Camera focusing system and focusing method of box girder inspection robot

Abstract

The invention discloses a camera focusing system and focusing method for a box girder inspection robot, and the system comprises a distance measuring sensor, a zoom lens, a focusing control module, and an image sensor, the distance measuring sensor is used for measuring the distance information from a camera to a target; the focusing control module is used for calculating a pre-focusing position according to the distance information and the lens parameters and driving the zoom lens to move to the pre-focusing position through a motor; the image sensor is used for capturing an image and supporting an automatic focusing algorithm based on contrast or phase detection; the focusing control module determines the direction and size of a focusing error based on phase detection pixels of the captured image so as to optimize a focusing position; or, by evaluating the contrast ratio in the captured image, when the contrast ratio is maximum, it is determined that the image is the optimal focusing position. The distance sensor is combined for distance measurement, the focal length of the camera is quickly adjusted according to the measurement result, and the image definition is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of box girder inspection, and in particular to a camera focusing system and a focusing method of a box girder inspection robot. Background Art

[0002] At present, the inspection of the interior of the continuous box girder belongs to the inspection of the hidden parts of the bridge structure. The inspection work inside the box girder faces problems such as high-altitude work, limited power and network, crossing mountains and water, and short window operation time, which brings certain safety hazards to manual work. At the same time, due to the wide distribution of bridges and the large geographical space span, there are many items that need to be inspected in the box girder, including water seepage, cracks, hollows, etc., which makes the manual work task heavy, high-intensity and time-consuming.

[0003] The lighting conditions inside the box girder are poor, the surface materials are diverse, and the space sizes vary greatly. Currently, there are two main methods for detecting internal defects of box girders: manual and inspection robots.

[0004] The manual method is to manually use a camera to perform a rough long-distance scan to determine the location of fatigue cracks; not only is the labor intensity high, but the test results are also biased. At the same time, the internal structure of the box girder is inconvenient to walk, and some are not equipped with lighting, especially the top plate cannot be observed at close range, and it is not convenient to mark and record. Manual inspection is interfered by many objective conditions and is also greatly affected by human subjectivity. In addition, if it is a manual focus camera, the shooting process is time-consuming and inefficient, and it is not suitable for fast and continuous inspection tasks; if it is an autofocus camera, especially for shooting the inner wall of the box girder in a low-contrast, low-light environment, multiple potential focuses appear in the same scene at the same time, and the autofocus camera may misjudge the focus position.

[0005] The inspection robot method uses an image acquisition device carried by the inspection robot to scan the inside of the box girder. However, it often faces poor lighting conditions inside the box girder, diverse surface materials, and large differences in space size, which makes the captured images blurry and low-contrast, making it difficult to accurately identify cracks and other defects. In addition, the image acquisition device uses a combination of multiple cameras or a rotating mechanical device to carry a camera to obtain images of the inner wall of the box girder. Both methods may cause inconsistent image brightness and contrast due to insufficient light inside the box girder. The internal environment of the box girder is complex, and it may be difficult to achieve automatic focusing, resulting in blurry captured images. Summary of the invention

[0006] In view of the above problems existing in the prior art, the present invention provides a camera focusing system and focusing method for a box girder inspection robot.

[0007] The present invention discloses a camera focusing system of a box girder inspection robot, comprising: an inspection robot body, an image acquisition device for inspecting and photographing the box girder is arranged on the inspection robot body, the image acquisition device comprises: a distance sensor, a zoom lens, a focus control module and an image sensor, the distance sensor and the image sensor are connected to the focus control module, and the focus control module is connected to the zoom lens;

[0008] The distance measuring sensor is used to measure the distance information from the camera to the target;

[0009] The focus control module is used to calculate the pre-focus position according to the distance information and the lens parameters, and drive the zoom lens to move to the pre-focus position through the motor;

[0010] The image sensor is used to capture images and support an autofocus algorithm based on contrast or phase detection; the focus control module determines the direction and size of the focus error based on the phase detection pixels of the captured image to optimize the focus position; or, by evaluating the contrast in the captured image, when the contrast is the largest, it is determined to be the best focus position.

[0011] As a further improvement of the present invention, the distance measuring sensor is a laser radar or an ultrasonic sensor, and the distance measuring sensor is coaxial with the camera or as close to the camera as possible to ensure the accuracy of measuring the distance.

[0012] As a further improvement of the present invention, a ring light frame is provided on the periphery of the image acquisition device, the ring light frame is connected to the inspection robot body, a plurality of independent fill lights and light sensors are evenly distributed on the ring light frame, the fill lights can be independently adjusted in brightness and angle, and all the fill lights and light sensors are connected to a control system;

[0013] The number of the light sensors is the same as that of the fill lights, and the light area formed by the N fill lights is divided into N small areas, and the N light sensors detect the light intensity of the corresponding small areas;

[0014] The control system adjusts the brightness and illumination angle of the corresponding fill light based on the light intensity detected by each light sensor;

[0015] The image acquisition device is connected to the control system, and the control system automatically adjusts the exposure parameters of the camera according to the light intensity detected by the light sensor to adapt to environmental changes.

[0016] As a further improvement of the present invention, the light sensor is installed at the front end of the corresponding fill light.

[0017] As a further improvement of the present invention, the control system adjusts the brightness and illumination angle of the corresponding fill light based on the light intensity detected by each light sensor; comprising:

[0018] Calculate the light intensity difference between the light intensity detected by each light sensor and its reference light intensity value, and compare the light intensity difference with the light intensity threshold. If it is greater than the light intensity threshold, calculate the driving current value corresponding to the fill light that needs to be adjusted, and send a dimming instruction to the control system, and the control system controls the corresponding fill light to adjust the brightness; wherein the reference light intensity value is: the average light intensity value of this area corresponding to the clearest image collected by the image acquisition device under laboratory environment testing;

[0019] The image acquisition device acquires an image of the inner wall of the box girder, calibrates the relationship between the image coordinates and the actual coordinates of the inner wall of the box girder, and draws a corresponding fill light area in the image coordinate area; according to a computer image processing algorithm, the color image acquired by the image acquisition device is converted into a grayscale image, and each divided small image area is divided into a plurality of small color blocks; the average pixel value of each small color block, that is, the brightness value, is calculated; the average value of the brightness value of each small color block that divides the small image area is taken as the brightness average value of the small image area; the brightness value of each small color block is compared with the brightness average value of the small image area, if the brightness values ​​of M consecutive small color blocks are all less than the brightness average value, then the center point of the central color block among the M color blocks is taken as a reference, and according to the relationship between the image coordinates and the actual coordinates of the inner wall of the box girder, the rotation angle of the fill light is calculated, and the angle of the fill light is adjusted through the control system to perform fill light processing on the small image area.

[0020] As a further improvement of the present invention, the fill light is an LED lamp, a xenon lamp or a laser lamp.

[0021] As a further improvement of the present invention, the annular light stand is a U-shaped stand with its opening facing downward and its two ends are detachably connected to the inspection robot body through quick-release structures. A reinforcement stand is installed in the middle of the annular light stand, and the top of the reinforcement stand is screwed to the cross bar of the annular light stand, and the bottom end is detachably connected to the inspection robot body through a quick-release structure. The quick-release structure includes one of a snap-on structure and a plug and slot matching structure.

[0022] As a further improvement of the present invention, the control system of the inspection robot body is connected to the controller of each fill light to achieve brightness adjustment of each fill light.

[0023] As a further improvement of the present invention, the fill light is installed on an electric pan-tilt platform, the electric pan-tilt platform is installed on the annular light stand, and the control system of the inspection robot body is connected to the controller of each electric pan-tilt platform to achieve angle adjustment of each fill light.

[0024] The present invention also discloses a camera focusing method for a box girder inspection robot, comprising:

[0025] Distance measurement: The distance measuring sensor transmits a pulse signal to the target (such as the inner wall of the box beam), receives the reflected signal, and calculates the distance information based on the time difference of the signal;

[0026] Pre-focus: The measured distance information is sent to the focus control module. The focus control module calculates the pre-focus position based on the distance information and lens parameters (such as focal length, depth of field, etc.), and drives the zoom lens to move to the pre-focus position through the motor;

[0027] Autofocus Optimization:

[0028] Enables the phase detection autofocus (PDAF) algorithm, which determines the direction and size of the focus error by analyzing the phase detection pixels on the image sensor to optimize the focus position; since there is already an approximate focus position provided by the laser sensor, PDAF can quickly lock the final focus position, reduce the search range, and speed up the focusing speed;

[0029] Enables the contrast-based autofocus (CDAF) algorithm, which continuously adjusts the position of the lens and evaluates the contrast in the image at each position. When the maximum contrast is found, the best focus position is considered to be reached; CDAF is usually slower than PDAF, but can provide more refined focus results, especially in low light or scenes with unclear textures.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] The present invention can perform automatic pre-focusing based on the measured distance, and further perform fine-tuning focusing based on the automatic focusing algorithm. Based on this, the present invention can realize automatic focusing of the camera, thereby enabling it to capture clear and high-quality images;

[0032] The present invention arranges an annular light frame on the periphery of the image acquisition device, and a plurality of fill lights are arranged on the annular light frame. The annular multi-fill light fill light structure arranged on the periphery of the image acquisition device can provide sufficient lighting for the inspection robot in a low-light or no-light environment, thereby ensuring that the image acquisition device and other sensors carried by the robot can capture clear, high-quality images and data.

[0033] The annular light stand of the present invention is installed on the inspection robot body by using a quick-disassembly structure, which facilitates the disassembly of the annular light stand and the inspection robot body, so that both can conveniently enter the box beam from the manhole at the bottom of the box beam;

[0034] The brightness and irradiation angle of each fill light of the present invention can be automatically adjusted based on the light intensity and position of the imaging area, thereby ensuring the light uniformity of the image shooting and further improving the imaging effect of the image acquisition device. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a schematic diagram of a camera focusing system of a box girder inspection robot disclosed in the present invention;

[0036] Figure 2 for Figure 1 The framework diagram of the image acquisition device;

[0037] Figure 3 for Figure 1 Schematic diagram of the middle ring light stand;

[0038] Figure 4 for Figure 1 Installation diagram of the fill light.

[0039] In the figure:

[0040] 1. Inspection robot body; 2. Image acquisition device; 3. Ring light stand; 4. Fill light; 5. Electric pan / tilt; 6. Light sensor; 7. Control system. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] The present invention is further described in detail below in conjunction with the accompanying drawings:

[0043] like Figure 1 , 2As shown, the present invention provides a camera focusing system for a box girder inspection robot, comprising: a focusing system arranged on the inspection robot body and a fill-in light system for automatically filling in light during the focusing process, the focusing system comprising an image acquisition device arranged on the inspection robot body and used for inspecting and photographing the box girder, the image acquisition device comprising: a ranging sensor, a zoom lens, a focus control module and an image sensor, the ranging sensor and the image sensor are connected to the focus control module, and the focus control module is connected to the zoom lens; wherein the ranging sensor is used to measure the distance information from the camera to the target; the focus control module is used to calculate a pre-focus position according to the distance information and lens parameters, and drive the zoom lens to move to the pre-focus position through a motor; the image sensor is used to capture images and support an autofocus algorithm based on contrast or phase detection; the focus control module determines the direction and size of the focus error based on the phase detection pixels of the captured image to optimize the focus position; or, by evaluating the contrast in the captured image, when the contrast is maximum, it is determined to be the best focus position.

[0044] Furthermore, the distance measuring sensor of the present invention is a laser radar or an ultrasonic sensor, and the distance measuring sensor is coaxial with the camera or as close to the camera as possible to ensure the accuracy of measuring the distance.

[0045] Furthermore, a serial interface is used to connect the laser sensor and the focus control module to ensure smooth data communication between the two devices.

[0046] The present invention provides a camera focusing method for a box girder inspection robot, comprising:

[0047] 1) Distance measurement: The distance measuring sensor transmits a pulse signal to the target (such as the inner wall of the box beam), receives the reflected signal, and calculates the distance information based on the time difference of the signal;

[0048] 2) Pre-focus: The measured distance information is sent to the focus control module, which calculates the pre-focus position based on the distance information and lens parameters (such as focal length, depth of field, etc.), and drives the zoom lens to move to the pre-focus position through the motor;

[0049] 3) Autofocus optimization:

[0050] Enables the phase detection autofocus (PDAF) algorithm, which determines the direction and size of the focus error by analyzing the phase detection pixels on the image sensor to optimize the focus position; since there is already an approximate focus position provided by the laser sensor, PDAF can quickly lock the final focus position, reduce the search range, and speed up the focusing speed;

[0051] Enables the contrast-based autofocus (CDAF) algorithm, which continuously adjusts the position of the lens and evaluates the contrast in the image at each position. When the maximum contrast is found, the best focus position is considered to be reached; CDAF is usually slower than PDAF, but can provide more refined focus results, especially in low light or scenes with unclear textures.

[0052] 4) Image capture: When autofocus is completed and reaches the optimal focus position, the camera captures a clear image.

[0053] When inspecting the inside of the box girder, the autofocus system of the present invention needs to work closely with other sensors and systems of the robot to ensure an efficient and accurate inspection process. The following is the collaborative working method of the autofocus system and the robot system:

[0054] Collaboration with the navigation system: The autofocus system receives the robot position and movement speed information provided by the navigation system, predicts and adjusts the focal length to adapt to the robot's movement and position changes.

[0055] Collaboration with the ambient light detection system: Based on the light intensity data provided by the ambient light detection system, the autofocus system adjusts the aperture size and exposure time to ensure image clarity under different lighting conditions.

[0056] Collaboration with distance measurement sensors: Using distance information provided by laser rangefinders or ultrasonic sensors, the autofocus system can adjust the focus more accurately to ensure that the target object is always in sharp focus.

[0057] Collaboration with image processing system: The autofocus system shares image data with the image processing system and adjusts the focus based on image clarity feedback. At the same time, the image processing system can further analyze the image adjusted by the autofocus system.

[0058] Collaboration with the camera mobile platform: The camera is fixed on the box girder robot by connecting the pan / tilt head, and the autofocus system can adjust the camera angle and focal length according to the tilt angle of the platform.

[0059] like Figure 1 , 3-4, the present invention provides a camera focusing system for a box girder inspection robot, wherein the fill light system for automatically filling light during the focusing process is an improvement on the fill light structure of the existing box girder inspection robot; the present invention is provided with an annular light stand 3 on the periphery of the image acquisition device, the annular light stand 3 is connected to the inspection robot body 1 through the above-mentioned quick-release structure, and a plurality of independent fill light lamps 4 and light sensors 6 are evenly distributed on the annular light stand 3. Specifically: the fill light 4 is an LED lamp, a xenon lamp or a laser lamp, the light sensor 6 is a photodiode, a phototransistor or a photoresistor, etc., the light sensor 6 can measure the light intensity in the current environment and convert the light intensity into an electrical signal, these sensors usually have high sensitivity and a wide dynamic range, and can accurately sense various lighting conditions from dark to bright. The annular light stand 3 is preferably a door-shaped frame (U-shaped frame), a circular frame or a C-shaped frame with the opening facing downward, so as to form an annular lighting area around the image acquisition device 2; the material of the annular light stand 3 is selected according to actual needs to meet the use requirements of the inspection robot; the image acquisition device 2 is located at the center of the annular light stand 3, such as Figure 2 The center position shown. Furthermore, when the annular light stand is a U-shaped stand with the opening facing downward, its two ends are detachably connected to the inspection robot body through a quick-release structure, and a reinforcement bracket is installed in the middle of the annular light stand. The top of the reinforcement bracket is screwed to the cross bar of the annular light stand, and the bottom is detachably connected to the inspection robot body through a quick-release structure. The quick-release device of the present invention can use a socket quick-release device similar to a seat belt, install a lock buckle at a preset position of the box girder inspection robot body, and install a plug at the tail end of the annular light stand 3. When the plug is inserted into the lock buckle, the annular light stand 3 can be stably installed on the box girder inspection robot body.

[0060] The fill light 4 of the present invention is installed on the annular light frame 3 through the electric pan-tilt 5. Specifically, the fill light 4 is installed on the movable end of the electric pan-tilt 5, and the fixed end of the electric pan-tilt 5 is fixedly installed on the annular light frame 3; the angle adjustment of the fill light in the horizontal and vertical directions can be achieved through the electric pan-tilt 5. The electric pan-tilt involved in the present invention is an existing conventional structure, so it will not be elaborated on here; for example, a household camera is a combination of a commonly used camera + an electric pan-tilt. Further, the control system of the inspection robot body of the present invention is connected to the controller of each electric pan-tilt to achieve angle adjustment of each fill light. Each fill light of the present invention can achieve independent control of brightness and angle to adapt to the environment and target detection objects at different positions inside the box girder, and avoid overexposure of the image due to excessive brightness or loss of image details due to excessive darkness.

[0061] The fill light of the present invention is evenly distributed in different parts of the robot, and can provide multi-angle lighting, so as to provide the best lighting effect in different detection scenes. The fill light system is powered by the internal battery of the robot and is centrally managed by the control system of the robot. The fill light system has a quick-release function, which is convenient for rapid deployment and recovery, and is also convenient for migration from the outside of the box beam to the inside of the box beam through the manhole at the bottom of the box beam.

[0062] The number of light sensors 6 of the present invention is consistent with that of the fill-in lights 4, and the illumination area formed by the N fill-in lights is divided into N small areas. The N light sensors detect the light intensity of the corresponding small areas. The light sensor 6 is installed at the front end of the corresponding fill-in light 4, and can be installed on the lampshade of the fill-in light 4 or on the movable end of the electric pan-tilt platform of the fill-in light. The control system 7 of the inspection robot body 1 is connected to the controller of each fill-in light 4 (which controls the brightness of the fill-in light and the movement of the electric pan-tilt platform) and each light sensor 6, so as to adjust the brightness and illumination angle of the corresponding fill-in light based on the light intensity detected by each light sensor. Figure 2 For example, the control system of the inspection robot body 1 of the present invention connects the controllers of all the fill-in lights in parallel through wires, and the controller can realize the switch and brightness adjustment of the fill-in lights, and then the switch and brightness adjustment of each fill-in light can be realized through the control system of the inspection robot body 1. The image acquisition device 4 is connected to the control system 7, and the control system 7 automatically adjusts the exposure parameters of the camera according to the light intensity detected by the light sensor to adapt to environmental changes.

[0063] The control system 7 adjusts the brightness and illumination angle of the corresponding fill light based on the light intensity detected by each light sensor, specifically including:

[0064] S1. Divide the area illuminated by the fill light into N small areas equal to the number of fill lights. The center of each small area is the illumination center of the fill light. The fill lights and the partitioned small areas are numbered one by one. Install a light sensor in the small area where the fill light is located to detect the light intensity of the nearby area.

[0065] S2. Test in a laboratory environment, adjust the fill light on the electric pan / tilt to the reference position (the central axis of the fill light is parallel to the normal of the measured area), adjust the brightness of the fill light, make the image acquired by the camera of the box girder robot to the clearest state, and record the brightness value of the light sensor in each small area, and take the average brightness value of each light sensor as the reference value for brightness adjustment.

[0066] S3. In the actual detection environment, the light sensor detects the light intensity of the small area where it is located and obtains the light intensity value of the small area; calculates the light intensity difference between the light intensity detected by each light sensor and its reference light intensity value, and compares the light intensity difference (absolute value) with the light intensity threshold. If it is greater than the light intensity threshold, the driving current value corresponding to the fill light that needs to be adjusted is calculated, and a dimming instruction is issued to the control system, and the control system controls the corresponding fill light to adjust the brightness; further, according to the different light intensities, multiple thresholds can be set to distinguish different lighting conditions, such as "bright", "medium", "dim", "dark" and other levels can be defined, and the adjustment strategy of the light source is judged by the IF-THEN rule. For example, if the light intensity difference (absolute value) is greater than the threshold X, the light source brightness is increased by Y%. At the same time, a closed-loop control system of a PID controller is used to continuously adjust the light source brightness through a light source compensation device according to the data fed back by the light sensor, and automatically turn on or adjust the brightness of the fill light according to the data fed back by the light sensor, while keeping the image quality within an ideal range.

[0067] S4, the image acquisition device acquires the image of the inner wall of the box girder, calibrates the relationship between the image coordinates and the actual box girder inner wall coordinates, and draws the corresponding fill light area in the image coordinate area; according to the computer image processing algorithm, the color image acquired by the image acquisition device is converted into a grayscale image, and each divided image small area is divided into a number of small color blocks of appropriate size; the average pixel value of each small color block, that is, the brightness value, is calculated; the average value of the brightness value of each small color block that divides the image small area is taken as the brightness average value of the small image area; the brightness value of each small color block is compared with the brightness average value of the small image area, if the brightness values ​​of M consecutive small color blocks are all less than the brightness average value, then the center point of the central color block among the M color blocks is taken as the reference (for example, if the brightness values ​​of 5 consecutive small color blocks are all less than the brightness average value, then the center point of the third color block among the 5 color blocks is taken as the reference), according to the relationship between the image coordinates and the actual box girder inner wall coordinates, the rotation angle of the fill light is calculated, and the angle of the fill light is adjusted by the control system to perform fill light processing on the small image area.

[0068] The advantages of the present invention are:

[0069] The present invention can perform automatic pre-focusing based on the measured distance, and further perform fine-tuning focusing based on the automatic focusing algorithm. Based on this, the present invention can realize automatic focusing of the camera, thereby enabling it to capture clear and high-quality images;

[0070] The present invention arranges an annular light frame on the periphery of the image acquisition device, and a plurality of fill lights are arranged on the annular light frame. The annular multi-fill light fill light structure arranged on the periphery of the image acquisition device can provide sufficient lighting for the inspection robot in a low-light or no-light environment, thereby ensuring that the image acquisition device and other sensors carried by the robot can capture clear, high-quality images and data.

[0071] The annular light stand of the present invention is installed on the inspection robot body by using a quick-disassembly structure, which facilitates the disassembly of the annular light stand and the inspection robot body, so that both can conveniently enter the box beam from the manhole at the bottom of the box beam;

[0072] The brightness and irradiation angle of each fill light of the present invention can be automatically adjusted based on the light intensity and position of the imaging area, thereby ensuring the light uniformity of the image shooting and further improving the imaging effect of the image acquisition device.

[0073] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A camera focusing system for a box girder inspection robot, comprising: The inspection robot body is provided with an image acquisition device for inspecting and photographing the box girder, wherein the image acquisition device comprises: a distance sensor, a zoom lens, a focus control module and an image sensor, wherein the distance sensor and the image sensor are connected to the focus control module, and the focus control module is connected to the zoom lens; The distance measuring sensor is used to measure the distance information from the camera to the target; The focus control module is used to calculate the pre-focus position according to the distance information and the lens parameters, and drive the zoom lens to move to the pre-focus position through the motor; The image sensor is used to capture images and support an autofocus algorithm based on contrast or phase detection; the focus control module determines the direction and size of the focus error based on the phase detection pixels of the captured image to optimize the focus position; or, by evaluating the contrast in the captured image, when the contrast is the largest, it is determined to be the best focus position.

2. The camera focusing system of the box girder inspection robot according to claim 1, characterized in that: The distance measuring sensor is a laser radar or an ultrasonic sensor, and the distance measuring sensor is coaxial with the camera or close to the camera.

3. The camera focusing system of the box girder inspection robot according to claim 1, characterized in that: An annular light frame is arranged on the periphery of the image acquisition device, and the annular light frame is connected to the inspection robot body. A plurality of independent fill lights and light sensors are evenly distributed on the annular light frame. The fill lights can be independently adjusted in brightness and angle, and all the fill lights and light sensors are connected to a control system; The number of the light sensors is the same as that of the fill lights, and the light area formed by the N fill lights is divided into N small areas, and the N light sensors detect the light intensity of the corresponding small areas; The control system adjusts the brightness and illumination angle of the corresponding fill light based on the light intensity detected by each light sensor; The image acquisition device is connected to the control system, and the control system automatically adjusts the exposure parameters of the camera according to the light intensity detected by the light sensor to adapt to environmental changes.

4. The camera focusing system of the box girder inspection robot according to claim 3, characterized in that: The light sensor is installed at the front end of the corresponding fill light.

5. The camera focusing system of the box girder inspection robot according to claim 3, characterized in that: The control system adjusts the brightness and illumination angle of the corresponding fill light based on the light intensity detected by each light sensor; including: Calculate the light intensity difference between the light intensity detected by each light sensor and its reference light intensity value, and compare the light intensity difference with the light intensity threshold. If it is greater than the light intensity threshold, calculate the driving current value corresponding to the fill light that needs to be adjusted, and send a dimming instruction to the control system, and the control system controls the corresponding fill light to adjust the brightness; wherein the reference light intensity value is: the average light intensity value of this area corresponding to the clearest image collected by the image acquisition device under laboratory environment testing; The image acquisition device acquires an image of the inner wall of the box girder, calibrates the relationship between the image coordinates and the actual coordinates of the inner wall of the box girder, and draws a corresponding fill light area in the image coordinate area; according to a computer image processing algorithm, the color image acquired by the image acquisition device is converted into a grayscale image, and each divided small image area is divided into a plurality of small color blocks; the average pixel value of each small color block, that is, the brightness value, is calculated; the average value of the brightness value of each small color block that divides the small image area is taken as the brightness average value of the small image area; the brightness value of each small color block is compared with the brightness average value of the small image area, if the brightness values ​​of M consecutive small color blocks are all less than the brightness average value, then the center point of the central color block among the M color blocks is taken as a reference, and according to the relationship between the image coordinates and the actual coordinates of the inner wall of the box girder, the rotation angle of the fill light is calculated, and the angle of the fill light is adjusted through the control system to perform fill light processing on the small image area.

6. The camera focusing system of the box girder inspection robot according to claim 3, characterized in that: The fill light is an LED lamp, a xenon lamp or a laser lamp.

7. The camera focusing system of the box girder inspection robot according to claim 3, characterized in that: The annular light stand is a U-shaped stand with its opening facing downward and its two ends are detachably connected to the inspection robot body through quick-release structures. A reinforcement stand is installed in the middle of the annular light stand. The top of the reinforcement stand is screwed to the cross bar of the annular light stand, and the bottom end is detachably connected to the inspection robot body through a quick-release structure. The quick-release structure includes one of a snap-on structure and a plug and slot matching structure.

8. The camera focusing system of the box girder inspection robot according to claim 3, characterized in that: The control system of the inspection robot body is connected to the controller of each fill light to adjust the brightness of each fill light.

9. The camera focusing system of the box girder inspection robot according to claim 3, characterized in that: The fill light is mounted on an electric pan-tilt platform, which is mounted on the annular light stand. The control system of the inspection robot body is connected to the controller of each electric pan-tilt platform to achieve angle adjustment of each fill light.

10. A camera focusing method based on a box girder inspection robot according to any one of claims 1 to 9, characterized in that: include: Distance measurement: The distance measuring sensor transmits a pulse signal to the target, receives the reflected signal, and calculates the distance information based on the time difference of the signal; Pre-focus: The measured distance information is sent to the focus control module, which calculates the pre-focus position based on the distance information and lens parameters, and drives the zoom lens to move to the pre-focus position through the motor; Autofocus Optimization: Enables phase detection based autofocus algorithm, which determines the direction and magnitude of focus error by analyzing the phase detection pixels on the image sensor to optimize the focus position; Enables the contrast-based autofocus algorithm, which continuously adjusts the position of the lens and evaluates the contrast in the image at each position. When the maximum contrast is found, the best focus position is considered to be reached.

Images