STEP*K step positions.

[0066] Step S4 iterates through steps S1-S3 until PD <= EP, where EP = STEP / CC, and EP is the predetermined phase difference focusing error. In other words, it terminates at a single step size, where K <= 1.

[0067] A method for acquiring facial iris images includes: a facial optical imaging unit, an eye detection unit, an optical mapping unit, a local region configuration unit, an iris optical imaging unit, and an image acquisition and processing unit; the image acquisition and processing unit is connected to the facial optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, and the iris optical imaging unit, respectively, and controls the facial optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, and the iris optical imaging unit;

[0068] The specific steps are as follows:

[0069] Step 1: The image acquisition and processing unit controls the face optical imaging unit to acquire face images of the entire region;

[0070] Step 2: The image acquisition and processing unit controls the human eye detection unit to detect the center pixel positions of the eyes of the face from the global face image acquired in Step 1.

[0071] Step 3: Based on the center pixel positions of the eyes of the face detected in Step 2, the image acquisition and processing unit controls the optical mapping unit to transform the optical geometric mapping relationship between the face optical imaging unit and the iris optical imaging unit.

[0072] Step four: Based on the object-image optical geometric mapping relationship obtained in step three, the image acquisition and processing unit synchronously and dynamically configures the local region of the iris image with a fixed pixel resolution corresponding to the iris optical imaging unit.

[0073] Step 5: The image acquisition and processing unit controls the iris optical imaging unit to acquire the iris image of the local area based on the local iris image obtained by synchronous dynamic configuration in step 4.

[0074] Furthermore, it also includes an iris image AF autofocus unit, wherein the image acquisition and processing unit is connected to the iris image AF autofocus unit and controls the iris image AF autofocus unit;

[0075] A sixth step is also set after step five;

[0076] Step six: The image acquisition and processing unit controls the automatic focusing of the local iris image based on the local iris image acquired in step five, and obtains the focused local iris image.

[0077] The advancements of this invention compared to existing technologies are as follows:

[0078] In this invention, by setting up an optical mapping unit and a local region configuration unit, the optical mapping unit can transform the object-image optical geometric mapping relationship between the face optical imaging unit and the iris optical imaging unit according to the center pixel position of the eyes of the face; the local region configuration unit then dynamically configures the local region of the iris image with a fixed pixel resolution corresponding to the iris optical imaging unit according to the object-image optical geometric mapping relationship. Under the action of the optical mapping unit and the local region configuration unit, the global face image acquired by the face optical imaging unit and the local iris image acquired by the iris optical imaging unit can be associated and mapped. When the face state corresponding to the global face image changes, it is only necessary to acquire the global face image again. Based on the re-acquired global face image and the association relationship between the previous global face image and the local iris image, a new local iris image is directly generated. Only the local iris image needs to be acquired, which greatly reduces the amount of image data. Therefore, this invention can meet the frame rate requirements of real-time image processing under high pixel resolution image data acquisition and transmission, achieve high-speed and accurate transmission, realize high-speed and accurate image processing of high pixel resolution images, improve image computing and processing speed performance, and realize high-speed and accurate automatic focusing acquisition of high-quality iris images.

[0079] It should be understood that the foregoing general description and the subsequent detailed description are illustrative and explanatory, and should not be used as limitations on the content claimed in this invention. Attached Figure Description

[0080] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the accompanying drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0081] Figure 1 This is a schematic diagram of the unit components of the device for acquiring facial iris images according to the present invention.

[0082] Figure 2 This is a logic diagram of the image acquisition and processing unit of the device for acquiring facial iris images according to the present invention. Detailed Implementation

[0083] The objects and functions of the present invention, as well as the methods for achieving these objects and functions, will be clarified by referring to exemplary embodiments. However, the present invention is not limited to the exemplary embodiments disclosed below; it can be implemented in various forms. The purpose of this specification is merely to help those skilled in the art to comprehensively understand the specific details of the invention.

[0084] In the following description, embodiments of the invention will be illustrated with reference to the accompanying drawings. In the drawings, the same reference numerals represent the same or similar parts, or the same or similar steps.

[0085] Example 1

[0086] like Figure 1 As shown, an embodiment provides a device for acquiring facial iris images, comprising:

[0087] The system includes a face optical imaging unit, an iris optical imaging unit, a human eye detection unit, an optical mapping unit, a local region configuration unit, an iris image AF autofocus unit, and an image acquisition and processing unit.

[0088] The human eye detection unit, optical mapping unit, and local area configuration unit constitute an optical imaging association unit.

[0089] The face optical imaging unit outputs a global area face image.

[0090] The optical imaging association unit is used to establish a synchronous association relationship between the pre-defined face optical imaging unit and iris optical imaging unit for outputting global-local region images.

[0091] The iris optical imaging unit outputs a corresponding local area iris image according to the synchronization correlation;

[0092] Furthermore,

[0093] The face optical imaging unit is used to acquire face images of the corresponding global region.

[0094] The human eye detection unit is used to detect the center pixel position of the corresponding human eyes based on the global region human face image corresponding to the human face optical imaging unit.

[0095] The optical mapping unit is used to realize the transformation of the optical geometric mapping relationship between the face optical imaging unit and the object image corresponding to the iris optical imaging unit, based on the center pixel position of the two eyes of the face corresponding to the human eye detection unit.

[0096] The local region configuration unit is used to realize the synchronous dynamic configuration of the ROI of the iris image with a fixed pixel resolution corresponding to the iris optical imaging unit according to the optical geometric mapping relationship corresponding to the optical mapping unit.

[0097] The iris optical imaging unit is used to acquire iris images of the corresponding fixed-pixel-resolution iris image ROI based on the fixed-pixel-resolution iris image ROI corresponding to the local region configuration unit.

[0098] The iris image AF autofocus unit is used to perform AF autofocus processing on the corresponding ROI of the iris image based on the ROI of the iris image corresponding to the iris optical imaging unit.

[0099] The image acquisition and processing unit is connected to the face optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, and the iris optical imaging unit, respectively, and controls the face optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, and the iris optical imaging unit to realize synchronous image acquisition, image processing calculation, and control of each unit.

[0100] The field of view (FOV) of the face optical imaging unit is greater than or equal to that of the iris optical imaging unit.

[0101] The human eye detection unit is used to detect the center pixel position of the eyes in the global region of the face image by the face optical imaging unit.

[0102] This invention utilizes the currently known convolutional neural network (CNN) cascade model based on deep learning to reliably and accurately achieve face region detection and output coordinates (PXe, PYe) for the center pixel position detection of the eyes.

[0103] The optical mapping unit is used to realize the object-image optical geometric mapping relationship transformation OMT, (PXe, PYe) : -> (PXi, PYi) based on the center pixel position of the face and eyes of the face / eye detection unit.

[0104] PXi-cPXi / 2

[0105] =Fi / (Zi-Fi)*(Xi) / PSi

[0106] =Fi / (Zi-Fi)*(Xe-Xoffset) / PSi

[0107] =Fi / (Zi-Fi)*[(Ze-Fe) / Fe*(PXe-cPXe / 2)*PSe-Xoffset)] / PSi.

[0108] PYi-cPYi / 2

[0109] =Fi / (Zi-Fi)*(Yi) / PSi

[0110] =Fi / (Zi-Fi)*(Ye-Yoffset) / PSi

[0111] =Fi / (Zi-Fi)*[(Ze-Fe) / Fe*(PYe-cPYe / 2)*PSe-Yoffset)] / PSi.

[0112] Specifically, when the conditions Ze >> Zoffset, Fi, Fe are satisfied,

[0113] (PXi-cPXi / 2)+Fi / (Zi-Fi)*Xoffset / PSi=(PXe-cPXe / 2)*Fi*PSe / (Fe*PSi)

[0114] (PYi-cPYi / 2)+Fi / (Zi-Fi)*Yoffset / PSi=(PYe-cPYe / 2)*Fi*PSe / (Fe*PSi)

[0115] Specifically, when the conditions Ze >> Xoffset, Yoffset, Zoffset, Fi, Fe are satisfied, the important fundamental characteristic of this invention is the inherent object-image optical geometric mapping relationship transformation factor OMTF:

[0116] (PXi-cPXi / 2) / (PXe-cPXe / 2)=

[0117] (PYi-cPYi / 2) / (PYe-cPYe / 2)=

[0118] Fi*PSe / (Fe*PSi)=OMTF

[0119] Where (PXe, PYe) are the XY coordinates of the center pixel position of the eyes on the image side of the face optical imaging unit.

[0120] (cPXe, cPYe) represents the pixel resolution of the XY side of the face optical imaging unit.

[0121] PSe is the unit pixel resolution of the face optical imaging unit, in μm / pixel.

[0122] Fe is the imaging optical focal length of the face optical imaging unit, in mm.

[0123] Ze is the imaging optical distance of the face optical imaging unit, in mm.

[0124] (Xe, Ye) are the XY position coordinates of the object space of the face optical imaging unit, in mm.

[0125] (PXi, PYi) represents the XY coordinates of the center pixel position of the eyes on the image side of the iris optical imaging unit.

[0126] (cPXi, cPYi) represents the pixel resolution of the XY side of the iris optical imaging unit.

[0127] PSi is the resolution per pixel of the iris optical imaging unit, in μm / pixel.

[0128] Fi is the imaging optical focal length of the iris optical imaging unit, in mm.

[0129] Zi represents the imaging optical distance of the iris optical imaging unit, in mm.

[0130] (Xi, Yi) represents the XY coordinates of the object space position of the iris optical imaging unit, in mm.

[0131] (Xoffset, Yoffset, Zoffset)

[0132] Xoffset = Xe-Xi, Yoffset = Ye-Yi, Zoffset = Ze-Zi are the XYZ coordinate offsets of the object space position of the face optical imaging unit relative to the iris optical imaging unit, in mm.

[0133] The inherent object-image optical geometry mapping transformation factor OMTF means that it is uncorrelated and does not depend on external imaging conditions.

[0134] The local region configuration unit of the present invention is used to generate a local region ROI of the iris image with a corresponding fixed pixel resolution in the iris optical imaging unit according to the optical geometric mapping relationship of the optical mapping unit, so as to achieve synchronous dynamic configuration.

[0135] This invention specifically configures a local region of interest (ROI) of the iris image with a fixed pixel resolution, centered at pixel positions (PXi, PYi), and the pixel range of the ROI is (PXroi, PYroi).

[0136] Where: PXroi and PYroi are the XY pixel ranges of the ROI (Region Area) in the iris image.

[0137] cPXi / 16<=PXroi<=cPXi / 2, cPYi / 16<=PYroi<=cPYi / 2

[0138] ROI=RECT(left, top, right, bottom)

[0139] =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXi+PXroi / 2, PYi+PYroi / 2);

[0140] Or the equivalent ROI = RECT(left, top, width, height)

[0141] =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXroi, PYroi).

[0142] In a specific embodiment of the present invention, the iris optical imaging unit acquires iris images based on the center pixel position (PXi, PYi) and pixel range (PXroi, PYroi) of the local region ROI of the iris image with a fixed pixel resolution corresponding to the above-mentioned ROI configuration unit.

[0143] Furthermore, in a specific embodiment of the present invention, iris image acquisition is performed by defining the center pixel position (PXi, PYi) and pixel range (PXroi, PYroi) of the local region (ROI) of the iris image. This is achieved by using the physical analog address array image output (X_ADD_START, Y_ADD_START, X_ADD_END, Y_ADD_END) corresponding to the imaging pixel array passpath of the iris optical imaging unit, or the corresponding digital offset array image output (X_CROP_OFFSET, Y_CROP_OFFSET, X_CROP_WIDTH, Y_CROP_HEIGHT), where...

[0144] X_ADD_START=left=PXi-PXroi / 2,

[0145] Y_ADD_START=top=PYi-PYroi / 2,

[0146] X_ADD_END=right=PXi+PXroi / 2,

[0147] Y_ADD_END=bottom=PYi+PYroi / 2

[0148] or

[0149] X_CROP_OFFSET=left=PXi-PXroi / 2,

[0150] Y_CROP_OFFSET=top=PYi-PYroi / 2,

[0151] X_CROP_WIDTH=width=PXroi,

[0152] Y_CROP_HEIGHT=height=PYroi.

[0153] Furthermore, specific embodiments of the present invention, by employing a defined local region ROI of the iris image, allow for variations in the error accuracy of the detection of the center pixel position of the face / eye detection unit. More importantly, it allows for a reduction in the optical mapping unit, and the execution association state between the local region configuration units is dynamically and synchronously updated. These are important basic characteristics of the present invention, which optimize image acquisition and processing performance and execution efficiency, thereby improving recognition rate and recognition speed.

[0154] Detailed description,

[0155] In a specific embodiment of the present invention, the current real-time dynamic position (PXe, PYe) satisfies the ROI condition limited to the local region of the iris image, and the current association state between the optical mapping unit and the local region configuration unit is maintained to avoid frequent execution of association state synchronization updates.

[0156] In a specific embodiment of the present invention, if the current real-time dynamic position (PXe, PYe) does not meet the condition of being limited to the ROI of the iris image, the optical mapping unit is executed, and the associated state between the local region configuration units is updated synchronously.

[0157] Furthermore, the aforementioned ROI (Region of Interest) constraints of the iris image are transformed into predetermined overlapping fixed partition constraints of the image space of the face optical imaging unit, as follows:

[0158] -PXroi / 2<=OMTF*(PXe-PXeo)<=PXroi / 2

[0159] -PYroi / 2<=OMTF*(PYe-PYeo)<=PYroi / 2

[0160] (PXe, PYe) are the XY coordinates of the center pixel positions of the eyes on the image side of the current real-time dynamic face optical imaging unit.

[0161] The predetermined overlapping fixed partitions are centered at (PXeo, PYeo) and have ranges of (PXezone, PYezone).

[0162] (PXeo, PYeo) = (cPXi / OMTF*n / N, cPYi / OMTF*m / M)

[0163] =(PXroi / OMTF*n,PYroi / OMTF*m)

[0164] n = [1, N-1]

[0165] m = [1, M-1]

[0166] N and M are the predetermined number of fixed unit partitions.

[0167] N = cPXi / PXroi

[0168] M = cPYi / PYroi

[0169] PXezone=PXroi / OMTF, PYezone=PYroi / OMTF

[0170] Essentially, the above mathematical transformations, which satisfy the given conditions, can be further understood to have the following mathematical relationship between point elements and sets:

[0171] (PXe,PYe)∈{(PXroi / OMTF*n±PXezone / 2,PYroi / OMTF*m±PYezone / 2)}

[0172] Specifically, by determining whether the coordinates (PXe, PYe) belong to the corresponding fixed unit partition (n, m) in the predetermined (N, M) fixed unit partition (set of point elements), the associated state of the corresponding fixed unit partition (n, m) is dynamically updated.

[0173] Or, equivalently,

[0174] The above mathematical transformations, satisfying the given conditions, further lead to the following mathematical spatial relationships:

[0175] PXroi / OMTF*n-PXezone / 2<=PXe<=PXroi / OMTF*n+PXezone / 2

[0176] PYroi / OMTF*m-PYezone / 2<=PYe<=PYroi / OMTF*m+PYezone / 2,

[0177] Specifically, by determining the coordinates (PXe, PYe) (point position) and positioning it in the corresponding fixed unit partition (n, m) within the predetermined (N, M) fixed unit partitions (spatial positions), the associated state of the corresponding fixed unit partition (n, m) is dynamically and synchronously updated.

[0178] This invention allows for control over the optical mapping unit and the dynamic synchronization update frequency of the execution association state between local region configuration units by defining a suitable local region ROI pixel range (PXroi, PYroi) of the iris image.

[0179] The key feature of this invention is that by using a limited ROI pixel range (PXroi, PYroi) for a local area of ​​the iris image, the amount of image data transmitted is reduced. Under the same transmission bandwidth, a higher frame rate is achieved, which means a higher image acquisition speed. In addition, the reduction in image data reduces the computational load and system resources for image processing, thereby improving image processing speed performance.

[0180] In a preferred embodiment of the present invention, the original pixel resolution of the iris optical imaging system is 64 Mpixels. By determining a suitable dynamic configuration, the local region ROI of the fixed pixel resolution iris image is limited to a pixel range of PXroi = cPXi / 4 and PYroi = cPYi / 4 centered on the pixel position (PXi, PYi) coordinates, that is, the pixel range of the local region ROI of the fixed pixel resolution iris image is limited to 4 Mpixels.

[0181] Reducing pixel resolution means going further, with higher frame rates such as 60fps and 120fps, increasing the image acquisition frame rate to achieve faster and more accurate image acquisition and processing, faster iris image AF autofocus speed, and improved recognition rate and speed.

[0182] The key feature of this invention is that by using a fixed pixel resolution in the local area configuration unit to achieve synchronous dynamic configuration, it can ensure that the resolution of the image data sending end of the iris optical imaging unit and the image data receiving end of the image acquisition and processing unit remains fixed and the same, without the need to adjust the image resolution transmission mode. During dynamic synchronous configuration, there is no delay and no configuration switching between different resolution modes.

[0183] In a specific embodiment of the present invention, the image acquisition and processing unit connects the above-mentioned units to achieve synchronous image acquisition and image processing calculation. More preferably, in this embodiment, the image acquisition and processing unit includes an ISP / NPU / CPU component, wherein the ISP is used for image acquisition, the NPU is used for image acquisition calculation, and the CPU is used for image acquisition processing. Figure 2 shows a logic diagram of the image acquisition and processing unit according to a specific embodiment of the present invention. Figure 2 This diagram illustrates the parallel pipeline execution architecture of the ISP / NPU / CPU components. This is used to optimize image acquisition and processing performance and efficiency, thereby improving recognition rate and speed.

[0184] The ISP component further includes a dual MIPI CSI for interfaceing the image data (sensor RAW data) input of the face optical imaging unit and the iris optical imaging unit, respectively.

[0185] Based on the input RAW image data, the ISP completes the predetermined image acquisition, including internal functions such as AE / AF, contrast / sharpness enhancement, and denoising.

[0186] The NPU component is used to execute deep learning-based convolutional neural network (CNN) cascade model calculations based on image data acquired by the ISP component. For example, it can implement the face region detection and face eye center pixel position detection output coordinates (PXe, PYe) of the above-mentioned human eye detection unit, and execute neural network model algorithms such as face / iris detection, segmentation feature extraction, etc.

[0187] The CPU component is used to execute feedback processing control of corresponding associated units based on the image data results calculated by the NPU component, such as the image data feedback processing control of the associated execution optical mapping unit, local region configuration unit, and iris image AF autofocus unit.

[0188] In a specific embodiment of the present invention, the image acquisition and processing unit adopts an image periodic parallel synchronous logic timing working mode, which executes the image calculation cycle (TCn+1) and the image processing cycle (TPn) in parallel and synchronously within the current image acquisition cycle (TAn+2).

[0189] Based on the timing requirements of the above parallel synchronous working mode, the timing relationship of the current cycle of this invention satisfies:

[0190] TAn+2>=TCn+1>=TPn

[0191] The time relationship requirements corresponding to the parallel synchronous working mode must meet:

[0192] TA>=TC>=TP

[0193] Image acquisition time (TA) is greater than or equal to image computation time (TC) and greater than or equal to image processing time (TP).

[0194] The iris image autofocus (AF) unit uses the imaging lens drive parameter control of the iris optical imaging unit. When controlling the parameters, the AF autofocus parameter control depends on the imaging lens drive type and is achieved by adjusting the imaging optical image distance or adjusting the imaging optical focal length / diopter.

[0195] The control of the imaging lens drive parameters of the iris optical imaging unit includes:

[0196] The STEP parameters are: number of steps 2*K+1 and step size (unit step length), where K is the number of steps;

[0197] To achieve continuous autofocus in one direction within the focusing range FR = (2 * K + 1) * STEP, e.g., K = 4,

[0198] in:

[0199] Step size STEP = β² * DOF or STEP = [β / (1+β)]² * DOF, depending on the imaging lens drive type. For example, STEP = β² * DOF for adjusting the imaging optical distance, and STEP = [β / (1+β)]² * DOF for adjusting the imaging optical focal length / diopter.

[0200] DOF = 2 * FNO * SOC * (1 + β) / β²

[0201] in:

[0202] β is the lateral optical magnification of the iris optical imaging unit;

[0203] DOF represents the depth of field of the iris optical imaging unit;

[0204] FNO is the aperture parameter of the iris optical imaging unit;

[0205] SOC is the minimum physical spot resolution parameter of the iris optical imaging unit;

[0206] A very important feature of this invention is that when β << 1, autofocus satisfies STEP = 2 * FNO * SOC regardless of any imaging lens drive type. The important characteristic of this method is that for a given iris optical imaging system, the step size has an inherent object-image optical constant and does not depend on the imaging lens drive type.

[0207] The iris image autofocus (AF) unit of this invention uses the imaging lens driving parameters of the iris optical imaging unit to generate a step number of 2*K+1, with a unit step size of STEP = 2*FNO*SOC. This achieves a depth of field within the predetermined focusing range FR that satisfies the object-side focusing accuracy of ±DOF / 2, while minimizing the step number. The unidirectional continuous drive control autofocus method avoids the bidirectional iterative searching of the maximum image focusing quality in traditional methods, ensuring that the autofocus control process can be completed within 0.1s.

[0208] Furthermore, the autofocus (AF) unit of this invention can also employ different types of imaging lens drivers to execute the imaging lens driving parameters of the iris optical imaging unit. For example, a PDAF type driver can achieve this by adjusting the imaging optical image distance or focal length. Based on the characteristics of the PDAF type driver, its essence is to directly respond to the relative distance of the corresponding focusing position through the image-side phase difference, such as a central motor, closed-loop motor, shape memory metal motor, piezoelectric motor, MEMS motor, or other linear or nonlinear motors.

[0209] The AF autofocus unit method based on PDAF of the present invention includes the following steps:

[0210] Step S1. First, calculate PD (ROI). PD is the phase difference corresponding to the ROI region, which depends on different phase pixel detection types such as left-right shield pixel, left-right micro-lens pixel, left-right dual pixel, 2*2 / 8PD omnidirectional phase pixel.

[0211] Step S2. Calculate the relative distance FP of the corresponding focusing position based on the corresponding phase difference.

[0212] FP=PD*CC=STEP*K,K=FP / STEP=CC*PD / STEP

[0213] Step size (unit step length) STEP parameter, STEP = 2 * FNO * SOC

[0214] CC is the conversion coefficient between the phase difference and the corresponding relative distance FP (i.e., the defocus distance, the relative distance from the focus position), which is also known as the defocus rate. The defocus rate is achieved by fitting the actual corresponding linear or nonlinear relationship.

[0215] Step S3. Based on the corresponding focal position relative distance FP, the PDAF type driver executes the imaging lens driving parameters of the iris optical imaging unit, directly driving the corresponding focal position relative distance FP. That is, directly driving STEP*K step positions;

[0216] Step S4. Iterate through steps S1-S3, terminating when PD<=EP, EP=STEP / CC, where EP is the predetermined phase difference focusing error, i.e., terminating at a single step size, K<=1.

[0217] This invention is attributed to the fact that the PDAF type driver directly corresponds to the focus position parameter driving. The above method of this invention can realize the AF autofocus to be completed within 3 iteration cycles, and the corresponding focus position relative distance parameter is executed no more than 3 times directly, so as to complete the object-side depth of field range that meets the focus accuracy ±DOF / 2, and can ensure that the autofocus control process can be completed within 0.03s.

[0218] Example 2

[0219] This embodiment provides a method for acquiring facial iris images, including: a face optical imaging unit, an eye detection unit, an optical mapping unit, a local region configuration unit, an iris optical imaging unit, an iris image autofocus unit, and an image acquisition and processing unit; the image acquisition and processing unit is connected to the face optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, the iris optical imaging unit, and the iris image autofocus unit, and controls the face optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, the iris optical imaging unit, and the iris image autofocus unit.

[0220] Step 1. The image acquisition and processing unit controls the face optical imaging unit to acquire the corresponding global area face image (FaceImage).

[0221] Step 2. The image acquisition and processing unit controls the eye detection unit to detect the center pixel position of the corresponding eyes based on the corresponding global area face image FaceImage;

[0222] Step 3. The image acquisition and processing unit controls the optical mapping unit to perform a transformation of the global region face image FaceImage in response to the iris image IrisImage based on the corresponding center pixel position of the eyes of the face;

[0223] Step 4. The image acquisition and processing unit controls the local area configuration unit to generate the corresponding fixed pixel resolution iris image local area ROI synchronous dynamic configuration based on the corresponding object-image optical geometric mapping relationship.

[0224] Step 5. The image acquisition and processing unit controls the iris optical imaging unit to acquire the corresponding fixed-pixel-resolution iris image (IrisImage) based on the local region ROI of the iris image with the corresponding fixed-pixel-resolution ROI.

[0225] Step 6. The image acquisition and processing unit controls the iris image AF autofocus unit to perform iris image AF autofocus based on the corresponding fixed pixel resolution iris image ROI (IrisImage).

[0226] The human eye detection unit is used to detect the center pixel position of the eyes in the global region of the face image by the face optical imaging unit.

[0227] This invention utilizes the currently known convolutional neural network (CNN) cascade model based on deep learning to reliably and accurately achieve face region detection and output coordinates (PXe, PYe) for the center pixel position detection of the eyes.

[0228] The optical mapping unit is used to realize the object-image optical geometric mapping relationship transformation OMT, (PXe, PYe) : -> (PXi, PYi) based on the center pixel position of the face and eyes of the face / eye detection unit.

[0229] PXi-cPXi / 2

[0230] =Fi / (Zi-Fi)*(Xi) / PSi

[0231] =Fi / (Zi-Fi)*(Xe-Xoffset) / PSi

[0232] =Fi / (Zi-Fi)*[(Ze-Fe) / Fe*(PXe-cPXe / 2)*PSe-Xoffset)] / PSi.

[0233] PYi-cPYi / 2

[0234] =Fi / (Zi-Fi)*(Yi) / PSi

[0235] =Fi / (Zi-Fi)*(Ye-Yoffset) / PSi

[0236] =Fi / (Zi-Fi)*[(Ze-Fe) / Fe*(PYe-cPYe / 2)*PSe-Yoffset)] / PSi.

[0237] Specifically, when the conditions Ze >> Zoffset, Fi, Fe are satisfied,

[0238] (PXi-cPXi / 2)+Fi / (Zi-Fi)*Xoffset / PSi=(PXe-cPXe / 2)*Fi*PSe / (Fe*PSi)

[0239] (PYi-cPYi / 2)+Fi / (Zi-Fi)*Yoffset / PSi=(PYe-cPYe / 2)*Fi*PSe / (Fe*PSi)

[0240] Specifically, when the conditions Ze >> Xoffset, Yoffset, Zoffset, Fi, Fe are satisfied, the important fundamental characteristic of this invention is the inherent object-image optical geometric mapping relationship transformation factor OMTF:

[0241] (PXi-cPXi / 2) / (PXe-cPXe / 2)=

[0242] (PYi-cPYi / 2) / (PYe-cPYe / 2)=

[0243] Fi*PSe / (Fe*PSi)=OMTF

[0244] Where (PXe, PYe) are the XY coordinates of the center pixel position of the eyes on the image side of the face optical imaging unit.

[0245] (cPXe, cPYe) represents the pixel resolution of the XY side of the face optical imaging unit.

[0246] PSe is the unit pixel resolution of the face optical imaging unit, in μm / pixel.

[0247] Fe is the imaging optical focal length of the face optical imaging unit, in mm.

[0248] Ze is the imaging optical distance of the face optical imaging unit, in mm.

[0249] (Xe, Ye) are the XY position coordinates of the object space of the face optical imaging unit, in mm.

[0250] (PXi, PYi) represents the XY coordinates of the center pixel position of the eyes on the image side of the iris optical imaging unit.

[0251] (cPXi, cPYi) represents the pixel resolution of the XY side of the iris optical imaging unit.

[0252] PSi is the resolution per pixel of the iris optical imaging unit, in μm / pixel.

[0253] Fi is the imaging optical focal length of the iris optical imaging unit, in mm.

[0254] Zi represents the imaging optical distance of the iris optical imaging unit, in mm.

[0255] (Xi, Yi) represents the XY coordinates of the object space position of the iris optical imaging unit, in mm.

[0256] (Xoffset, Yoffset, Zoffset)

[0257] Xoffset = Xe-Xi, Yoffset = Ye-Yi, Zoffset = Ze-Zi are the XYZ coordinate offsets of the object space position of the face optical imaging unit relative to the iris optical imaging unit, in mm.

[0258] The inherent object-image optical geometry mapping transformation factor OMTF means that it is uncorrelated and does not depend on external imaging conditions.

[0259] The local region configuration unit of the present invention is used to generate a local region ROI of the iris image with a corresponding fixed pixel resolution in the iris optical imaging unit according to the optical geometric mapping relationship of the optical mapping unit, so as to achieve synchronous dynamic configuration.

[0260] This invention specifically configures a local region of interest (ROI) of the iris image with a fixed pixel resolution, centered at pixel positions (PXi, PYi), and the pixel range of the ROI is (PXroi, PYroi).

[0261] Where: PXroi and PYroi are the XY pixel ranges of the ROI (Region Area) in the iris image.

[0262] cPXi / 16<=PXroi<=cPXi / 2, cPYi / 16<=PYroi<=cPYi / 2

[0263] ROI=RECT(left, top, right, bottom)

[0264] =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXi+PXroi / 2, PYi+PYroi / 2);

[0265] Or the equivalent ROI = RECT(left, top, width, height)

[0266] =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXroi, PYroi).

[0267] In a specific embodiment of the present invention, the iris optical imaging unit acquires iris images based on the center pixel position (PXi, PYi) and pixel range (PXroi, PYroi) of the local region ROI of the iris image with a fixed pixel resolution corresponding to the above-mentioned ROI configuration unit.

[0268] Furthermore, in a specific embodiment of the present invention, iris image acquisition is performed by defining the center pixel position (PXi, PYi) and pixel range (PXroi, PYroi) of the local region (ROI) of the iris image. This is achieved by using the physical analog address array image output (X_ADD_START, Y_ADD_START, X_ADD_END, Y_ADD_END) corresponding to the imaging pixel array passpath of the iris optical imaging unit, or the corresponding digital offset array image output (X_CROP_OFFSET, Y_CROP_OFFSET, X_CROP_WIDTH, Y_CROP_HEIGHT), where...

[0269] X_ADD_START=left=PXi-PXroi / 2,

[0270] Y_ADD_START=top=PYi-PYroi / 2,

[0271] X_ADD_END=right=PXi+PXroi / 2,

[0272] Y_ADD_END=bottom=PYi+PYroi / 2

[0273] or

[0274] X_CROP_OFFSET=left=PXi-PXroi / 2,

[0275] Y_CROP_OFFSET=top=PYi-PYroi / 2,

[0276] X_CROP_WIDTH=width=PXroi,

[0277] Y_CROP_HEIGHT=height=PYroi.

[0278] Furthermore, specific embodiments of the present invention, by employing a defined local region ROI of the iris image, allow for variations in the error accuracy of the detection of the center pixel position of the face / eye detection unit. More importantly, it allows for a reduction in the optical mapping unit, and the execution association state between the local region configuration units is dynamically and synchronously updated. These are important basic characteristics of the present invention, which optimize image acquisition and processing performance and execution efficiency, thereby improving recognition rate and recognition speed.

[0279] Detailed description,

[0280] In a specific embodiment of the present invention, the current real-time dynamic position (PXe, PYe) satisfies the ROI condition limited to the local region of the iris image, and the current association state between the optical mapping unit and the local region configuration unit is maintained to avoid frequent execution of association state synchronization updates.

[0281] In a specific embodiment of the present invention, if the current real-time dynamic position (PXe, PYe) does not meet the condition of being limited to the ROI of the iris image, the optical mapping unit is executed, and the associated state between the local region configuration units is updated synchronously.

[0282] Furthermore, the aforementioned ROI (Region of Interest) constraints of the iris image are transformed into predetermined overlapping fixed partition constraints of the image space of the face optical imaging unit, as follows:

[0283] -PXroi / 2<=OMTF*(PXe-PXeo)<=PXroi / 2

[0284] -PYroi / 2<=OMTF*(PYe-PYeo)<=PYroi / 2

[0285] (PXe, PYe) are the XY coordinates of the center pixel positions of the eyes on the image side of the current real-time dynamic face optical imaging unit.

[0286] The predetermined overlapping fixed partitions are centered at (PXeo, PYeo) and have ranges of (PXezone, PYezone).

[0287] (PXeo, PYeo) = (cPXi / OMTF*n / N, cPYi / OMTF*m / M)

[0288] =(PXroi / OMTF*n,PYroi / OMTF*m)

[0289] n = [1, N-1]

[0290] m = [1, M-1]

[0291] N and M are the predetermined number of fixed unit partitions.

[0292] N = cPXi / PXroi

[0293] M = cPYi / PYroi

[0294] PXezone=PXroi / OMTF, PYezone=PYroi / OMTF

[0295] Essentially, the above mathematical transformations, which satisfy the given conditions, can be further understood to have the following mathematical relationship between point elements and sets:

[0296] (PXe,PYe)∈{(PXroi / OMTF*n±PXezone / 2,PYroi / OMTF*m±PYezone / 2)}

[0297] Specifically, by determining whether the coordinates (PXe, PYe) belong to the corresponding fixed unit partition (n, m) in the predetermined (N, M) fixed unit partition (set of point elements), the associated state of the corresponding fixed unit partition (n, m) is dynamically updated.

[0298] Or, equivalently,

[0299] The above mathematical transformations, satisfying the given conditions, further lead to the following mathematical spatial relationships:

[0300] PXroi / OMTF*n-PXezone / 2<=PXe<=PXroi / OMTF*n+PXezone / 2

[0301] PYroi / OMTF*m-PYezone / 2<=PYe<=PYroi / OMTF*m+PYezone / 2,

[0302] Specifically, by determining the coordinates (PXe, PYe) (point position) and positioning it in the corresponding fixed unit partition (n, m) within the predetermined (N, M) fixed unit partitions (spatial position), the associated state of the corresponding fixed unit partition (n, m) is dynamically and synchronously updated.

[0303] This invention allows for control over the optical mapping unit and the dynamic synchronization update frequency of the execution association state between local region configuration units by using a defined appropriate ROI pixel range (PXroi, PYroi) of the iris image local region.

[0304] The key feature of this invention is that by using a limited ROI pixel range (PXroi, PYroi) for a local area of ​​the iris image, the amount of image data transmitted is reduced. Under the same transmission bandwidth, a higher frame rate is achieved, which means a higher image acquisition speed. In addition, the reduction in image data reduces the computational load and system resources for image processing, thereby improving image processing speed performance.

[0305] In a preferred embodiment of the present invention, the original pixel resolution of the iris optical imaging system is 64 Mpixels. By determining a suitable dynamic configuration, the local region ROI of the fixed pixel resolution iris image is limited to a pixel range of PXroi = cPXi / 4 and PYroi = cPYi / 4 centered on the pixel position (PXi, PYi) coordinates, that is, the pixel range of the local region ROI of the fixed pixel resolution iris image is limited to 4 Mpixels.

[0306] Reducing pixel resolution means going further, with higher frame rates such as 60fps and 120fps, increasing the image acquisition frame rate to achieve faster and more accurate image acquisition and processing, faster iris image AF autofocus speed, and improved recognition rate and speed.

[0307] The key feature of this invention is that by using a fixed pixel resolution in the local area configuration unit to achieve synchronous dynamic configuration, it can ensure that the resolution of the image data sending end of the iris optical imaging unit and the image data receiving end of the image acquisition and processing unit remains fixed and the same, without the need to adjust the image resolution transmission mode. During dynamic synchronous configuration, there is no delay and no configuration switching between different resolution modes.

[0308] In a specific embodiment of the present invention, the image acquisition and processing unit connects the above-mentioned units to achieve synchronous image acquisition and image processing calculation. More preferably, in this embodiment, the image acquisition and processing unit includes an ISP / NPU / CPU component, wherein the ISP is used for image acquisition, the NPU is used for image acquisition calculation, and the CPU is used for image acquisition processing. Figure 2 shows a logic diagram of the image acquisition and processing unit according to a specific embodiment of the present invention. Figure 2 This diagram illustrates the parallel pipeline execution architecture of the ISP / NPU / CPU components. This is used to optimize image acquisition and processing performance and efficiency, thereby improving recognition rate and speed.

[0309] The ISP component further includes a dual MIPI CSI for interfaceing the image data (sensor RAW data) input of the face optical imaging unit and the iris optical imaging unit, respectively.

[0310] Based on the input RAW image data, the ISP completes the predetermined image acquisition, including internal functions such as AE / AF, contrast / sharpness enhancement, and denoising.

[0311] The NPU component is used to execute deep learning-based convolutional neural network (CNN) cascade model calculations based on image data acquired by the ISP component. For example, it can implement the face region detection and face eye center pixel position detection output coordinates (PXe, PYe) of the above-mentioned human eye detection unit, and execute neural network model algorithms such as face / iris detection, segmentation feature extraction, etc.

[0312] The CPU component is used to execute feedback processing control of corresponding associated units based on the image data results calculated by the NPU component, such as the image data feedback processing control of the associated execution optical mapping unit, local region configuration unit, and iris image AF autofocus unit.

[0313] In a specific embodiment of the present invention, the image acquisition and processing unit adopts an image periodic parallel synchronous logic timing working mode, which executes the image calculation cycle (TCn+1) and the image processing cycle (TPn) in parallel and synchronously within the current image acquisition cycle (TAn+2).

[0314] Based on the timing requirements of the above parallel synchronous working mode, the timing relationship of the current cycle of this invention satisfies:

[0315] TAn+2>=TCn+1>=TPn

[0316] The time relationship requirements corresponding to the parallel synchronous working mode must meet:

[0317] TA>=TC>=TP

[0318] Image acquisition time (TA) is greater than or equal to image computation time (TC) and greater than or equal to image processing time (TP).

[0319] The iris image autofocus (AF) unit uses the imaging lens drive parameter control of the iris optical imaging unit. When controlling the parameters, the AF autofocus parameter control depends on the imaging lens drive type and is achieved by adjusting the imaging optical image distance or adjusting the imaging optical focal length / diopter.

[0320] The control of the imaging lens drive parameters of the iris optical imaging unit includes:

[0321] The STEP parameters are: number of steps 2*K+1 and step size (unit step length), where K is the number of steps;

[0322] To achieve continuous autofocus in one direction within the focusing range FR = (2 * K + 1) * STEP, e.g., K = 4,

[0323] in:

[0324] Step size STEP = β² * DOF or STEP = [β / (1+β)]² * DOF, depending on the imaging lens drive type. For example, STEP = β² * DOF for adjusting the imaging optical distance, and STEP = [β / (1+β)]² * DOF for adjusting the imaging optical focal length / diopter.

[0325] DOF = 2 * FNO * SOC * (1 + β) / β²

[0326] in:

[0327] β is the lateral optical magnification of the iris optical imaging unit;

[0328] DOF represents the depth of field of the iris optical imaging unit;

[0329] FNO is the aperture parameter of the iris optical imaging unit;

[0330] SOC is the minimum physical spot resolution parameter of the iris optical imaging unit;

[0331] A very important feature of this invention is that when β << 1, autofocus satisfies STEP = 2 * FNO * SOC regardless of any imaging lens drive type. The important characteristic of this method is that for a given iris optical imaging system, the step size has an inherent object-image optical constant and does not depend on the imaging lens drive type.

[0332] The iris image autofocus (AF) unit of this invention uses the imaging lens driving parameters of the iris optical imaging unit to generate a step number of 2*K+1, with a unit step size of STEP = 2*FNO*SOC. This achieves a depth of field within the predetermined focusing range FR that satisfies the object-side focusing accuracy of ±DOF / 2, while minimizing the step number. The unidirectional continuous drive control autofocus method avoids the bidirectional iterative searching of the maximum image focusing quality in traditional methods, ensuring that the autofocus control process can be completed within 0.1s.

[0333] Furthermore, the autofocus (AF) unit of this invention can also employ different types of imaging lens drivers to execute the imaging lens driving parameters of the iris optical imaging unit. For example, a PDAF type driver can achieve this by adjusting the imaging optical image distance or focal length. Based on the characteristics of the PDAF type driver, its essence is to directly respond to the relative distance of the corresponding focusing position through the image-side phase difference, such as a central motor, closed-loop motor, shape memory metal motor, piezoelectric motor, MEMS motor, or other linear or nonlinear motors.

[0334] The AF autofocus unit method based on PDAF of the present invention includes the following steps:

[0335] Step S1. First, calculate PD (ROI). PD is the phase difference corresponding to the ROI region, which depends on different phase pixel detection types such as left-right shield pixel, left-right micro-lens pixel, left-right dual pixel, 2*2 / 8PD omnidirectional phase pixel.

[0336] Step S2. Calculate the relative distance FP of the corresponding focusing position based on the corresponding phase difference.

[0337] FP=PD*CC=STEP*K,K=FP / STEP=CC*PD / STEP

[0338] Step size (unit step length) STEP parameter, STEP = 2 * FNO * SOC

[0339] CC is the conversion coefficient between the phase difference and the corresponding relative distance FP (i.e., the defocus distance, the relative distance from the focus position), which is also known as the defocus rate. The defocus rate is achieved by fitting the actual corresponding linear or nonlinear relationship.

[0340] Step S3. Based on the corresponding focal position relative distance FP, the PDAF type driver executes the imaging lens driving parameters of the iris optical imaging unit, directly driving the corresponding focal position relative distance FP. That is, directly driving STEP*K step positions;

[0341] Step S4. Iterate through steps S1-S3, terminating when PD<=EP, EP=STEP / CC, where EP is the predetermined phase difference focusing error, i.e., terminating at a single step size, K<=1.

[0342] This invention is attributed to the fact that the PDAF type driver directly corresponds to the focus position parameter driving. The above method of this invention can realize the AF autofocus to be completed within 3 iteration cycles, and the corresponding focus position relative distance parameter is executed no more than 3 times directly, so as to complete the object-side depth of field range that meets the focus accuracy ±DOF / 2, and can ensure that the autofocus control process can be completed within 0.03s.

[0343] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations, such as parameter adjustments, component changes, or step substitutions. However, any obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A device for acquiring facial iris images, characterized in that, include: Face optical imaging unit, iris optical imaging unit, optical imaging association unit, image acquisition and processing unit; The image acquisition and processing unit is connected to the face optical imaging unit, the iris optical imaging unit, and the optical imaging association unit, respectively, and controls the face optical imaging unit, the iris optical imaging unit, and the optical imaging association unit. The face optical imaging unit outputs a global area face image; The optical imaging association unit is used to establish a synchronous association relationship between the pre-defined face optical imaging unit and iris optical imaging unit for outputting global-local region images; The iris optical imaging unit outputs a corresponding local area iris image according to the synchronization correlation; The imaging field of view of the face optical imaging unit is greater than or equal to the imaging field of view of the iris optical imaging unit. The optical imaging association unit includes: a human eye detection unit, an optical mapping unit, and a local region configuration unit; The face optical imaging unit is used to acquire global area face images; The human eye detection unit is used to detect the position of the center pixels of the two eyes of the face from the global area face image acquired by the face optical imaging unit; The optical mapping unit is used to transform the optical geometric mapping relationship between the face optical imaging unit and the iris optical imaging unit based on the center pixel position of the two eyes detected by the human eye detection unit. The local region configuration unit is used to synchronously and dynamically configure the local region of the iris image with a fixed pixel resolution corresponding to the iris optical imaging unit according to the object-image optical geometric mapping relationship obtained by the optical mapping unit. The iris optical imaging unit is used to acquire local iris images based on a local region of the iris image with a fixed pixel resolution synchronized by the local region configuration unit.

2. The device for acquiring facial iris images according to claim 1, characterized in that, When the optical mapping unit transforms the object-image optical geometric mapping relationship of the face optical imaging unit in response to the iris optical imaging unit according to the center pixel position of the human eye detection unit, the calculation method of the transformation factor OMTF is as follows: OMTF = Fi * PSe / (Fe * PSe) = (PXi - cPXi / 2) / (PXe - cPXe / 2) = (PYi - cPYi / 2) / (PYe - cPYe / 2) Wherein, OMTF is the transformation factor with inherent object-image optical geometric mapping relationship; Fi is the imaging optical focal length of the iris optical imaging unit; PSe is the unit pixel resolution of the face optical imaging unit; Fe is the imaging optical focal length of the face optical imaging unit; PSi is the unit pixel resolution of the iris optical imaging unit; (PXi, PYi) are the XY coordinates of the center pixel position of the eyes on the image side of the iris optical imaging unit; (cPXi, cPYi) are the XY pixel resolution of the iris optical imaging unit; (PXe, PYe) are the XY coordinates of the center pixel position of the eyes on the image side of the face optical imaging unit; and (cPXe, cPYe) are the XY pixel resolution of the face optical imaging unit.

3. The device for acquiring facial iris images according to claim 1, characterized in that, By using a limited local region of the iris image, the frequency of dynamic synchronization update of the execution association state between the optical mapping unit and the local region configuration unit is controlled.

4. The device for acquiring facial iris images according to claim 1, characterized in that, When the iris optical imaging unit acquires a local iris image based on the iris image local area synchronized by the local area configuration unit, the local iris image is the image acquired within the area enclosed by the pixel position (PXi, PYi) coordinates and the pixel range (PXroi, PYroi). Where: PXroi and PYroi are the XY pixel ranges of a local region of the iris image; cPXi / 16<=PXroi<=cPXi / 2, cPYi / 16<=PYroi<=cPYi / 2 ROI=RECT(left, top, right, bottom) =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXi+PXroi / 2, PYi+PYroi / 2); Or ROI = RECT (left, top, width, height) =RECT(PXi-PXroi / 2, PYi-PYroi / 2, PXroi, PYroi).

5. The device for acquiring facial iris images according to claim 4, characterized in that, When acquiring local iris images based on pixel positions (PXi, PYi) and pixel ranges (PXroi, PYroi), the physical simulation local array image output (X_ADD_START, Y_ADD_START, X_ADD_END, Y_ADD_END) or the corresponding digital offset local array image output (X_CROP_OFFSET, Y_CROP_OFFSET, X_CROP_WIDTH, Y_CROP_HEIGHT) corresponding to the imaging pixel array path of the iris optical imaging unit is used. in: X_ADD_START=left=PXi-PXroi / 2, Y_ADD_START=top=PYi-PYroi / 2, X_ADD_END=right=PXi+PXroi / 2, Y_ADD_END=bottom=PYi+PYroi / 2; or X_CROP_OFFSET=left=PXi-PXroi / 2, Y_CROP_OFFSET=top=PYi-PYroi / 2, X_CROP_WIDTH=width=PXroi, Y_CROP_HEIGHT=height=PYroi.

6. The device for acquiring facial iris images according to claim 1, characterized in that, It also includes an iris image AF autofocus unit, the image acquisition and processing unit is connected to the iris image AF autofocus unit and controls the iris image AF autofocus unit; The iris image AF autofocus unit is used to automatically focus the local area iris image based on the local area iris image corresponding to the iris optical imaging unit, so as to obtain the focused local area iris image.

7. The apparatus for acquiring facial iris images according to claim 6, characterized in that, The iris image AF autofocus unit is controlled by the imaging lens drive parameters of the iris optical imaging unit.

8. The apparatus for acquiring facial iris images according to claim 7, characterized in that, The parameter control of the iris image AF autofocus unit depends on the imaging lens drive type, and is achieved by adjusting the imaging optical image distance or adjusting the imaging optical focal length / diopter.

9. The apparatus for acquiring facial iris images according to claim 7, characterized in that, The imaging lens driving parameter control of the iris optical imaging unit includes: step number 2*K+1 and step size STEP parameter, where K is the step number; Achieve continuous autofocus in one direction within the predetermined focusing range FR = (2 * K + 1) * STEP; in: STEP = 2 * FNO * SOC; FNO is the aperture parameter of the iris optical imaging unit; SOC is the minimum physical spot resolution parameter of the iris optical imaging unit.

10. The apparatus for acquiring facial iris images according to claim 9, characterized in that, The method for controlling the imaging lens driving parameters of the iris optical imaging unit includes: Step S1: Calculate PD, where PD is the phase difference corresponding to a local region of the iris image; Step S2: Calculate the relative distance FP of the corresponding focusing position based on the phase difference obtained in step S1. The calculation formula is as follows: FP = PD * CC; Where CC is the conversion coefficient between the phase difference and the relative distance FP of the corresponding focus position, i.e., the defocus rate. The defocus rate is achieved by fitting the actual corresponding linear or nonlinear relationship. Step S3: Based on the relative distance FP of the focal position obtained in step S2, drive the PDAF type driver to execute the imaging lens driving parameters of the iris optical imaging unit, and drive the corresponding relative distance FP of the focal position, that is, directly drive STEP*K step positions. Step S4 iterates through steps S1-S3 until PD <= EP, where EP = STEP / CC, and EP is the predetermined phase difference focusing error. In other words, it terminates at a single step size, where K <= 1.

11. A method for acquiring facial iris images, characterized in that, include: Face optical imaging unit, eye detection unit, optical mapping unit, local region configuration unit, iris optical imaging unit, and image acquisition and processing unit; The image acquisition and processing unit is connected to the face optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, and the iris optical imaging unit, respectively, and controls the face optical imaging unit, the eye detection unit, the optical mapping unit, the local region configuration unit, and the iris optical imaging unit. The specific steps are as follows: Step 1: The image acquisition and processing unit controls the face optical imaging unit to acquire face images of the entire region; Step 2: The image acquisition and processing unit controls the human eye detection unit to detect the center pixel positions of the eyes of the face from the global face image acquired in Step 1. Step 3: Based on the center pixel positions of the eyes of the face detected in Step 2, the image acquisition and processing unit controls the optical mapping unit to transform the optical geometric mapping relationship between the face optical imaging unit and the iris optical imaging unit. Step four: Based on the object-image optical geometric mapping relationship obtained in step three, the image acquisition and processing unit synchronously and dynamically configures the local region of the iris image with a fixed pixel resolution corresponding to the iris optical imaging unit. Step 5: The image acquisition and processing unit controls the iris optical imaging unit to acquire the iris image of the local area based on the local iris image obtained by synchronous dynamic configuration in step 4.

12. The method for acquiring facial iris images as described in claim 11, characterized in that, It also includes an iris image AF autofocus unit, the image acquisition and processing unit is connected to the iris image AF autofocus unit and controls the iris image AF autofocus unit; A sixth step is set after step five; Step six: The image acquisition and processing unit controls the automatic focusing of the local iris image based on the local iris image acquired in step five, and obtains the focused local iris image.

Images