AA focusing method and focusing system of HD lens module

Abstract

The invention relates to the technical field of lens module focusing, in particular to an AA focusing method and focusing system of an HD lens module. The AA focusing method of the HD lens module comprises the following steps: step 1, data acquisition and preprocessing; 2, constructing a deep reinforcement learning model; step 3, dynamic focusing, and step 4, fixing and retesting: fixing the adjusted lens and light source by dispensing UV and thermosetting glue, retesting the optical performance after the glue is cured, and confirming that the glue is cured so as to meet the optical performance requirement. According to the AA focusing method of the HD lens module, the powerful feature extraction capability of deep learning and the dynamic optimization capability of reinforcement learning are combined, the focusing task can be efficiently completed, and it is ensured that the optical performance meets the requirement.

Description

Technical Field

[0001] The present invention relates to the technical field of lens module focusing, and in particular to an AA focusing method and a focusing system for an HD lens module. Background Art

[0002] With the rapid development of the automotive lighting field, intelligence and pixelation have become one of the mainstream trends in the future development of vehicle lights. Among them, HD pixel headlights with Micro-LED as the carrier have the dual advantages of high pixels and low cost, and have become an optimal solution with high mass production potential; however, the assembly accuracy requirements for HD pixel headlights have also increased synchronously.

[0003] During the assembly process of Micro-LED and the lens module, focusing is required to make the projected pattern clear. After focusing, screws are generally used for locking to fix the lens and lock the focal length. During the process of using screw locking, the torque of the screw will cause the lens to deflect slightly, and the gap between the two structures (the lens and the light source) will be compacted during the locking process, making the originally clear projection effect blurred; moreover, the focusing process is slow, and manual focusing and human eye judgment are required. Manual focusing is difficult to meet the high-precision requirements and is prone to introducing errors; the focusing takes a long time and cannot meet the requirements of large-scale production; for complex optical systems, it is difficult to dynamically adjust manual focusing to adapt to different light source and lens combinations.

[0004] In a lens module with perfect precision, the optical axis center of the lens and the optical axis center of the light source overlap. However, the errors in the manufacturing process are inevitable. Using the traditional two-dimensional focusing (only adjusting the distance in the Z-axis direction between the lens and the sensor) process to manufacture the lens module can no longer meet the requirements of high pixels. Therefore, AA focusing came into being. AA focusing adjusts through multiple degrees of freedom, including translation in the X, Y, and Z directions and adjustment of the tilt angle, to ensure that the optical axis of the lens is concentric with the optical axis of the image sensor. The existing AA focusing collects the white screen pattern (light pattern distribution) in real time through a camera, and feeds back the MTF (Modulation Transfer Function) values of each measurement point to the focusing device through an algorithm to achieve automatic focusing, that is, gradually translate in the X, Y, and Z directions according to the errors of each measurement point, and finally adjust the tilt angle. Such a gradual adjustment process will also bring errors and requires repeated adjustment, and it is impossible to further improve the accuracy and efficiency. Summary of the Invention

[0005] The technical problem to be solved by the present invention is: to provide an AA focusing method for an HD lens module in order to solve the problems existing in the prior art in the above-mentioned background art.

[0006] The technical solution adopted by the present invention to solve its technical problems is: an AA focusing method for an HD lens module, including the following steps: S1. Data acquisition and preprocessing: Collect data on the initial positions of the light source and the lens and the light pattern distribution in the HD lens module through an optical sensor and a position sensor, and preprocess the collected data. S2. Construct a deep reinforcement learning model: Input the preprocessed data into a deep learning model for training, extract the light pattern distribution features, optimize the focusing parameters through a reinforcement learning algorithm, and guide the model to find the optimal solution through a reward mechanism. S3. Dynamic focusing: Adjust the positions of the lens and the light source according to the initial parameters output by the deep reinforcement learning model, monitor the light pattern distribution in real time through an optical sensor, calculate the error from the target distribution, optimize the focusing parameters using the reinforcement learning algorithm based on the error data, adjust the positions of the lens and the light source, and continue to calculate the error data during the adjustment process until the adjustment of the lens and the light source stops after reaching the preset accuracy. S4. Fixing and remeasurement: Fix the adjusted lens and light source with point UV + thermosetting glue, and perform an optical performance remeasurement after the glue cures to confirm the glue curing to meet the optical performance requirements.

[0007] Furthermore, in step S1, the initial position of the light source includes the coordinates and angle of the light source, the initial position of the lens includes the focal length and tilt angle of the lens, and the light pattern distribution includes the spot shape and light intensity distribution.

[0008] Even further, the preprocessing in step S1 includes normalization, noise reduction, and feature extraction. The normalization is to convert the data into a distribution with a mean of 0 and a standard deviation of 1, and the calculation formula is:

[0009] where μ is the mean of the initial position data of the light source and the lens, and σ is the standard deviation of the initial position data of the light source and the lens; The noise reduction is to compress the input initial position data of the light source and the lens into low-dimensional features using the encoder part of an autoencoder, and the decoder part attempts to reconstruct the input data to remove noise through training; The feature extraction is to automatically learn the features of the light pattern distribution using a convolutional neural network.

[0010] Furthermore, in step S2, the deep learning model uses a convolutional neural network; the reinforcement learning algorithm uses Q-learning, and its goal is to learn a Q-value function Q(s, a) , which represents the expected return of taking action s in state a . The update formula of the Q-value function Q(s, a) is:

[0011] wherer is an immediate reward; γ is the discount factor; s′ is the next state; a′ the next action of; α is the learning rate; is the next state s′ the maximum Q-value of all possible actions; A convolutional neural network is used to approximate the Q-value function, and the training process is stabilized through experience replay and a target network. Among them, the experience replay buffer stores state transitions ( s, a, r, s′ ), and a batch of samples is randomly selected for update during training; in reinforcement learning, the optimal policy is obtained by maximizing the cumulative reward, and the reward signal r represents the immediate feedback after taking a certain action in a certain state and is used to update the Q-value function.

[0012] Further, the input of the deep reinforcement learning model in step S2 includes the initial position parameters of the light source and the lens and the light pattern distribution data, and its output is the adjustment parameters of the lens and the light source, and the adjustment parameters include the displacement amount and the rotation angle.

[0013] There is also provided an AA focusing system for an HD lens module, including a hardware module and a software module, The hardware module includes a position sensor, an optical sensor, a focusing tooling, and a motor driving device. The HD lens module is installed on the focusing tooling. The position sensor is arranged on the mounting bracket of the HD lens module. The optical sensor is a CCD camera, which is arranged at a certain distance away from the HD lens module. The motor driving device is installed on the focusing tooling of the HD lens module; The software module includes a deep reinforcement learning model, a focusing algorithm, and a control interface. The deep reinforcement learning model is used to analyze the image data collected by the optical sensor and predict the best focusing position; the focusing algorithm controls the electric driving device to adjust the positions of the lens and the light source on the HD lens module according to the prediction result of the deep reinforcement learning model; the control interface has a CAN data acquisition module, which is connected to the CAN bus of the vehicle through the CAN interface to realize the communication between the hardware module and the software module.

[0014] Further, the HD lens module includes a lens, a light source, and a mounting bracket. The lens and the light source are arranged on the mounting bracket and are distributed front and back. The mounting bracket is arranged on the focusing tooling; A reduction lens is arranged on the front side of the lens, a white screen is arranged on the front side of the reduction lens, and a CCD camera for photographing the light pattern on the white screen is arranged above the reduction lens.

[0015] Further, the light pattern on the white screen is obtained by projecting the preset light pattern on the light source onto the white screen after the light emitted by the Micro-LED is divided by the lens and the reduction lens.

[0016] Further, the focusing tooling includes an X-axis module, a Y-axis module, a Z-axis module, and a rotation module. The X-axis module is installed on the bottom plate, the Y-axis module is arranged on the X-axis module, the Z-axis module is arranged on the Y-axis module, and the rotation module is arranged on the Z-axis module.

[0017] Furthermore, the motor driving device includes a first adjustment motor for driving the X-axis module to move, a second adjustment motor for driving the Y-axis module to move, a third adjustment motor for driving the Z-axis module to move, and a fourth adjustment motor for driving the rotation module to rotate horizontally.

[0018] Advantages of the present invention: The AA focusing method of the HD lens module of the present invention combines the powerful feature extraction ability of deep learning and the dynamic optimization ability of reinforcement learning, can efficiently complete the focusing task, and ensure that the optical performance meets the requirements; Based on the deep reinforcement learning method, by predicting the defocus distance, only a small number of images are required to complete focusing, and the focusing time is only 15% - 24% of the traditional method; this method directly predicts the optimal focusing position through a deep learning model, avoiding the focusing error and multiple adjustments caused by local extreme points in the traditional method; The deep reinforcement learning model can learn complex image features, so as to more accurately predict the focusing position; Through multi-degree-of-freedom adjustment, the imaging quality of the high-pixel lens module is ensured; after being trained with a large number of samples, the deep reinforcement learning model can adapt to different shooting scenes and samples, and has good versatility; The deep reinforcement learning model can dynamically adjust the focusing parameters according to real-time feedback, adapting to complex environments and different working conditions; during the focusing process, the model can dynamically adjust the lens position according to the light pattern distribution error monitored in real time until the preset accuracy is reached; It provides strong support for the automation and intelligence of the optical imaging system, and helps the technological upgrading of related fields. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention will be further described below with reference to the drawings and embodiments.

[0020] Figure 1 is a schematic structural diagram of the AA focusing system of the HD lens module of the present invention.

[0021] Figure 2 is Figure 1 the schematic structural diagram of the HD lens module in

[0022] Figure 3Yes Figure 1 Focus adjustment and optical path schematic diagram.

[0023] Figure 4 It is a simulation result diagram under the cumulative tolerance of the actual sample of the HD lens module.

[0024] Figure 5 It is a flowchart of the AA focus adjustment method for the HD lens module of the present invention.

[0025] Figure 6 It is the initial image using the AA focus adjustment method of the present invention.

[0026] Figure 7 It is the image after X-axis adjustment using the AA focus adjustment method of the present invention.

[0027] Figure 8 It is the image after Y-axis adjustment using the AA focus adjustment method of the present invention.

[0028] Figure 9 It is the image after Z-axis adjustment using the AA focus adjustment method of the present invention.

[0029] Figure 10 It is the image after horizontal rotation using the AA focus adjustment method of the present invention.

[0030] In the figure: 1. Optical sensor; 2. Focus adjustment tooling; 21. X-axis module; 22. Y-axis module; 23. Z-axis module; 24. Rotation module; 3. HD lens module; 31. Lens; 32. Light source; 33. Mounting bracket; 4. Reduction lens; 5. White screen; 10. Base plate. Detailed implementation manners

[0031] Now, the present invention will be further described in detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic manner, so they only show the components related to the present invention.

[0032] Embodiment 1 As Figure 1 shown, the AA focus adjustment system of the HD lens module includes a hardware module and a software module. The hardware module includes a position sensor (not shown in the figure), an optical sensor 1, a focus adjustment tooling 2, and a motor drive device. The HD lens module 3 is installed on the focus adjustment tooling 2. The position sensor is arranged on the mounting bracket 33 of the HD lens module 3. The optical sensor 1 is a CCD camera, which is arranged at a certain distance away from the HD lens module 3. The motor drive device is installed on the focus adjustment tooling 2 of the HD lens module 3.

[0033] As Figure 2As shown, the HD lens module 3 includes a lens 31, a light source 32, and a mounting bracket 33. The lens 31 and the light source 32 are arranged on the mounting bracket 33 in a front-to-back distribution, and the mounting bracket 33 is arranged on the focusing tooling 2. Combining Figure 1 , the focusing tooling 2 includes an X-axis module 21, a Y-axis module 22, a Z-axis module 23, and a rotation module 24. The X-axis module 21 is installed on the bottom plate 10, the Y-axis module 22 is arranged on the X-axis module 21, the Z-axis module 23 is arranged on the Y-axis module 22, and the rotation module 24 is arranged on the Z-axis module 23. Correspondingly, the motor drive device includes a first adjustment motor for driving the X-axis module 21 to move, a second adjustment motor for driving the Y-axis module 22 to move, a third adjustment motor for driving the Z-axis module 23 to move, and a fourth adjustment motor for driving the rotation module 24 to rotate horizontally.

[0034] Since the displacement amount of adjustment and the rotation angle are relatively small, Figure 2 What is shown is an adjustment handle. The adjustment handle is connected to a screw rod, and the screw rod is connected to the slider of the corresponding module (the Z-axis module 23 is a swing rod; the rotation module 24 is a rotating block). Rotating the adjustment handle drives the screw rod to rotate, causing the corresponding slider to move in the corresponding direction (the rotating block rotates). In order to cooperate with the control of the software module and improve the focusing efficiency, the adjustment handle is replaced with an adjustment motor.

[0035] As Figure 1 and Figure 3 shown, a reduction lens 4 is arranged on the front side of the lens 31, a white screen 5 is arranged on the front side of the reduction lens 4, and a CCD camera 6 for photographing the light pattern on the white screen 5 is arranged above the reduction lens 4. The Micro-LED is lit, and a pre-designed light pattern is generated on the light source 32. After the light emitted by the Micro-LED passes through the lens 31 and the reduction lens 4, the preset light pattern on the light source 32 is projected onto the white screen 5, and the optical sensor 1 (CCD camera) can capture the light pattern on the white screen 5.

[0036] According to the actual assembly effect of the HD lens module 3, it is required to deflect ±0.05° around the lens 31 in the Z-axis direction. The sample part tolerance conversion is shown in Table 1: Table 1 Sample Part Tolerance Conversion Table

[0037] Based on the above total cumulative tolerance, when the HD lens module 3 makes an X-axis direction adjustment, with a Y-axis direction tolerance of +0.225 mm, a Z-axis direction tolerance of +0.22 mm, and an X-axis direction tolerance of +0.02 mm, the simulation results are as Figure 4As shown, the image clarity can basically meet the focusing requirements, but the image center is deviated. Through optical simulation verification, the lens deflection is relatively sensitive and needs to be controlled within 0.1°, so it is necessary to increase the adjustment function of the lens deflection. Therefore, the focusing tooling 2 needs to be adjusted in the Z axis, Y axis, Z axis and horizontal rotation direction. Of course, if the installation structure allows, a vertical rotation module can also be set.

[0038] In view of this, it is necessary to optimize the software module of the AA focusing system of the HD lens module. The software module of this embodiment includes a deep reinforcement learning model, a focusing algorithm and a control interface. The deep reinforcement learning model is used to analyze the image data collected by the optical sensor 1 and predict the optimal focusing position; the focusing algorithm controls the electric drive device to adjust the position of the lens 31 and the light source 32 on the HD lens module 3 according to the prediction results of the deep reinforcement learning model; the control interface has a CAN data acquisition module, which is connected to the vehicle's CAN bus through the CAN interface to realize communication between the hardware module and the software module.

[0039] Embodiment 2 like Figure 5 As shown, the focusing method of the AA focusing system of the HD lens module of the first embodiment includes the following steps: S1, data collection and preprocessing: collect the data of the initial position and light pattern distribution of the light source 32 and the lens 31 in the HD lens module 3 through the optical sensor 1 and the position sensor, and preprocess the collected data; S2. Build a deep reinforcement learning model: input the preprocessed data into the deep learning model for training, extract the light distribution characteristics, optimize the focusing parameters through the reinforcement learning algorithm, and guide the model to find the optimal solution through the reward mechanism; S3, dynamic focusing: adjust the position of the lens 31 and the light source 32 according to the initial parameters output by the deep reinforcement learning model, monitor the light pattern distribution in real time through the optical sensor 1, calculate the error with the target distribution, optimize the focusing parameters according to the error data using the reinforcement learning algorithm, adjust the position of the lens 31 and the light source 32, and continue to calculate the error data during the adjustment process until the preset accuracy is reached and stop adjusting the lens 31 and the light source 32; S4, fixation and retest: fix the adjusted lens 31 and light source 32 by applying UV+thermosetting glue, and retest the optical performance after the glue is cured to confirm that the glue is cured to meet the optical performance requirements.

[0040] Specifically, the initial position of the light source 32 includes the coordinates and angle of the light source 32, the initial position of the lens 31 includes the focal length and tilt angle of the lens 31, and the light pattern distribution includes the light spot shape and light intensity distribution. Preprocessing includes normalization, noise reduction and feature extraction. Normalization is to convert the data into a distribution with a mean of 0 and a standard deviation of 1. The calculation formula is:

[0041] Among them, μ is the mean of the initial position data of the light source 32 and the lens 31, and σ is the standard deviation of the initial position data of the light source 32 and the lens 31; for noise reduction, the encoder part of the autoencoder is used to compress the input initial position data of the light source 32 and the lens 31 into low-dimensional features, and the decoder part attempts to reconstruct the input data, and the noise is removed through training; for feature extraction, the convolutional neural network is used to automatically learn the features of the light pattern distribution.

[0042] The deep learning model uses a convolutional neural network; the reinforcement learning algorithm uses Q-learning, and its goal is to learn a Q-value function Q(s, a) , representing the expected return of taking action s in state a , and the update formula of the Q-value function Q(s, a) is:

[0043] Among them, r is the immediate reward; γ is the discount factor; s′ is the next state; a′ the next action of α is the learning rate; is the maximum Q-value of all possible actions in the next state s′ ; A convolutional neural network is used to approximate the Q-value function, and the training process is stabilized through experience replay and target network. Among them, the experience replay buffer stores state transitions ( s, a, r, s′ ), and a batch of samples is randomly selected for update during training; in reinforcement learning, the optimal policy is obtained by maximizing the cumulative reward, and the reward signal r represents the immediate feedback after taking a certain action in a certain state and is used to update the Q-value function.

[0044] The input of the deep reinforcement learning model includes the initial position parameters of the light source 32 and the lens 31 and the light pattern distribution data, and its output is the adjustment parameters of the lens 31 and the light source 32, and the adjustment parameters include the displacement amount and the rotation angle.

[0045] During actual focusing, the initial data is shown in Table 2, and the presented image is as Figure 6 shown, and the image clarity is very poor.

[0046] Table 2 Initial data of the lens barrel and the light source

[0047] The focusing method of Embodiment 2 is used for automatic focusing, as Figures 7 - 10As shown, the adjustment is carried out successively in the X-axis, Y-axis, Z-axis, and horizontal rotation directions. During the adjustment process, real-time compensation will be performed. For example, when adjusting the Y-axis, the adjustment parameters of the lens 31 and the light source 32 output by the deep reinforcement learning model will compensate for the displacement of the previous X-axis adjustment; when adjusting the Z-axis, the adjustment parameters of the lens 31 and the light source 32 output by the deep reinforcement learning model will compensate for the displacements of the X-axis and Y-axis adjustments until all adjustments are completed. The image effect after focusing is as Figure 10 shown. Its image clarity meets the focusing requirements, and there is no deviation in the center of the image. The adjusted data is shown in Table 3.

[0048] Table 3 Data after the adjustment of the lens barrel and the light source

[0049] The focusing of the prior art takes 5 to 10 minutes. The focusing time using the focusing method of Embodiment 2 is about 1 minute, which improves the efficiency and reduces the labor cost. And through the precise focusing process, the optical performance consistency of each lens module can be guaranteed, reducing the defective rate, thereby reducing the production cost. The focusing parameters optimized by deep reinforcement learning can make the optical performance of the lens module reach the best state and improve the imaging quality of the product. The optimized focusing parameters and the fixed lens module can maintain stable optical performance during long-term use, extending the service life of the product.

[0050] Inspired by the ideal embodiments of the present invention described above, through the above description, relevant staff can make various changes and modifications within the scope not deviating from the technical idea of the present invention. The technical scope of the present invention is not limited to the content in the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. An AA focusing method for an HD lens module, characterized in that, The steps include: S1, data collection and preprocessing: using an optical sensor (1) and a position sensor, collecting data on the initial position of the light source (32) and the lens (31) in the HD lens module (3) and light pattern distribution, and preprocessing the collected data; S2. Build a deep reinforcement learning model: input the preprocessed data into the deep learning model for training, extract the light distribution characteristics, optimize the focusing parameters through the reinforcement learning algorithm, and guide the model to find the optimal solution through the reward mechanism; S3, dynamic focusing: adjusting the positions of the lens (31) and the light source (32) according to the initial parameters output by the deep reinforcement learning model, monitoring the light pattern distribution in real time through the optical sensor (1), calculating the error with the target distribution, optimizing the focusing parameters using the reinforcement learning algorithm according to the error data, adjusting the positions of the lens (31) and the light source (32), and continuing to calculate the error data during the adjustment process until the preset accuracy is reached, and then stopping the adjustment of the lens (31) and the light source (32); S4, fixation and retest: fix the adjusted lens (31) and light source (32) by applying UV+thermosetting glue, and retest the optical performance after the glue is cured to confirm that the glue is cured to meet the optical performance requirements.

2. The AA focusing method of the HD lens module according to claim 1, wherein: In step S1, the initial position of the light source (32) includes the coordinates and angle of the light source (32), the initial position of the lens (31) includes the focal length and tilt angle of the lens (31), and the light pattern distribution includes the light spot shape and light intensity distribution.

3. The AA focusing method of the HD lens module according to claim 2, wherein: The preprocessing in step S1 includes normalization, noise reduction and feature extraction. The normalization is to convert the data into a distribution with a mean of 0 and a standard deviation of 1. The calculation formula is: Wherein, μ is the mean of the initial position data of the light source (32) and the lens (31), and σ is the standard deviation of the initial position data of the light source (32) and the lens (31); The noise reduction is to use the encoder part of the autoencoder to compress the input initial position data of the light source (32) and the lens (31) into low-dimensional features, and the decoder part attempts to reconstruct the input data and remove noise through training; The feature extraction is to automatically learn the features of light type distribution using a convolutional neural network.

4. The AA focusing method of the HD lens module according to claim 1, wherein: In step S2, the deep learning model uses a convolutional neural network; the reinforcement learning algorithm uses Q-learning, and its goal is to learn a function of Q value Q (s, a) , representing the expected return of taking action s in state a , and the update formula of the function Q(s,a) of Q value is as follows: Among them, r is the immediate reward; γ is the discount factor; s′ is the next state; a′ is the next action of α is the learning rate; is the next state s′ is the maximum Q value of all possible actions; Use a convolutional neural network to approximate the Q-value function, and stabilize the training process through experience replay and target networks. Among them, the experience replay buffer stores state transitions ( s,a,r,s′ ), and randomly samples a batch of samples for update during training; in reinforcement learning, the optimal policy is obtained by maximizing the cumulative reward, and the reward signal r represents the immediate feedback after taking a certain action in a certain state and is used to update the Q-value function.

5. The AA focusing method of the HD lens module according to claim 1, wherein: The input of the deep reinforcement learning model in step S2 includes the initial position parameters of the light source and the lens and the light type distribution data, and the output thereof is the adjustment parameters of the lens and the light source, and the adjustment parameters include the displacement and the rotation angle.

6. An AA focusing system for an HD lens module, characterized in that: Including hardware modules and software modules, The hardware module comprises a position sensor, an optical sensor (1), a focusing fixture (2) and a motor drive device, the HD lens module (3) is mounted on the focusing fixture (2), the position sensor is arranged on a mounting frame (33) of the HD lens module (3), the optical sensor (1) is a CCD camera, which is arranged at a certain distance away from the HD lens module (3), and the motor drive device is mounted on the focusing fixture of the HD lens module (3); The software module includes a deep reinforcement learning model, a focusing algorithm, and a control interface. The deep reinforcement learning model is used to analyze the image data collected by the optical sensor (1) and predict the optimal focusing position. The focusing algorithm controls the electric drive device to adjust the positions of the lens (31) and the light source (32) on the HD lens module (3) according to the prediction result of the deep reinforcement learning model. The control interface has a CAN data acquisition module, which is connected to the vehicle's CAN bus through the CAN interface to realize the communication between the hardware module and the software module.

7. The AA focusing system of the HD lens module according to claim 6, characterized in that: The HD lens module (3) includes a lens (31), a light source (32), and a mounting bracket (33). The lens (31) and the light source (32) are arranged on the mounting bracket (33) in a front-back distribution, and the mounting bracket (33) is arranged on the focusing tooling (2). A reduction lens (4) is arranged on the front side of the lens (31), a white screen (5) is arranged on the front side of the reduction lens (4), and a CCD camera for photographing the light pattern on the white screen (5) is arranged above the reduction lens (4).

8. The AA focusing system of the HD lens module according to claim 7, wherein: The light pattern on the white screen (5) is obtained by the light emitted by the Micro-LED passing through the lens (31) and the reduction lens (4) and projecting the preset light pattern on the light source (32) onto the white screen (5).

9. The AA focusing system of the HD lens module according to claim 6, wherein: The focusing tooling (2) includes an X-axis module (21), a Y-axis module (22), a Z-axis module (23), and a rotation module (24). The X-axis module (21) is installed on the bottom plate (10), the Y-axis module (22) is arranged on the X-axis module (21), the Z-axis module (23) is arranged on the Y-axis module (22), and the rotation module (24) is arranged on the Z-axis module (23).

10. The AA focusing system of the HD lens module according to claim 9, characterized in that: The motor drive device includes a first adjustment motor for driving the X-axis module (21) to move, a second adjustment motor for driving the Y-axis module (22) to move, a third adjustment motor for driving the Z-axis module (23) to move, and a fourth adjustment motor for driving the rotation module (24) to rotate horizontally.

Images