A laser deicing method based on image recognition

Abstract

The present invention relates to the technical field of image analysis, and in particular to a laser deicing method based on image recognition, comprising: driving a deicing device to a tunnel entrance and exit, detecting a posture of the deicing device, adjusting the posture of the deicing device based on a posture detection result of the deicing device, collecting arch structure image data of the tunnel entrance and exit through the deicing device; capturing icicle images in the arch structure image data of the tunnel entrance and exit, locating corner points in the icicle images, and creating a laser deicing path based on the corner points in the icicle images; the present invention takes snowfall scenes into consideration and can be applied to snowfall and non-snowfall scenes, thereby improving the robustness of the application scenarios of the method; in the deicing stage, the icicle cleaning priority and the state of vehicles traveling in the scene are comprehensively considered, and the speed is matched for the icicle cleaning, thereby realizing unmanned, intelligent and safe autonomous icicle cleaning, and there is no need to prohibit the tunnel from driving during the cleaning process, and the vehicle-free gaps in the tunnel can be effectively used to achieve detailed cleaning.

Description

Technical Field

[0001] The present invention relates to the technical field of image analysis, and in particular to a laser deicing method based on image recognition. Background Art

[0002] In cold seasons, the ambient temperature of the tunnel is low, and some tunnels are located in special terrains such as mountainous areas, with good ventilation conditions, resulting in rapid heat loss, causing icicles and cones to condense at the top arch of the tunnel entrance and exit.

[0003] A Chinese invention patent with application number 201810624553.2 discloses a laser deicing method based on path planning, which includes the following steps: Step 1, select a laser according to the working distance; Step 2, determine the deicing mode according to the icing conditions, and there are four modes in total: Mode 1: When there is ice in the lower layer of the ice, turn the laser to aim at the center of the ice and cable node, and use the thermal stress effect to de-ice; Mode 2: When there is no ice in the lower layer of the ice and the ice thickness is less than 1.5 cm, turn the laser to perform repeated translational scanning along the axis of the cable, and use the thermal melting effect to de-ice; Mode 3: When there is no ice in the lower layer of the ice and When the ice thickness is between 1.5cm and 3cm, the laser is turned in turn to perform repeated translational scanning along the axis of the upper / lower ice layer, and the thermal melting effect is used to remove the ice; Mode 4: When there is no ice in the lower layer of the ice and the ice thickness is greater than 3cm, the laser is first turned to perform repeated translational scanning along the axis of the cable, and the thermal melting effect is used to melt the ice. Then, the laser is turned to perform wavy line scanning along the axis of the cable, and the thermal stress is used to remove the lower layer of ice. The above thermal melting and thermal stress deicing process is repeated until the lower layer of ice falls off; finally, repeated translational scanning is performed along the axis of the upper layer of ice, and the thermal melting effect is used to remove the upper layer of ice.

[0004] The application aims to solve the problem that "the laser deicing method mainly performs repeated translational scanning movements by adjusting the laser power, the scanning path and mode are relatively simple, it lacks applicability, and the deicing operation progresses slowly and is inefficient."

[0005] However, in the icicle clearing scenario at the top arch of the tunnel entrance and exit, the existing technology can only perform simple icicle clearing and cannot take into account the vehicles driving in and out of the tunnel in real time. Therefore, in this icicle clearing scenario, road management measures are often required. When the icicles are being cleared, vehicles are prohibited from passing. On the one hand, this method consumes a certain amount of human resources, and on the other hand, it undoubtedly creates certain safety risks compared to tunnel traffic.

[0006] To this end, we propose a laser deicing method based on image recognition. Summary of the invention

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides a laser deicing method based on image recognition, which solves the technical problems raised in the above-mentioned background technology.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0009] A laser deicing method based on image recognition, comprising:

[0010] Drive the de-icing equipment to the tunnel entrance and exit, detect the posture of the de-icing equipment, adjust the posture of the de-icing equipment based on the posture detection result of the de-icing equipment, and collect the image data of the arch structure at the tunnel entrance and exit through the de-icing equipment;

[0011] Capture the icicle image in the arch structure image data at the tunnel entrance and exit, locate the corner points in the icicle image, and create a laser deicing path based on the corner points in the icicle image;

[0012] Traversing the corresponding laser deicing paths of each icicle image, analyzing the urgency of the icicles to which the corresponding icicle images belong based on each laser deicing path, setting the laser deicing path processing priority based on the urgency analysis result, and applying the laser deicing paths in sequence based on the laser deicing path processing priority setting result;

[0013] Real-time measurement of the speed and distance of vehicles entering and exiting the tunnel, and decision-making on the laser deicing path based on the collected results;

[0014] The decision logic of the laser deicing path travel speed is expressed as:

[0015] ;

[0016] Where: The time required for a vehicle to enter or exit a tunnel; is the distance between the vehicle and the tunnel entrance and exit; is the vehicle speed; for compensation;

[0017] compensate Customized by the user, compensation >0, compensation The initial default value is 1 second. Calculate and further calculate the corresponding The difference between the two values ​​is selected, and the corresponding source vehicle of the maximum difference calculation result is selected. , recorded as , then the available time domain of the speed path in the laser deicing path is ;

[0018] Calculate the time required for the ice column to fall to the ground based on the speed path in the ice column image, and update the available time domain of the speed path in the laser deicing path based on the calculation result;

[0019] ;

[0020] Where: The time required for the cut-off portion of the icicle to fall to the ground; The distance between the cut-off part of the icicle and the road surface; is the acceleration due to gravity;

[0021] The speed path in the laser deicing path can be updated in the time domain as follows: ;

[0022] The laser deicing path travel speed is:

[0023] ;

[0024] Where: is the moving speed of the pace path; is the pace path length.

[0025] Furthermore, the deicing equipment is integrated by a laser device, a speed and distance measuring camera, a camera and a drone, the drone has a built-in flight path, the drone carries the laser device, the speed and distance measuring camera, and the camera to arrive at the tunnel entrance and exit based on the flight path, and the laser device is any one of a carbon dioxide laser and a fiber laser;

[0026] The drone is equipped with a gyroscope. When the drone carries a laser device, a speed and distance measuring camera, and a camera and arrives at the tunnel entrance and exit based on the flight path, that is, the end of the flight path, the drone senses the attitude of the drone based on the gyroscope. When the drone senses that the drone is tilted, the drone automatically adjusts its attitude based on the gyroscope perception result until it is level.

[0027] The image data of the arched structure at the tunnel entrance and exit is collected by a camera mounted on the de-icing equipment. After the image data of the arched structure at the tunnel entrance and exit is collected, it is synchronously identified whether there is an occluded area in the image data. When the identification result is yes, the image data is completed.

[0028] Furthermore, the step of completing the image data of the arch structure at the tunnel entrance and exit includes:

[0029] S1: Acquire image data of the arch structure at the tunnel entrance and exit, convert the image data into a grayscale image, set the occlusion area capture logic, and capture the occlusion area in the image data based on the occlusion area capture logic;

[0030] S2: If the capture result is empty, the image data completion processing operation ends, otherwise, the occluded area is completed;

[0031] The logic of occlusion area completion processing is expressed as:

[0032] ;

[0033] Where: Iterate the target grayscale value for the pixels in the occluded area; is the total number of neighboring pixels of the pixel in the occluded area; is the gray value of the i-th adjacent pixel; is the judgment function;

[0034] The iterative target grayscale value of each pixel in the occluded area is obtained based on the above formula, and the grayscale value of the pixel in the corresponding occluded area is adjusted based on the obtained iterative target grayscale value to complete the completion processing of the occluded area;

[0035] Decision function The value is 0 or the original value of the parameter in the brackets. When the pixel pointed to by the parameter in the brackets and the target pixel of the overall calculation formula are all pixels in the occluded area, =0, otherwise, the judgment function Take the original value of the parameter in brackets, is the judgment function The number of times the value is 0, when the occluded area is completed based on the above formula, the pixels in the occluded area are obtained Afterwards, based on After completing the pixel, the completion operation of the next pixel in the occluded area is performed. When the completion operation of the occluded area is performed in the occluded area, the completion direction is from the periphery to the center.

[0036] Furthermore, the logic for capturing the blocked area in the image data of the arch structure at the tunnel entrance and exit is:

[0037] Set the grayscale value interval of the occluded area, apply the grayscale value interval of the occluded area to compare with each pixel in the image data, and capture the pixels that meet the grayscale value interval of the occluded area;

[0038] Identify whether the adjacent pixel of each captured pixel is another captured pixel. When the identification result is yes, combine the two adjacent pixels and record them as a pixel block. Further identify whether the adjacent pixel of the pixel block is another captured pixel. When the identification result is yes, combine the pixel block with the pixel and record them as a pixel block. Repeat this process. When the identification result is no each time, record the identification target as an occluded area.

[0039] Each occlusion region includes at least one pixel block, and the occlusion region is generated in the image data when collecting image data of the arch structure at the entrance and exit of the tunnel in a snowfall scene.

[0040] Furthermore, during the stage of collecting image data of the arch structure at the tunnel entrance and exit and the initial stage of the laser deicing path application, the deicing equipment continues to maintain its original position and posture, and the capture logic of the icicle image is:

[0041] Obtain the image data of the arch structure at the entrance and exit of the tunnel, segment the image data, and define the segmentation accuracy by the user. The length and width of each sub-image obtained by segmentation are equal, and the size of each sub-image obtained by segmentation is the same. Calculate the correlation, energy, and entropy of each sub-image.

[0042] ;

[0043] Where: , Determine conditions for icicles; is the gray value of the sub-image; is the sub-image contrast; The gray value determination threshold for the icicle image; is the icicle image contrast determination threshold;

[0044] in, , When the sum is 2, the sub-image is determined to be an icicle image, otherwise it is determined to be a non-icicle image. , Customized by the user end, when the number of sub-images judged as icicle images in the image data does not exceed 1 or there are no adjacent sub-images, the image data segmentation accuracy is adjusted to increase until the number of sub-images judged as icicle images exceeds 1 and there are adjacent sub-images, then the process ends. The combination of adjacent sub-images is the icicle image.

[0045] Furthermore, the corner points in the icicle image are the center points of the sub-images constituting the icicle image. After the corner points in the icicle image are located, the position information of the deicing equipment itself and the position information of the tunnel entrance and exit are further obtained. A straight path is determined based on the connection between the two sets of position information. A plane is created at the points on the path representing the position information of the tunnel entrance and exit, so that the path is perpendicular to the plane. The position information of the corner points located in the area where the icicle image is located on the plane is determined with reference to the position information of the deicing equipment itself.

[0046] After the position information of the positioning corner points is determined, the positioning corner points are adjacently connected to each other based on the position information of each positioning corner point to create a laser deicing path.

[0047] Furthermore, the urgency analysis logic of the icicle to which the icicle image belongs is expressed as:

[0048] ;

[0049] Where: is the urgency of the icicle to which the icicle image belongs; is the bounding box area of ​​the icicle image corresponding to the laser deicing path; is the maximum width that can be taken in the bounding box of the icicle image corresponding to the laser deicing path; is the normalization factor;

[0050] Among them, the normalization factor >0, used for control The value of is too large or too small, and the urgency of the icicles to which each icicle image belongs is calculated based on the above formula. ,by The icicles with large values ​​in the icicle images correspond to the laser deicing paths, which are the priority application targets.

[0051] Furthermore, the speed and distance of the vehicles entering and exiting the tunnel are measured based on the speed and distance measuring camera on the deicing equipment. The laser deicing path is in a Z-shape. In the decision-making stage of the laser deicing path travel pace, the path points on both sides of the laser deicing path are captured, and the section between two adjacent path points is used as the pace path, that is, the number of pace paths is the number of laser deicing path travel paces;

[0052] Among them, the laser deicing path is applied to the flight of the UAV in the deicing equipment. After the laser deicing path is created, the UAV flies based on the laser deicing path, so that the end of the laser equipment carried by the UAV is opposite to the lowest end point of the laser deicing path, and enters the waiting stage of flying based on the laser deicing path.

[0053] Furthermore, the moving speed of the pace path is >0, the pace path moving speed is applied, otherwise, it jumps to the stage of measuring the speed and distance of vehicles entering and exiting the tunnel;

[0054] Among them, the moving speed of the pace path When applied, drones in de-icing equipment wait Post-application pace path moving speed Move from the starting point of the pace path to the end of the pace path, and calculate the new pace path moving speed The process continues like this until the end of the speed setting path coincides with the end of the laser deicing path.

[0055] Furthermore, when the icicle image in the tunnel entrance and exit arch structure image data is not unique, after each icicle image corresponding to the icicle completes the de-icing operation, the de-icing equipment returns to the position and posture of the tunnel entrance and exit arch structure image data collection stage.

[0056] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0057] The present invention provides a laser deicing method based on image recognition. During the execution of the method, the snowfall scene is taken into consideration and the method can be applicable to both snowfall and non-snowfall scenes, thereby improving the robustness of the application scenarios of the method. In the deicing stage, the icicle cleaning priority and the state of the vehicles in the scene are comprehensively considered to pace the icicle cleaning, thereby realizing unmanned, intelligent and safe autonomous icicle cleaning. During the cleaning process, there is no need to prohibit tunnels from driving, and the vehicle-free gaps in the tunnel can be effectively used to achieve detailed cleaning. Moreover, since the cleaning of icicles is distributed, further cleaning of the icicles is further avoided, thereby ensuring that the icicle cleaning work in the icicle cleaning scene at the top arch position of the tunnel entrance and exit is carried out stably and safely. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work:

[0059] Figure 1 A schematic diagram of a laser deicing method based on image recognition;

[0060] Figure 2 This is a logical schematic diagram for creating the icicle laser deicing path in the present invention. DETAILED DESCRIPTION

[0061] 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.

[0062] The present invention will be further described below in conjunction with the embodiments.

[0063] Embodiment 1:

[0064] A laser deicing method based on image recognition in this embodiment, such as Figure 1 As shown, including:

[0065] Drive the de-icing equipment to the tunnel entrance and exit, detect the posture of the de-icing equipment, adjust the posture of the de-icing equipment based on the posture detection result of the de-icing equipment, and collect the image data of the arch structure at the tunnel entrance and exit through the de-icing equipment;

[0066] Capture the icicle image in the arch structure image data at the tunnel entrance and exit, locate the corner points in the icicle image, and create a laser deicing path based on the corner points in the icicle image;

[0067] During the data collection phase of the arch structure image at the tunnel entrance and exit and the initial stage of the laser deicing path application, the deicing equipment continues to maintain its original position and posture, and the capture logic of the icicle image is as follows:

[0068] Obtain the image data of the arch structure at the entrance and exit of the tunnel, segment the image data, and define the segmentation accuracy by the user. The length and width of each sub-image obtained by segmentation are equal, and the size of each sub-image obtained by segmentation is the same. Calculate the correlation, energy, and entropy of each sub-image.

[0069] ;

[0070] Where: , Determine conditions for icicles; is the gray value of the sub-image; is the sub-image contrast; The gray value determination threshold for the icicle image; is the contrast determination threshold of the icicle image;

[0071] in, , When the sum is 2, the sub-image is determined to be an icicle image, otherwise it is determined to be a non-icicle image. , Customized by the user end, when the number of sub-images determined as icicle images in the image data does not exceed 1 or there are no adjacent sub-images, the image data segmentation accuracy is adjusted to increase until the number of sub-images determined as icicle images exceeds 1 and there are adjacent sub-images, then the segmentation ends, and the combination of adjacent sub-images is the icicle image;

[0072] Through the above logic formula, it is possible to determine whether each sub-image obtained by segmentation is a designated part of the icicle image, thereby providing necessary execution parameter support for further execution of the method in this embodiment.

[0073] The corner points in the icicle image are the center points of the sub-images that make up the icicle image. After the corner points in the icicle image are located, the position information of the deicing equipment itself and the position information of the tunnel entrance and exit are further obtained. A straight path is determined based on the connection between the two sets of position information. A plane is created at the points on the path that represent the position information of the tunnel entrance and exit, so that the path is perpendicular to the plane. The position information of the corner points located in the area where the icicle image is located on the plane is determined with reference to the position information of the deicing equipment itself.

[0074] After the position information of the positioning corner points is determined, the positioning corner points are adjacently connected to each other based on the position information of each positioning corner point to create a laser deicing path;

[0075] Traversing the corresponding laser deicing paths of each icicle image, analyzing the urgency of the icicles to which the corresponding icicle images belong based on each laser deicing path, setting the laser deicing path processing priority based on the urgency analysis result, and applying the laser deicing paths in sequence based on the laser deicing path processing priority setting result;

[0076] Real-time measurement of the speed and distance of vehicles entering and exiting the tunnel, and decision-making on the laser deicing path based on the collected results;

[0077] The logic of the urgency analysis of the icicle to which the icicle image belongs is expressed as:

[0078] ;

[0079] Where: is the urgency of the icicle to which the icicle image belongs; is the bounding box area of ​​the icicle image corresponding to the laser deicing path; is the maximum width that can be taken in the bounding box of the icicle image corresponding to the laser deicing path; is the normalization factor;

[0080] Among them, the normalization factor >0, used for control The value of is too large or too small, and the urgency of the icicles to which each icicle image belongs is calculated based on the above formula. ,by The laser deicing path corresponding to the icicle image with a large value is taken as the priority application target;

[0081] Through the above logic formula, the priority of icicle cleaning is identified, so that the icicle cleaning work based on this method can be carried out more logically, safely and stably.

[0082] The decision logic of the laser deicing path travel speed is expressed as:

[0083] ;

[0084] Where: The time required for a vehicle to enter or exit a tunnel; is the distance between the vehicle and the tunnel entrance and exit; is the vehicle speed; for compensation;

[0085] compensate Customized by the user, compensation >0, compensation The initial default value is 1 second. Calculate and further calculate the corresponding The difference between the two values ​​is selected, and the corresponding source vehicle of the maximum difference calculation result is selected. , recorded as , then the available time domain of the speed path in the laser deicing path is ;

[0086] Calculate the time required for the ice column to fall to the ground based on the speed path in the ice column image, and update the available time domain of the speed path in the laser deicing path based on the calculation result;

[0087] ;

[0088] Where: The time required for the cut-off portion of the icicle to fall to the ground; The distance between the cut-off part of the icicle and the road surface; is the acceleration due to gravity;

[0089] The speed path in the laser deicing path can be updated in the time domain as follows: ;

[0090] The laser deicing path travel speed is:

[0091] ;

[0092] Where: is the moving speed of the pace path; is the pace path length;

[0093] Through the above logic formula, the speed of each speed path in the laser deicing path is finally output.

[0094] The speed and distance of vehicles entering and exiting the tunnel are measured based on the speed and distance measuring camera on the deicing equipment. The laser deicing path is Z-shaped. In the decision-making stage of the laser deicing path travel pace, the path points on both sides of the laser deicing path are captured, and the section between two adjacent path points is used as the pace path, that is, the number of pace paths is the number of laser deicing path travel paces;

[0095] Among them, the laser deicing path is applied to the flight of the UAV in the deicing equipment. After the laser deicing path is created, the UAV flies based on the laser deicing path, so that the end of the laser equipment carried by the UAV is opposite to the lowest end point of the laser deicing path, and enters the waiting stage of flying based on the laser deicing path;

[0096] Pace path moving speed >0, the pace path moving speed is applied, otherwise, it jumps to the stage of measuring the speed and distance of vehicles entering and exiting the tunnel;

[0097] Among them, the moving speed of the pace path When applied, drones in de-icing equipment wait Post-application pace path moving speed Move from the starting point of the pace path to the end of the pace path, and calculate the new pace path moving speed Then, the operation is continued until the end of the speed setting path coincides with the end of the laser deicing path, and then the operation ends;

[0098] When the icicle image in the arch structure image data at the tunnel entrance and exit position is not unique, after each icicle image corresponding to the icicle completes the de-icing operation, the de-icing equipment returns to the position and posture of the arch structure image data collection stage at the tunnel entrance and exit position.

[0099] In this embodiment, by executing the method in the above embodiment, a cleaning technology is effectively provided for the icicles condensed at the top arch position of the tunnel entrance and exit. Compared with the existing technology, it is more intelligent, does not need to prohibit the road for icicle cleaning work, and is safer for vehicles entering and exiting the tunnel. In addition, the broken ice cleaning after wireless icicle cleaning provides safer communication guarantee for winter tunnel road traffic.

[0100] See also Figure 2 As shown, based on the arrow indication, an icicle image composed of sub-images is shown, which is then defined by the sub-image grid in the icicle image, showing the process from acquisition to laser deicing path.

[0101] Embodiment 2:

[0102] In terms of specific implementation, based on Example 1, this example refers to Figure 1 The laser deicing method based on image recognition in Example 1 is further described in detail as shown below:

[0103] The de-icing equipment is integrated by laser equipment, speed and distance measuring cameras, cameras and drones. The drone has a built-in flight path. The drone carries the laser equipment, speed and distance measuring cameras, and cameras to reach the tunnel entrance and exit based on the flight path. The laser equipment is any one of a carbon dioxide laser and a fiber laser.

[0104] The drone is equipped with a gyroscope. When the drone carries a laser device, a speed and distance measuring camera, and a camera and arrives at the tunnel entrance and exit based on the flight path, that is, the end of the flight path, the drone senses the drone's attitude based on the gyroscope. When the drone senses that it is tilted, it automatically adjusts its attitude based on the gyroscope's perception results until it is level.

[0105] The image data of the arch structure at the tunnel entrance and exit is collected by the camera mounted on the deicing equipment. After the image data of the arch structure at the tunnel entrance and exit is collected, it is synchronously identified whether there is an occluded area in the image data. When the identification result is yes, the image data is completed;

[0106] The steps of completing the image data of the arch structure at the tunnel entrance and exit include:

[0107] S1: Acquire image data of the arch structure at the tunnel entrance and exit, convert the image data into a grayscale image, set the occlusion area capture logic, and capture the occlusion area in the image data based on the occlusion area capture logic;

[0108] S2: If the capture result is empty, the image data completion processing operation ends, otherwise, the occluded area is completed;

[0109] The logic of occlusion area completion processing is expressed as:

[0110] ;

[0111] Where: Iterate the target grayscale value for the pixels in the occluded area; is the total number of neighboring pixels of the pixel in the occluded area; is the gray value of the i-th adjacent pixel; is the judgment function;

[0112] The iterative target grayscale value of each pixel in the occluded area is obtained based on the above formula, and the grayscale value of the pixel in the corresponding occluded area is adjusted based on the obtained iterative target grayscale value to complete the completion processing of the occluded area;

[0113] Decision function The value is 0 or the original value of the parameter in the brackets. When the pixel pointed to by the parameter in the brackets and the target pixel of the overall calculation formula are all pixels in the occluded area, =0, otherwise, the judgment function Take the original value of the parameter in brackets, is the judgment function The number of times the value is 0, when the occluded area is completed based on the above formula, the pixels in the occluded area are obtained Afterwards, based on After completing the pixel, the completion operation of the next pixel in the occluded area is performed. When the completion operation of the occluded area is performed in the occluded area, the completion direction is from the periphery to the center.

[0114] Through the above logic formula, the area to be completed in the arch structure image data at the tunnel entrance and exit position is provided with a specified gray value iteration logic, thereby ensuring the completeness of the arch structure image data at the tunnel entrance and exit position based on the gray value iteration operation.

[0115] The logic for capturing the blocked area in the image data of the arched structure at the tunnel entrance and exit is:

[0116] Set the grayscale value interval of the occluded area, apply the grayscale value interval of the occluded area to compare with each pixel in the image data, and capture the pixels that meet the grayscale value interval of the occluded area;

[0117] Identify whether the adjacent pixel of each captured pixel is another captured pixel. When the identification result is yes, combine the two adjacent pixels and record them as a pixel block. Further identify whether the adjacent pixel of the pixel block is another captured pixel. When the identification result is yes, combine the pixel block with the pixel and record them as a pixel block. Repeat this process. When the identification result is no each time, record the identification target as an occluded area.

[0118] Each occlusion region includes at least one pixel block, and the occlusion region is generated in the image data when collecting image data of the arch structure at the entrance and exit of the tunnel in a snowfall scene.

[0119] In this embodiment, through the above settings, the method in Example 1 can be applied to both scenes with snowfall and scenes without snowfall, thereby enhancing the robustness of the method in Example 1 in the application stage.

[0120] In summary, during the execution of the method in the above embodiment, by taking the snowfall scene into consideration, the method can be applicable to both snowfall and non-snowfall scenes, thereby improving the robustness of the application scenario of the method. In the de-icing stage, the icicle cleaning priority and the state of the vehicles in the scene are comprehensively considered to pace the icicle cleaning, thereby realizing unmanned, intelligent and safe autonomous icicle cleaning. During the cleaning process, there is no need to ban tunnels from driving, and the vehicle-free gaps in the tunnel can be effectively utilized to achieve detailed cleaning. Moreover, since the cleaning of icicles is distributed, further cleaning of the icicles is further avoided, thereby ensuring that the icicle cleaning work in the icicle cleaning scenario at the top arch position of the tunnel entrance and exit is carried out stably and safely.

[0121] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A laser deicing method based on image recognition, characterized in that: include: Drive the de-icing equipment to the tunnel entrance and exit, detect the posture of the de-icing equipment, adjust the posture of the de-icing equipment based on the posture detection result of the de-icing equipment, and collect the image data of the arch structure at the tunnel entrance and exit through the de-icing equipment; Capture the icicle image in the arch structure image data at the tunnel entrance and exit, locate the corner points in the icicle image, and create a laser deicing path based on the corner points in the icicle image; Traversing the corresponding laser deicing paths of each icicle image, analyzing the urgency of the icicles to which the corresponding icicle images belong based on each laser deicing path, setting the laser deicing path processing priority based on the urgency analysis result, and applying the laser deicing paths in sequence based on the laser deicing path processing priority setting result; Real-time measurement of the speed and distance of vehicles entering and exiting the tunnel, and decision-making on the laser deicing path based on the collected results; The decision logic of the laser deicing path travel speed is expressed as: ； Where: The time required for a vehicle to enter or exit a tunnel; is the distance between the vehicle and the tunnel entrance and exit; is the vehicle speed; for compensation; compensate Customized by the user, compensation >0, compensation The initial default value is 1 second. Calculate and further calculate the corresponding The difference between the two values ​​is selected, and the corresponding source vehicle of the maximum difference calculation result is selected. , recorded as , then the available time domain of the speed path in the laser deicing path is ; Calculate the time required for the ice column to fall to the ground based on the speed path in the ice column image, and update the available time domain of the speed path in the laser deicing path based on the calculation result; ； Where: The time required for the cut-off portion of the icicle to fall to the ground; The distance between the cut-off part of the icicle and the road surface; is the acceleration due to gravity; The speed path in the laser deicing path can be updated in the time domain as follows: ; The laser deicing path travel speed is: ； Where: is the moving speed of the pace path; is the pace path length.

2. The laser deicing method based on image recognition according to claim 1, characterized in that: The deicing equipment is integrated by laser equipment, speed and distance measuring cameras, cameras and drones. The drone has a built-in flight path. The drone carries the laser equipment, speed and distance measuring cameras, and cameras to arrive at the tunnel entrance and exit based on the flight path. The laser equipment is any one of a carbon dioxide laser and a fiber laser. The drone is equipped with a gyroscope. When the drone carries a laser device, a speed and distance measuring camera, and a camera and arrives at the tunnel entrance and exit based on the flight path, that is, the end of the flight path, the drone senses the attitude of the drone based on the gyroscope. When the drone senses that the drone is tilted, the drone automatically adjusts its attitude based on the gyroscope perception result until it is level. The image data of the arched structure at the tunnel entrance and exit is collected by a camera mounted on the de-icing equipment. After the image data of the arched structure at the tunnel entrance and exit is collected, it is synchronously identified whether there is an occluded area in the image data. When the identification result is yes, the image data is completed.

3. The laser deicing method based on image recognition according to claim 2, characterized in that: The step of completing the image data of the arch structure at the tunnel entrance and exit comprises: S1: Acquire image data of the arch structure at the tunnel entrance and exit, convert the image data into a grayscale image, set the occlusion area capture logic, and capture the occlusion area in the image data based on the occlusion area capture logic; S2: If the capture result is empty, the image data completion processing operation ends, otherwise, the occluded area is completed; The logic of occlusion area completion processing is expressed as: ； Where: Iterate the target grayscale value for the pixels in the occluded area; is the total number of neighboring pixels of the pixel in the occluded area; is the gray value of the i-th adjacent pixel; is the judgment function; The iterative target grayscale value of each pixel in the occluded area is obtained based on the above formula, and the grayscale value of the pixel in the corresponding occluded area is adjusted based on the obtained iterative target grayscale value to complete the completion processing of the occluded area; Decision function The value is 0 or the original value of the parameter in the brackets. When the pixel pointed to by the parameter in the brackets and the target pixel of the overall calculation formula are all pixels in the occluded area, =0, otherwise, the judgment function Take the original value of the parameter in brackets, is the judgment function The number of times the value is 0, when the occluded area is completed based on the above formula, the pixels in the occluded area are obtained Afterwards, based on After completing the pixel, the completion operation of the next pixel in the occluded area is performed. When the completion operation of the occluded area is performed in the occluded area, the completion direction is from the periphery to the center.

4. The laser deicing method based on image recognition according to claim 3, characterized in that: The logic for capturing the blocked area in the image data of the arched structure at the tunnel entrance and exit is: Set the grayscale value interval of the occluded area, apply the grayscale value interval of the occluded area to compare with each pixel in the image data, and capture the pixels that meet the grayscale value interval of the occluded area; Identify whether the adjacent pixel of each captured pixel is another captured pixel. When the identification result is yes, combine the two adjacent pixels and record them as a pixel block. Further identify whether the adjacent pixel of the pixel block is another captured pixel. When the identification result is yes, combine the pixel block with the pixel and record them as a pixel block. Repeat this process. When the identification result is no each time, record the identification target as an occluded area. Each occlusion region includes at least one pixel block, and the occlusion region is generated in the image data when collecting image data of the arch structure at the entrance and exit of the tunnel in a snowfall scene.

5. The laser deicing method based on image recognition according to claim 1, characterized in that: During the data collection phase of the arch structure image at the tunnel entrance and exit and the initial phase of the laser deicing path application, the deicing equipment continues to maintain its original position and posture. The capture logic of the icicle image is as follows: Obtain the image data of the arch structure at the entrance and exit of the tunnel, segment the image data, and define the segmentation accuracy by the user. The length and width of each sub-image obtained by segmentation are equal, and the size of each sub-image obtained by segmentation is the same. Calculate the correlation, energy, and entropy of each sub-image. ； Where: , Determine conditions for icicles; is the gray value of the sub-image; is the sub-image contrast; The gray value determination threshold for the icicle image; is the contrast determination threshold of the icicle image; in, , When the sum is 2, the sub-image is determined to be an icicle image, otherwise it is determined to be a non-icicle image. , Customized by the user end, when the number of sub-images judged as icicle images in the image data does not exceed 1 or there are no adjacent sub-images, the image data segmentation accuracy is adjusted to increase until the number of sub-images judged as icicle images exceeds 1 and there are adjacent sub-images, then the process ends. The combination of adjacent sub-images is the icicle image.

6. The laser deicing method based on image recognition according to claim 1, characterized in that: The corner point in the icicle image is the center point of each sub-image that constitutes the icicle image. After the corner point in the icicle image is located, the position information of the deicing equipment itself and the position information of the tunnel entrance and exit are further obtained. A straight path is determined based on the connection between the two sets of position information. A plane is created at the points on the path that represent the position information of the tunnel entrance and exit, so that the path is perpendicular to the plane. The position information of the corner point is determined in the area where the icicle image is located on the plane with reference to the position information of the deicing equipment itself. After the position information of the positioning corner points is determined, the positioning corner points are adjacently connected to each other based on the position information of each positioning corner point to create a laser deicing path.

7. The laser deicing method based on image recognition according to claim 1, characterized in that: The urgency analysis logic of the icicle to which the icicle image belongs is expressed as follows: ； Where: is the urgency of the icicle to which the icicle image belongs; is the bounding box area of ​​the icicle image corresponding to the laser deicing path; is the maximum width that can be taken in the bounding box of the icicle image corresponding to the laser deicing path; is the normalization factor; Among them, the normalization factor >0, used for control The value of is too large or too small, and the urgency of the icicles to which each icicle image belongs is calculated based on the above formula. ,by The icicles with large values ​​in the icicle images correspond to the laser deicing paths, which are the priority application targets.

8. The laser deicing method based on image recognition according to claim 1, characterized in that: The speed and distance of the vehicles entering and exiting the tunnel are measured based on the speed and distance measuring camera on the deicing equipment. The laser deicing path is in a Z-shape. In the decision-making stage of the laser deicing path travel speed pacing, the path points on both sides of the laser deicing path are captured, and the section between two adjacent path points is used as the speed pacing path, that is, the number of speed pacing paths is the number of laser deicing path travel speed pacing times; Among them, the laser deicing path is applied to the flight of the UAV in the deicing equipment. After the laser deicing path is created, the UAV flies based on the laser deicing path, so that the end of the laser equipment carried by the UAV is opposite to the lowest end point of the laser deicing path, and enters the waiting stage of flying based on the laser deicing path.

9. The laser deicing method based on image recognition according to claim 1, characterized in that: The moving speed of the pace path >0, the pace path moving speed is applied, otherwise, it jumps to the stage of measuring the speed and distance of vehicles entering and exiting the tunnel; Among them, the moving speed of the pace path When applied, drones in de-icing equipment wait Post-application pace path moving speed Move from the starting point of the pace path to the end of the pace path, and calculate the new pace path moving speed The process continues like this until the end of the speed setting path coincides with the end of the laser deicing path.

10. The laser deicing method based on image recognition according to claim 1, characterized in that: When the icicle image in the tunnel entrance and exit arch structure image data is not unique, after each icicle image corresponding to the icicle completes the de-icing operation, the de-icing equipment returns to the position and posture of the tunnel entrance and exit arch structure image data collection stage.

Images