Vehicle-mounted camera self-cleaning and image enhancement system and control method under harsh working conditions

Abstract

The application discloses a self-cleaning and image enhancement system and control method for a vehicle-mounted camera under harsh working conditions, and relates to the technical field of unmanned vehicles, comprising an unmanned vehicle working condition analysis module and a camera self-cleaning device. The application can actively analyze the visual picture quality of the camera module of the unmanned vehicle, and automatically start a cleaning program when detecting that the image quality is reduced due to rain, snow, mud, dust and the like. The application ensures that the vehicle-mounted camera can still provide clear visual information under various harsh weather or road conditions, thereby guaranteeing the safety and reliability of the automatic driving or auxiliary driving system. The telescopic camera side cover protects the camera lens area during cleaning, prevents cleaning liquid or pollutants from splashing onto the lens itself, and also provides a space for circulating cleaning liquid and ultrasonic waves, thereby improving the targeting of cleaning. Through the image enhancement algorithm of the GPU chip, the problems of image blurring, color distortion and the like caused by weather, light or slight pollution can be effectively improved.

Description

Technical Field

[0001] This invention relates to the field of unmanned vehicle technology, and in particular to a self-cleaning and image enhancement system and control method for vehicle-mounted cameras under harsh working conditions. Background Technology

[0002] With the rapid development of intelligent driving technology, vehicle cameras, as a core component of the vehicle perception system, directly affect driving safety. However, vehicle cameras face severe challenges in practical applications: environmental factors such as severe weather, dust, mud, and snow can seriously affect the image quality of the cameras. Harsh operating conditions have multifaceted effects on vehicle-mounted cameras, including not only lens obstruction caused by physical contamination but also image quality degradation due to changes in environmental conditions. These factors collectively constitute the severe challenges faced by vehicle-mounted cameras in practical applications.

[0003] Regarding physical contamination, rain, snow, dust, and mud can adhere to the camera's surface, causing blurred or completely obstructed images. These contaminants may originate from dust kicked up by the vehicle itself or from debris splashed by other vehicles. Studies have shown that even tiny particles, if adhering to the camera's surface, can significantly degrade image quality, making images blurry. For example, raindrops in rainy weather can form irregular transparent areas on the lens, distorting light and causing image distortion; snowflakes in snowy weather can accumulate on the lens, partially obscuring the view; and dust and mud can form an opaque coating, which in severe cases can completely blind the camera.

[0004] Therefore, developing a self-cleaning and image enhancement system for vehicle cameras that can maintain effective perception capabilities under harsh working conditions has become a key issue that urgently needs to be addressed in the field of intelligent driving. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a self-cleaning and image enhancement system and control method for vehicle-mounted cameras under harsh working conditions. The technical solution adopted is as follows: A self-cleaning and image enhancement system for vehicle-mounted cameras under harsh operating conditions includes an unmanned vehicle operating condition analysis module and a camera self-cleaning device. The unmanned vehicle operating condition analysis module is communicatively connected to the unmanned vehicle-mounted camera module and determines whether a harsh operating condition has occurred based on the visual image quality analysis of the unmanned vehicle-mounted camera module. The camera self-cleaning device includes a telescopic camera side cover, an ultrasonic-based circulating cleaning fluid cleaning component, and a chip-based cleaning controller. The telescopic camera side cover is installed in a slot around the camera. The circulating cleaning fluid cleaning component's circulating cleaning fluid pipe is connected to the circulating cleaning fluid path inside the telescopic camera side cover. The cleaning controller is communicatively connected to the unmanned vehicle operating condition analysis module. When the unmanned vehicle operating condition analysis module analyzes the visual image quality as greater than or equal to a set scoring threshold, the cleaning controller controls the telescopic camera side cover to be retracted, and the circulating cleaning fluid cleaning component does not work. When the unmanned vehicle operating condition analysis module analyzes the visual image quality as less than the set scoring threshold, the cleaning controller performs a cleaning action. The cleaning action involves controlling the telescopic camera side cover to be extended, controlling the circulating cleaning fluid cleaning component to start circulating cleaning fluid and ultrasonic cleaning, continuing for a set time, and then retracting. This cleaning action is repeated three times according to the set time.

[0006] By adopting the above technical solution, the system can proactively analyze the visual image quality of the unmanned vehicle's camera module. When it detects image quality degradation caused by rain, snow, mud, dust, etc. (below the scoring threshold), it automatically initiates a cleaning process. This ensures that the vehicle's camera can still provide clear visual information under various adverse weather or road conditions, thereby guaranteeing the safety and reliability of autonomous driving or assisted driving systems.

[0007] The retractable camera side cover design can protect the camera lens area during cleaning, preventing cleaning fluid or contaminants from splashing onto the lens itself. It also provides space for circulating cleaning fluid and ultrasonic waves to work, improving the targeted nature of the cleaning process.

[0008] The combination of ultrasonic cleaning and circulating cleaning fluid offers superior cleaning power compared to traditional pure water rinsing or simple wiping. Ultrasonic waves generate minute cavitation effects, effectively removing stubborn dirt, oil films, or ice layers adhering to the lens surface, while the circulating cleaning fluid carries away the removed contaminants, preventing secondary contamination.

[0009] The design of three repeated cleaning cycles avoids the impact of excessively long single cleaning time on visual image capture. Of course, it can also stop the cleaning process within 10-20 seconds of the cleaning action, ensuring that even if the stains are relatively heavy, a good cleaning effect can be achieved through multiple cycles, thus improving the reliability of cleaning.

[0010] Timely and effective cleaning reduces the long-term erosion and wear of dirt on camera lenses and internal components. Especially in harsh environments containing sand, gravel, or chemicals (such as de-icing agents), it can significantly extend the lifespan of camera modules and reduce maintenance costs.

[0011] The system only activates the cleaning device when needed (i.e., when the image quality is below a threshold), otherwise remaining retracted and in standby mode. This avoids unnecessary energy consumption, achieving cleaning on demand—a smart and energy-saving design. The retracted state of the telescopic side shields also helps reduce wind resistance.

[0012] Optionally, an image enhancement module is also included. The image enhancement module includes a cache and a GPU chip. The cache is communicatively connected to the unmanned vehicle camera module and the vehicle control system, respectively. The GPU chip is communicatively connected to the unmanned vehicle condition analysis module and the cache, respectively. If the unmanned vehicle condition analysis module determines that the visual image quality is less than a set scoring threshold, the image enhancement module intervenes. The GPU chip enhances the visual image captured by the unmanned vehicle camera module based on the image enhancement algorithm and stores it in the cache. The vehicle control system uses the enhanced visual image stored in the cache.

[0013] By adopting the above technical solution, when the image quality of the camera deteriorates due to contamination, even if the self-cleaning device is activated, the cleaning may not be complete or may require time to recover. The image enhancement module can process the image before, during, or after the cleaning action, or even as a supplement to the cleaning effect. By executing advanced image enhancement algorithms (such as dehazing, contrast enhancement, edge sharpening, low-light enhancement, etc.) through the GPU chip, problems such as image blurring and color distortion caused by weather (fog, rain, snow), lighting (strong light, weak light), or slight contamination can be effectively improved.

[0014] This allows the vehicle control system to obtain relatively clearer and more informative images even when the cameras are not physically cleaned to their optimal condition, improving the perception accuracy and decision-making reliability of the autonomous driving system in harsh environments.

[0015] It not only restores image quality through physical means (self-cleaning), but also compensates for the deficiencies or delays of physical cleaning through digital processing (image enhancement).

[0016] This "hardware and software combined" strategy greatly improves the vision system's ability to cope with harsh working conditions. Even if there are still a few stains or water droplets on the physical lens that are difficult to completely remove, image enhancement can restore image details to a certain extent, provide usable visual information, and reduce the risk of system performance degradation due to the failure of a single component.

[0017] Optionally, the unmanned vehicle operating condition analysis module includes a memory, a vision analysis chip, and a data analysis chip. The memory is communicatively connected to the unmanned vehicle's onboard camera module, the vision analysis chip is communicatively connected to the memory, and the data analysis chip is communicatively connected to the vision analysis chip.

[0018] Optionally, the telescopic camera side cover includes an electric push rod, a side cover base plate, and a side plate. The electric push rod is installed inside the unmanned vehicle shell, and a side cover passage hole is opened around the unmanned vehicle shell surrounding the unmanned vehicle camera module. One side of the side cover base plate is mounted on the piston rod of the electric push rod, and one end of the side plate is mounted on the other side of the side cover base plate. When the piston rod of the electric push rod extends, the side plate surrounds the unmanned vehicle camera module. When the piston rod of the electric push rod retracts, the side plate retracts, and its end is flush with the unmanned vehicle shell.

[0019] Optionally, the circulating cleaning fluid cleaning assembly includes a circulating cleaning fluid tank, a micro pump, a connecting hose, an ultrasonic generator, and multiple nozzles. The circulating cleaning fluid tank is installed inside the unmanned vehicle shell. The side cover bottom plate and the inside of the side plate are provided with a circulating cleaning fluid channel. The micro pump connects the circulating cleaning fluid channel and the circulating cleaning fluid tank through the connecting hose. The multiple nozzles are respectively installed at multiple spray holes on the inside of the side plate. The ultrasonic generator is installed inside the circulating cleaning fluid channel.

[0020] By adopting the above technical solution, precise control over the extension and retraction of the side cover base plate and side plates, including position and speed, can be achieved using an electric push rod. This allows the side cover base plate and side plates to stably and accurately surround the unmanned vehicle-mounted camera module. The design of the side plates surrounding the camera can form a surrounding jet-type ultrasonic cleaning with the cooperation of multiple nozzles. The ultrasonic waves generated by the ultrasonic generator create a cavitation effect in the cleaning fluid, which can effectively remove stubborn stains (such as oil film, salt crystals, and partially dried mud) attached to the lens surface. This physical cleaning method is more thorough and has higher cleaning efficiency than simple water rinsing.

[0021] The electric push rods and most of the side enclosure structures (base plate, side plates) are designed inside the autonomous vehicle's outer shell, extending only when cleaning is required. This design makes the vehicle's appearance cleaner and more aesthetically pleasing, without adding extra external structures during normal driving, and reduces the risk of damage from impacts with foreign objects.

[0022] A circulation system consisting of a circulating cleaning fluid tank and a micro-pump continuously supplies liquid with a certain pressure and cleaning power. Compared to single spraying, the circulation system ensures multiple, continuous rinsings of the lens within a set time, guaranteeing effective cleaning. The cleaning fluid can be replaced periodically to maintain its cleaning capability.

[0023] By integrating the circulating cleaning fluid channel directly into the bottom plate and inside the side cover, the structural space of the side cover itself is cleverly utilized, making the entire cleaning assembly more compact, reducing additional pipes and connectors, and lowering the risk of leakage and installation complexity.

[0024] Multiple nozzles are distributed on the inside of the side panel, which can rinse the camera lens from different angles to ensure that the entire lens surface is covered and to avoid cleaning dead corners.

[0025] Optionally, the cleaning controller includes an instruction buffer and a control chip. The instruction buffer is communicatively connected to the unmanned vehicle operating condition analysis module, and the control chip is communicatively connected to the instruction buffer, and controls the execution actions of the electric push rod, the micro pump and the ultrasonic generator respectively.

[0026] A method for controlling the self-cleaning and image enhancement of vehicle-mounted cameras under harsh working conditions is proposed. This method employs a self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions to control their self-cleaning and image enhancement processes. The method includes the following steps: Step 1: The unmanned vehicle-mounted camera module continuously collects raw images, and the raw image data is stored in the memory of the working condition analysis module. Step 2: The visual analysis chip performs multi-dimensional analysis to evaluate the visual image quality and outputs a visual image quality score. Step 3: Set a scoring threshold. When the visual image quality score output by the data analysis chip is less than the scoring threshold, the instruction buffer stores the cleaning action instruction. Step 4: The control chip executes the cleaning action command, controls the piston rod of the electric push rod to extend, controls the micro pump to start, and controls the ultrasonic generator to start; the cleaning action command lasts for a set time; the cleaning action command is executed 3 times in a cycle every set time.

[0027] Optionally, when the visual image quality score output by the data analysis chip is less than the score threshold, the GPU chip obtains the original image from the cache and sequentially performs bad pixel repair, adaptive dehazing, HDR synthesis, and dynamic sharpening to obtain an enhanced visual image.

[0028] Optionally, step 2, which assesses the visual image quality, includes the following specific steps: Step 21: Calculate the Laplacian variance of the image, calculate the sharpness index based on the Laplacian variance of the image, and set the weight of the sharpness index. Step 22: Segment the HSV color space, calculate the percentage of smudged pixels to obtain the smudge coverage index, and set the weight of the smudge coverage index. Step 23: Perform optical interference detection to identify halo features and obtain optical interference indexes, and set the weights of the optical interference indexes. Step 24: Multiply each indicator by its corresponding weight and add them together to obtain the visual image quality score.

[0029] Optionally, in step 4, the time can be set to 10 to 20 seconds.

[0030] In summary, the present invention has at least one of the following beneficial technical effects: This invention provides a self-cleaning and image enhancement system and control method for vehicle-mounted cameras under harsh working conditions. The system can actively analyze the visual image quality of unmanned vehicle-mounted camera modules and automatically start the cleaning program when it detects that the image quality has deteriorated due to rain, snow, mud, dust, etc.

[0031] This ensures that the vehicle-mounted camera can still provide clear visual information under various adverse weather or road conditions, thereby guaranteeing the safety and reliability of autonomous driving or driver assistance systems.

[0032] The retractable camera side cover protects the camera lens area during cleaning, preventing cleaning fluid or contaminants from splashing onto the lens itself. It also provides space for circulating cleaning fluid and ultrasonic waves to work, improving the effectiveness of cleaning.

[0033] The combination of ultrasonic cleaning and circulating cleaning fluid offers superior cleaning power compared to traditional pure water rinsing or simple wiping. Ultrasonic waves generate minute cavitation effects, effectively removing stubborn dirt, oil films, or ice layers adhering to the lens surface, while the circulating cleaning fluid carries away the removed contaminants, preventing secondary contamination.

[0034] Timely and effective cleaning reduces the long-term erosion and wear of dirt on camera lenses and internal components, significantly extending the lifespan of camera modules and reducing maintenance costs.

[0035] The system activates the cleaning device only when needed, remaining retracted and in standby mode otherwise. This avoids unnecessary energy consumption, achieving cleaning on demand—a smart and energy-saving design. The retracted state of the telescopic side shields also helps reduce wind resistance.

[0036] Image enhancement algorithms using GPU chips can effectively improve problems such as image blurring and color distortion caused by weather, lighting, or slight pollution.

[0037] Even if there are still a few stains or water droplets on the physical lens that are difficult to completely remove, image enhancement can restore image details to a certain extent, provide usable visual information, and reduce the risk of system performance degradation due to the failure of a single component. Attached Figure Description

[0038] Figure 1 This is a schematic diagram of the electrical component connection principle of the vehicle-mounted camera self-cleaning and image enhancement system under harsh working conditions according to the present invention; Figure 2 This is a schematic diagram of the camera self-cleaning device in the initial state of the vehicle-mounted camera self-cleaning and image enhancement system under harsh working conditions of the present invention. Figure 3 This is a schematic diagram of the camera self-cleaning device in the clean state of the vehicle-mounted camera self-cleaning and image enhancement system under harsh working conditions according to the present invention.

[0039] Figure reference numerals: 1. Unmanned vehicle operating condition analysis module; 11. Memory; 12. Visual analysis chip; 13. Data analysis chip; 2. Unmanned vehicle onboard camera module; 311. Electric push rod; 312. Side cover base plate; 313. Side plate; 321. Circulating cleaning fluid tank; 322. Micro pump; 323. Connecting hose; 324. Ultrasonic generator; 325. Nozzle; 33. Cleaning controller; 34. Command buffer; 35. Control chip; 4. Image enhancement module; 41. Buffer; 42. GPU chip; 100. Vehicle control system. Detailed Implementation

[0040] The present invention will be further described in detail below with reference to the accompanying drawings.

[0041] This invention discloses a self-cleaning and image enhancement system and control method for vehicle-mounted cameras under harsh working conditions.

[0042] Reference Figures 1-3 Example 1: A self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions, including an unmanned vehicle working condition analysis module 1 and a camera self-cleaning device. The unmanned vehicle working condition analysis module 1 is communicatively connected to the unmanned vehicle-mounted camera module 2. Based on the visual image quality analysis of the unmanned vehicle-mounted camera module 2, it determines whether a harsh working condition has occurred. The camera self-cleaning device includes a telescopic camera side cover, an ultrasonic-based circulating cleaning fluid cleaning component, and a chip-based cleaning controller 33. The telescopic camera side cover is installed in a slot around the camera. The circulating cleaning fluid pipe of the circulating cleaning fluid cleaning component is connected to the circulating cleaning fluid path inside the telescopic camera side cover. The cleaning controller 33 is connected to the unmanned vehicle condition analysis module 1. When the unmanned vehicle condition analysis module 1 analyzes the visual image quality as greater than or equal to the set scoring threshold, the cleaning controller 33 controls the telescopic camera side cover to be in a retracted state, and the circulating cleaning fluid cleaning component does not work. When the unmanned vehicle condition analysis module 1 analyzes the visual image quality as less than the set scoring threshold, the cleaning controller 33 executes a cleaning action. The cleaning action involves controlling the telescopic camera side cover to be in an extended state, controlling the circulating cleaning fluid cleaning component to start circulating cleaning fluid and ultrasonic cleaning, continuing for a set time, and then retracting. The cleaning action is repeated three times according to the set time.

[0043] The system can proactively analyze the visual image quality of the unmanned vehicle's onboard camera module 2. When it detects image quality degradation caused by rain, snow, mud, dust, etc. (below the scoring threshold), it automatically initiates a cleaning process. This ensures that the onboard camera can still provide clear visual information under various adverse weather or road conditions, thereby guaranteeing the safety and reliability of autonomous driving or assisted driving systems.

[0044] The retractable camera side cover design can protect the camera lens area during cleaning, preventing cleaning fluid or contaminants from splashing onto the lens itself. It also provides space for circulating cleaning fluid and ultrasonic waves to work, improving the targeted nature of the cleaning process.

[0045] The combination of ultrasonic cleaning and circulating cleaning fluid offers superior cleaning power compared to traditional pure water rinsing or simple wiping. Ultrasonic waves generate minute cavitation effects, effectively removing stubborn dirt, oil films, or ice layers adhering to the lens surface, while the circulating cleaning fluid carries away the removed contaminants, preventing secondary contamination.

[0046] The design of three repeated cleaning cycles avoids the impact of excessively long single cleaning time on visual image capture. Of course, it can also stop the cleaning process within 10-20 seconds of the cleaning action, ensuring that even if the stains are relatively heavy, a good cleaning effect can be achieved through multiple cycles, thus improving the reliability of cleaning.

[0047] Timely and effective cleaning reduces the long-term erosion and wear of dirt on camera lenses and internal components. Especially in harsh environments containing sand, gravel, or chemicals (such as de-icing agents), it can significantly extend the lifespan of camera modules and reduce maintenance costs.

[0048] The system only activates the cleaning device when needed (i.e., when the image quality is below a threshold), otherwise remaining retracted and in standby mode. This avoids unnecessary energy consumption, achieving cleaning on demand—a smart and energy-saving design. The retracted state of the telescopic side shields also helps reduce wind resistance.

[0049] Example 2 also includes an image enhancement module 4, which includes a cache 41 and a GPU chip 42. The cache 41 is communicatively connected to the unmanned vehicle camera module 2 and the vehicle control system 100, respectively. The GPU chip 42 is communicatively connected to the unmanned vehicle condition analysis module 1 and the cache 41, respectively. If the unmanned vehicle condition analysis module 1 determines that the visual image quality is less than a set scoring threshold, the image enhancement module 4 intervenes. The GPU chip 42 enhances the visual image acquired by the unmanned vehicle camera module 2 based on the image enhancement algorithm and stores it in the cache 41. The vehicle control system 100 uses the enhanced visual image stored in the cache 41.

[0050] When camera image quality deteriorates due to contamination, even with a self-cleaning device activated, incomplete cleaning or time-consuming recovery may occur. The image enhancement module can process the image before, during, or after the cleaning process, or even supplement the cleaning effect. By executing advanced image enhancement algorithms (such as dehazing, contrast enhancement, edge sharpening, and low-light enhancement) through the GPU chip 42, it can effectively improve image blurring and color distortion caused by weather (fog, rain, snow), lighting (strong light, weak light), or minor contamination.

[0051] This allows the vehicle control system to obtain relatively clearer and more informative images even when the cameras are not physically cleaned to their optimal condition, improving the perception accuracy and decision-making reliability of the autonomous driving system in harsh environments.

[0052] It not only restores image quality through physical means (self-cleaning), but also compensates for the deficiencies or delays of physical cleaning through digital processing (image enhancement).

[0053] This "hardware and software combined" strategy greatly improves the vision system's ability to cope with harsh working conditions. Even if there are still a few stains or water droplets on the physical lens that are difficult to completely remove, image enhancement can restore image details to a certain extent, provide usable visual information, and reduce the risk of system performance degradation due to the failure of a single component.

[0054] Example 3: The unmanned vehicle condition analysis module 1 includes a memory 11, a vision analysis chip 12, and a data analysis chip 13. The memory 11 is communicatively connected to the unmanned vehicle camera module 2, the vision analysis chip 12 is communicatively connected to the memory 11, and the data analysis chip 13 is communicatively connected to the vision analysis chip 12.

[0055] Example 4: The telescopic camera side cover includes an electric push rod 311, a side cover base plate 312, and a side plate 313. The electric push rod 311 is installed inside the unmanned vehicle shell, and a side cover passage hole is opened around the unmanned vehicle shell surrounding the unmanned vehicle camera module 2. One side of the side cover base plate 312 is installed on the piston rod of the electric push rod 311, and one end of the side plate 313 is installed on the other side of the side cover base plate 312. When the piston rod of the electric push rod 311 extends, the side plate 313 surrounds the unmanned vehicle camera module 2. When the piston rod of the electric push rod 311 retracts, the side plate 313 retracts, and its end is flush with the unmanned vehicle shell.

[0056] Example 5: The circulating cleaning fluid cleaning assembly includes a circulating cleaning fluid tank 321, a micro pump 322, a connecting hose 323, an ultrasonic generator 324, and multiple nozzles 325. The circulating cleaning fluid tank 321 is installed inside the unmanned vehicle shell. The side cover bottom plate 312 and the side plate 313 are provided with a circulating cleaning fluid channel. The micro pump 322 connects the circulating cleaning fluid channel and the circulating cleaning fluid tank 321 through the connecting hose 323. Multiple nozzles 325 are respectively installed at multiple spray holes on the inner side of the side plate 313. The ultrasonic generator 324 is installed in the circulating cleaning fluid channel.

[0057] The electric push rod 311 enables precise control over the extension and retraction of the side cover base plate 312 and side plate 313, including position and speed. This allows the side cover base plate 312 and side plate 313 to stably and accurately surround the unmanned vehicle camera module 2. The design of the side plate 313 surrounding the camera can form a surrounding jet-type ultrasonic cleaning with the cooperation of multiple nozzles 325. The ultrasonic waves generated by the ultrasonic generator 324 create a cavitation effect in the cleaning fluid, which can effectively remove stubborn stains (such as oil film, salt crystals, and partially dried mud) attached to the lens surface. This physical cleaning method is more thorough and efficient than simple water rinsing.

[0058] The electric push rod 311 and most of the side cover structure (base plate, side plates) are designed inside the autonomous vehicle's outer shell, extending only when cleaning is required. This design makes the vehicle's appearance cleaner and more aesthetically pleasing, without adding extra external structures during normal driving, and reducing the risk of damage from impacts with foreign objects.

[0059] A circulation system consisting of a circulating cleaning fluid tank 321 and a micro pump 322 continuously supplies liquid with a certain pressure and cleaning capacity. Compared to single spraying, the circulation system ensures multiple, continuous rinsings of the lens within a set time, guaranteeing cleaning effectiveness. The cleaning fluid can be replaced periodically to maintain its cleaning power.

[0060] By directly integrating the circulating cleaning fluid channel inside the side cover base plate 312 and side plate 313, the structural space of the side cover itself is cleverly utilized, making the entire cleaning assembly more compact, reducing additional pipes and connectors, and lowering the risk of leakage and installation complexity.

[0061] Multiple nozzles 325 are distributed on the inside of the side panel, which can rinse the camera lens from different angles to ensure that the entire lens surface is covered and to avoid cleaning dead corners.

[0062] Example 6: The cleaning controller 33 includes an instruction buffer 34 and a control chip 35. The instruction buffer 34 is communicatively connected to the unmanned vehicle condition analysis module 1, and the control chip 35 is communicatively connected to the instruction buffer 34, and controls the execution actions of the electric push rod 311, the micro pump 322 and the ultrasonic generator 324 respectively.

[0063] Example 7: A method for controlling the self-cleaning and image enhancement of an onboard camera under harsh working conditions. This method employs an onboard camera self-cleaning and image enhancement system for harsh working conditions to control the self-cleaning and image enhancement of the onboard camera under harsh working conditions, including the following steps: Step 1: The unmanned vehicle-mounted camera module 2 continuously collects raw images, and the raw image data is stored in the memory 11 of the working condition analysis module 1. Step 2: The visual analysis chip 12 performs multi-dimensional analysis to evaluate the visual image quality and outputs a visual image quality score. Step 3: Set a scoring threshold. When the visual image quality score output by the data analysis chip 13 is less than the scoring threshold, the instruction buffer 34 stores the cleaning action instruction. Step 4: The control chip 35 executes the cleaning action command, controls the piston rod of the electric push rod 311 to extend, controls the micro pump 322 to start, and controls the ultrasonic generator 324 to start; the cleaning action command continues for a set time; the cleaning action command is executed 3 times in a cycle every set time.

[0064] In Example 8, when the visual image quality score output by the data analysis chip 13 is less than the score threshold, the GPU chip 42 obtains the original image from the cache 41 and sequentially performs bad pixel repair, adaptive dehazing, HDR synthesis, and dynamic sharpening to obtain the enhanced visual image.

[0065] Example 9, step 2, includes the following specific steps for evaluating the visual image quality: Step 21: Calculate the Laplacian variance of the image, calculate the sharpness index based on the Laplacian variance of the image, and set the weight of the sharpness index. Step 22: Segment the HSV color space, calculate the percentage of smudged pixels to obtain the smudge coverage index, and set the weight of the smudge coverage index. Step 23: Perform optical interference detection to identify halo features and obtain optical interference indexes, and set the weights of the optical interference indexes. Step 24: Multiply each indicator by its corresponding weight and add them together to obtain the visual image quality score.

[0066] In Example 10, step 4, the set time is 10 to 20 seconds.

[0067] The implementation principle of the present invention is illustrated below through specific embodiments: A driverless taxi is driving on a muddy road after rain. Its forward-facing onboard camera module 2 (for example, for lane line recognition and obstacle detection) begins to capture a live video stream. The rain and splashes of water on the road have caused a thin layer of mud and water mixture to adhere to the lens. Although it does not completely obstruct the view, the image quality has already noticeably decreased.

[0068] The camera module 2 continuously transmits the acquired raw image data to the unmanned vehicle condition analysis module 1 and stores it in its memory 11.

[0069] The visual analysis chip 12 analyzes the latest image in the memory 11. It quickly calculates the image sharpness (judging the degree of edge blurring by indicators such as Laplacian variance), detects a large cluster of blue / green pixels (judged as water / mud stains in the HSV color space), and identifies a slight halo effect (optical interference) at the image edges. These analysis results are quantified into a score.

[0070] The data analysis chip 13 receives various index scores output by the visual analysis chip 12, and performs a weighted summation according to preset weights (e.g., clarity has a higher weight, stain coverage has a second higher weight, and optical interference has the lowest weight) to obtain a comprehensive visual image quality score. This score (e.g., a score of 45 points) is lower than a preset scoring threshold (e.g., a threshold of 60 points).

[0071] The data analysis chip 13 sends a signal below a threshold to the cleaning controller 33. The instruction buffer 34 in the cleaning controller 33 receives the instruction and stores the instruction sequence of "execute three cleaning actions".

[0072] Self-cleaning process: First cleaning: The control chip 35 reads instructions from the instruction buffer 34 and first controls the electric push rod 311 to move. Its piston rod extends, causing the side cover bottom plate 312 and the side plate 313 to move outward from the slot inside the unmanned vehicle shell until the side plate 313 completely surrounds the camera module 2, with its end flush with the vehicle shell. At this time, the nozzle 325 installed on the inside of the side plate is also located around the lens.

[0073] The control chip 35 then instructs the micro pump 322 to start, drawing cleaning fluid (possibly an aqueous solution containing a small amount of surfactant) from the circulating cleaning fluid tank 321, and delivering the cleaning fluid to each nozzle 325 through the connecting hose 323 and the circulating cleaning fluid channel inside the side cover.

[0074] At the same time, the control chip 35 instructs the ultrasonic generator 324 to start, generating ultrasonic waves in the circulating cleaning fluid channel. The cleaning fluid, carrying the energy of the ultrasonic waves, is evenly sprayed from the nozzle 325 onto the lens surface.

[0075] This combination of "extend + spray + ultrasonic" action lasts for a set time, such as 15 seconds. The cavitation effect of the ultrasonic waves helps to remove some of the mud-water mixture, while the circulating cleaning fluid washes it away and flows back to the cleaning fluid tank 321 (or is filtered and recirculated) through the drain hole (or return channel) of the side cover bottom plate 312.

[0076] Fifteen seconds later, control chip 35 instructs electric actuator 311 to retract, the side cover retracts, and the cleaned lens is exposed. The micro pump and ultrasonic generator also stop working.

[0077] Second cleaning: The system pauses briefly (e.g., a few seconds), then repeats the entire cleaning process of extending, spraying, ultrasonically agitating, and retracting, lasting another 15 seconds. This cleaning further removes any remaining stains.

[0078] Third cleaning: Similarly, the system waits a few seconds before performing a third, complete 15-second cleaning cycle to ensure the lens is in optimal clean condition.

[0079] Cleaning and image enhancement (optional): After three cleaning cycles, the side cover retracts, and camera module 2 resumes image acquisition. The condition analysis module 1 reassesses the image quality; assuming the score has now increased to 75 points, exceeding the threshold of 60 points, the system determines the cleaning was successful.

[0080] During this period, if the image enhancement module 4 is configured to intervene immediately during or after cleaning, the GPU chip 42 will start working when the data analysis chip 13 first emits a signal below a threshold. It retrieves the original blurred image acquired before or during the cleaning process from the buffer 41 and quickly executes a series of image enhancement algorithms: first, it repairs any bad pixels in the image; then, it performs adaptive dehazing to restore clarity in distant areas; next, it uses HDR (High Dynamic Range) synthesis technology to balance details in bright and dark areas; and finally, it performs dynamic sharpening to enhance edges and textures. The processed enhanced image is then stored back in the buffer 41.

[0081] The vehicle control system 100 prioritizes reading from the buffer 41 and using these enhanced images for subsequent perception, decision-making and control. Even if there are still a few water stains on the lens that are difficult to completely remove, the enhanced images provide clearer and more useful information.

[0082] As the vehicle continues driving, if the weather improves or the road surface dries, the operating condition analysis module 1 continuously monitors the image quality. When the score stabilizes above the threshold again, the system will keep the side shields retracted and the cleaning components in standby mode until the next time cleaning is detected. Throughout the process, the system achieves on-demand, efficient, and intelligent self-cleaning and image optimization.

[0083] The above are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, all equivalent changes made in accordance with the structure, shape and principle of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions, characterized in that: The system includes an unmanned vehicle operating condition analysis module (1) and a camera self-cleaning device. The unmanned vehicle operating condition analysis module (1) is communicatively connected to the unmanned vehicle camera module (2). Based on the visual image quality analysis of the unmanned vehicle camera module (2), it judges whether a severe operating condition has occurred. The camera self-cleaning device includes a telescopic camera side cover, an ultrasonic-based circulating cleaning fluid cleaning component, and a chip-based cleaning controller (33). The telescopic camera side cover is installed in a slot set around the camera. The circulating cleaning fluid cleaning component's circulating cleaning fluid pipe is connected to the circulating cleaning fluid path inside the telescopic camera side cover. The cleaning controller (33) 33) It communicates with the unmanned vehicle condition analysis module (1). When the unmanned vehicle condition analysis module (1) analyzes the visual image quality to be greater than or equal to the set scoring threshold, the cleaning controller (33) controls the telescopic camera side cover to be in the retracted state, and the circulating cleaning fluid cleaning component does not work. When the unmanned vehicle condition analysis module (1) analyzes the visual image quality to be less than the set scoring threshold, the cleaning controller (33) performs the cleaning action. The cleaning action is to control the telescopic camera side cover to be in the extended state, control the circulating cleaning fluid cleaning component to start the circulating cleaning fluid and ultrasonic cleaning, and retract it after a set time. The cleaning action is repeated three times according to the set time. The specific steps involved in evaluating visual image quality are as follows: Step 21: Calculate the Laplacian variance of the image, calculate the sharpness index based on the Laplacian variance of the image, and set the weight of the sharpness index. Step 22: Segment the HSV color space, calculate the percentage of smudged pixels to obtain the smudge coverage index, and set the weight of the smudge coverage index. Step 23: Perform optical interference detection to identify halo features and obtain optical interference indexes, and set the weights of the optical interference indexes. Step 24: Multiply each indicator by its corresponding weight and add them together to obtain the visual image quality score.

2. The self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions according to claim 1, characterized in that: It also includes an image enhancement module (4), which includes a buffer (41) and a GPU chip (42). The buffer (41) is connected to the unmanned vehicle camera module (2) and the vehicle control system (100) respectively. The GPU chip (42) is connected to the unmanned vehicle condition analysis module (1) and the buffer (41) respectively. If the unmanned vehicle condition analysis module (1) determines that the visual image quality is less than the set scoring threshold, the image enhancement module (4) intervenes. The GPU chip (42) enhances the visual image collected by the unmanned vehicle camera module (2) based on the image enhancement algorithm and stores it in the buffer (41). The vehicle control system (100) uses the enhanced visual image stored in the buffer (41).

3. The self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions according to claim 2, characterized in that: The unmanned vehicle operating condition analysis module (1) includes a memory (11), a vision analysis chip (12) and a data analysis chip (13). The memory (11) is communicatively connected to the unmanned vehicle camera module (2), the vision analysis chip (12) is communicatively connected to the memory (11), and the data analysis chip (13) is communicatively connected to the vision analysis chip (12).

4. The self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions according to claim 3, characterized in that: The telescopic camera side cover includes an electric push rod (311), a side cover base plate (312), and a side plate (313). The electric push rod (311) is installed inside the unmanned vehicle shell, and a side cover through hole is opened around the unmanned vehicle shell surrounding the unmanned vehicle camera module (2). One side of the side cover base plate (312) is installed on the piston rod of the electric push rod (311), and one end of the side plate (313) is installed on the other side of the side cover base plate (312). When the piston rod of the electric push rod (311) extends, the side plate (313) surrounds the unmanned vehicle camera module (2). When the piston rod of the electric push rod (311) retracts, the side plate (313) retracts, and its end is flush with the unmanned vehicle shell.

5. The self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions according to claim 4, characterized in that: The circulating cleaning fluid cleaning assembly includes a circulating cleaning fluid tank (321), a micro pump (322), a connecting hose (323), an ultrasonic generator (324), and multiple nozzles (325). The circulating cleaning fluid tank (321) is installed inside the unmanned vehicle shell. The side cover bottom plate (312) and the side plate (313) are provided with a circulating cleaning fluid channel. The micro pump (322) connects the circulating cleaning fluid channel and the circulating cleaning fluid tank (321) through the connecting hose (323). The multiple nozzles (325) are respectively installed at multiple spray holes on the inside of the side plate (313). The ultrasonic generator (324) is installed inside the circulating cleaning fluid channel.

6. The self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions according to claim 5, characterized in that: The cleaning controller (33) includes an instruction buffer (34) and a control chip (35). The instruction buffer (34) is connected to the unmanned vehicle condition analysis module (1) and the control chip (35) is connected to the instruction buffer (34) and controls the execution actions of the electric push rod (311), the micro pump (322) and the ultrasonic generator (324) respectively.

7. A method for self-cleaning and image enhancement control of vehicle-mounted cameras under harsh working conditions, characterized in that: The self-cleaning and image enhancement system for vehicle-mounted cameras under harsh working conditions, as described in claim 6, controls the self-cleaning and image enhancement of vehicle-mounted cameras under harsh working conditions, comprising the following steps: Step 1: The unmanned vehicle camera module (2) continuously collects raw images, and the raw image data is stored in the memory (11) of the working condition analysis module (1). Step 2: The visual analysis chip (12) performs multi-dimensional analysis to evaluate the visual image quality and outputs a visual image quality score. Step 3: Set a scoring threshold. When the visual image quality score output by the data analysis chip (13) is less than the scoring threshold, the instruction buffer (34) stores the cleaning action instruction. Step 4: The control chip (35) executes the cleaning action command, controls the piston rod of the electric push rod (311) to extend, controls the micro pump (322) to start, and controls the ultrasonic generator (324) to start; the cleaning action command continues for a set time; the cleaning action command is executed (3) times in a cycle every set time.

8. The self-cleaning and image enhancement control method for vehicle-mounted cameras under harsh working conditions according to claim 7, characterized in that: When the visual image quality score output by the data analysis chip (13) is less than the score threshold, the GPU chip (42) obtains the original image from the cache (41) and sequentially performs bad pixel repair, adaptive dehazing, HDR synthesis, and dynamic sharpening to obtain the enhanced visual image.

9. The self-cleaning and image enhancement control method for vehicle-mounted cameras under harsh working conditions according to claim 8, characterized in that: In step 4, the time is set to 10 to 20 seconds.

Images