Track planning method for unmanned aerial vehicle operation

Abstract

The invention discloses a trajectory planning method for unmanned aerial vehicle operation, and belongs to the technical field of unmanned aerial vehicle application, and the method specifically comprises the steps: building the connection with a positioning satellite through an unmanned aerial vehicle-mounted positioning module, and obtaining the position information of a to-be-detected building and an unmanned aerial vehicle; the unmanned aerial vehicle is controlled to fly to the position above the to-be-detected building in sequence, an initial front image of the character sign of the to-be-detected building is obtained according to the character sign orientation of the to-be-detected building, preprocessing and direction correction are conducted on the initial front image, and a corrected image is obtained; performing character recognition on the corrected image, verifying the recognized character information and the planning character information of the to-be-detected building, and when the recognition is wrong, marking as a character mark needing to be updated; when the identification is correct, carrying out image comparison on the corrected image and the planning character sign sample of the building to be detected, and if the overlap ratio is lower than a preset threshold value, marking the corrected image as a character sign needing to be updated; according to the invention, high-efficiency and low-cost flight is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle applications, and in particular to a trajectory planning method for unmanned aerial vehicle operations. Background Art

[0002] With the rapid advancement of science and technology, drone technology has achieved significant breakthroughs and has been widely used in many fields. Especially in the field of inspection, the application of drones has demonstrated its significant advantages of efficiency, flexibility and safety. Through high-definition cameras and advanced sensors, drones can conduct detailed visual inspections of urban facilities and identify potential problems. The use of drones not only greatly improves inspection efficiency and reduces labor intensity, but also improves the safety of inspection work and avoids the risk of personnel working in high-risk environments. Therefore, the integration of drone technology has brought revolutionary changes to the field of urban inspection and improved the quality and efficiency of urban management.

[0003] In building maintenance and urban planning, the text signs on buildings are important information carriers. They not only provide the name, function and historical information of the building, but are also an indispensable part of urban management. However, due to environmental factors and the influence of years of use, building text signs may be worn, faded or contaminated, which will affect their recognizability and aesthetics. Therefore, it is necessary to regularly inspect and maintain the integrity and clarity of building text signs.

[0004] In existing drone image recognition systems, drone images are often contaminated by various noises due to the imperfections of imaging systems, transmission media, recording equipment, and changes in weather conditions, making it difficult to accurately identify characters in drone images. Summary of the invention

[0005] The purpose of the present invention is to provide a trajectory planning method for unmanned aerial vehicle operations to solve the following technical problems:

[0006] In existing drone image recognition systems, drone images are often contaminated by various noises due to the imperfections of imaging systems, transmission media, recording equipment, and changes in weather conditions, making it difficult to accurately identify characters in drone images.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A trajectory planning method for unmanned aerial vehicle operation comprises the following steps:

[0009] The drone-mounted positioning module establishes a connection with the positioning satellite to obtain the location information of the buildings to be inspected and the drone, and obtains the shortest route to traverse all the buildings to be inspected based on the location information;

[0010] Control the drone to fly along the shortest route, fly to the top of the buildings to be inspected in turn, adjust the shooting angle of the drone according to the orientation of the text signs of the buildings to be inspected, obtain the initial front image of the text signs of the buildings to be inspected, and pre-process and correct the direction of the initial front image to obtain a corrected image;

[0011] Perform text recognition on the corrected image, and verify the recognized text information with the planned text information of the building to be inspected. When there is a recognition error, mark the building to be inspected as requiring text update;

[0012] When the recognition is correct, the corrected image is compared with the planned text sign sample of the building to be inspected. If the overlap is lower than the preset threshold, the building to be inspected is also marked as requiring updated text signs.

[0013] As a further solution of the present invention: the generation process of the planning text mark sample is:

[0014] Obtain the font, color and spacing corresponding to the planning text information, as well as the color of the signboard where the text is located and the proportion relative to the text, use the signboard as the base map, overlay the text symbols, and generate a planning text sign sample.

[0015] As a further solution of the present invention: the pretreatment process is:

[0016] The initial front image is gray-processed to generate a gray-scale image, a highlight area in the gray-scale image whose gray-scale value is higher than a preset threshold d is identified, the initial front image is obtained, and the layer of the high-brightness area is deleted from the initial front image and replaced with a blank layer.

[0017] As a further solution of the present invention: the process of direction correction is:

[0018] Perform edge detection on the grayscale image, segment the text area and the signboard area, binarize the grayscale image after edge detection to obtain a binary image, select the text area in the binary image, divide the text area into a plurality of character areas, gradually diffuse all the single character areas from the center to the outside until the adjacent single character areas after diffusion contact each other to obtain one or more strip areas, obtain the corresponding straight line equations for the strip areas through median filtering, calculate the average slope of the plurality of straight line equations, perform angle correction on the text area in the initial grayscale image according to the average slope, and convert the text direction to a horizontal direction.

[0019] As a further solution of the present invention: the process of dividing the character area is:

[0020] Scan the corrected image of the sign along the long side of the text area, obtain the grayscale value of each column of pixels in turn, distinguish the pixel grayscale values ​​of the text and the sign, detect the column where the ratio of the number of text pixels to the total number of pixels in the current column is greater than the preset threshold, that is, the starting pixel column A1 of the text, continue to scan the text area until the column where the ratio of the number of text pixels to the total number of pixels in the current column is less than the preset threshold, that is, the ending pixel column A2 of the first character, calculate the pixel width w between A2 and A1, then w is the pixel width of a single character, continue to scan the text area, identify the characters whose ratio of the number of text pixels is greater than the preset threshold and whose pixel width is w, and divide each single character area to obtain several character areas.

[0021] As a further solution of the present invention: when the building text logo is located on the side of the building, the drone is controlled to fly to the side of the building corresponding to the text logo; when the building text logo is located on the top of the building, the drone is controlled to descend to a preset height from the top of the building.

[0022] As a further solution of the present invention: the comparison process between the positive image and the text mark sample is:

[0023] The corrected image and the text logo sample are displayed at the same image resolution, and the proportions a and b of the pixels corresponding to the characters in the entire image are calculated respectively. a / b is the degree of overlap. When the degree of overlap is lower than the preset threshold, it means that the text logo of the building is highly worn.

[0024] Beneficial effects of the present invention:

[0025] A fast static collection route planning method based on the global navigation satellite system onboard an unmanned aerial vehicle is provided. The method can automatically plan the shortest route, efficiently collect images of building text signs, and perform accurate image processing and analysis. When there is an error in text recognition, it indicates that the signboard is severely worn. When the text recognition is correct, further judgment is made by comparing the signboard image with the sample to judge the degree of wear, thereby improving the efficiency and accuracy of building text sign detection, and also improving safety and adaptability. It has significant practical application value for urban building maintenance and planning management. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The present invention will be further described below in conjunction with the accompanying drawings.

[0027] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0029] See also Figure 1 As shown, the present invention is a trajectory planning method for unmanned aerial vehicle operation, comprising the following steps:

[0030] The drone-mounted positioning module establishes a connection with the positioning satellite to obtain the location information of the buildings to be inspected and the drone, and obtains the shortest route to traverse all the buildings to be inspected based on the location information.

[0031] In this step, the drone-mounted positioning module establishes a connection with the positioning satellite to obtain the location information of all buildings to be inspected in a certain area. The drone-mounted positioning module is also used to determine the location information of the drone. The drone-mounted positioning module generates a flight map based on the location information of the buildings to be inspected and the location information of the drone, simulates the flight route on the flight map, and finally determines the shortest route that traverses all the buildings to be inspected. This can reduce the operation time of the drone, improve the efficiency of collecting text sign samples of the buildings to be inspected, and thus improve the overall operation efficiency.

[0032] The drone is controlled to fly along the shortest route and fly above the buildings to be inspected in turn. The shooting angle of the drone is adjusted according to the orientation of the text signs of the buildings to be inspected, an initial front image of the text signs of the buildings to be inspected is obtained, and the initial front image is preprocessed and direction-corrected to obtain a corrected image.

[0033] In this embodiment, the drone can first fly above the building to be inspected, take a bird's-eye view of the building to find the wall where the text logo is located, and then quickly fly to the wall to take pictures, saving shooting time.

[0034] The corrected image is subjected to text recognition, and the recognized text information is verified with the planned text information of the building to be inspected. When there is a recognition error, the building to be inspected is marked as requiring text update.

[0035] In this step, when the drone collects the initial front image, the corresponding building mark to be inspected is added to the initial front image, and the building mark to be inspected can be the building name, building number, etc. When there is a recognition error, that is, it is found that the text information of a building to be inspected does not correspond to the planned text information of the building to be inspected, the corresponding building mark to be inspected can be determined according to the initial front image.

[0036] When the recognition is correct, the corrected image is compared with the planned text sign sample of the building to be inspected. If the overlap is lower than the preset threshold, the building to be inspected is also marked as requiring updated text signs.

[0037] The present invention uses an unmanned aerial vehicle-mounted positioning module to quickly locate the building to be inspected and calculates the shortest route through all buildings, which greatly reduces the time and labor required for manual investigation and improves the efficiency of the inspection process.

[0038] The drone can automatically adjust the shooting angle according to the direction of the text sign of the building to be inspected, ensuring a high-quality initial front image, which is conducive to subsequent image processing and analysis.

[0039] By performing automated steps such as grayscale processing, highlight area deletion, edge detection, and binarization processing on the initial front image, the interference information in the image is effectively removed, the direction of the text is accurately corrected, and the accuracy of text recognition is improved.

[0040] Performing text recognition on the corrected image and verifying it with the planned text information can accurately detect errors in text logos and promptly mark buildings that need updated text logos.

[0041] By calculating the overlap between the corrected image and the planned text sign samples, an objective evaluation standard is provided to determine the degree of wear of the building text signs and thus determine whether they need to be updated.

[0042] The use of drones to inspect text signs on buildings avoids the risks of manual high-altitude operations and improves the safety of workers.

[0043] The present invention helps the city management department to timely discover and update unclear or erroneous building text signs, thereby improving the city's information level and image.

[0044] In another preferred embodiment of the present invention, the planned text logo sample can be generated according to the captured initial front image.

[0045] The generation process of the planning text logo sample is as follows:

[0046] The font, color and spacing corresponding to the planned text information, as well as the color of the signboard where the text is located and the ratio of the text to the signboard are obtained from the initial front image. The signboard is used as the base map, and text symbols are superimposed to generate a planned text sign sample.

[0047] Among them, the font, font color, and signboard color can be directly identified from the initial front image.

[0048] The ratio of the text to the sign needs to be identified based on the text lines and the size of the sign in the image. Specifically, the ratio can be determined based on the pixel points of the text lines and all the pixel points of the initial front image.

[0049] In another preferred embodiment of the present invention, the pretreatment process is:

[0050] The initial front image is gray-processed to generate a gray-scale image, and a highlight area in the gray-scale image whose gray-scale value is higher than a preset threshold d is identified to obtain the initial front image. The layer of the high-brightness area is deleted from the initial front image and replaced with a blank layer, which helps to improve the accuracy of text recognition and reduce noise interference.

[0051] Eliminate the glass or window areas on the building surface to avoid recognizing the text of other buildings reflected in them.

[0052] During the drone shooting process, some interference images of non-inspected buildings may be captured, such as advertising texts set on the inspected buildings. Such interference images also need to be removed. Specifically, the interference images carrying other texts can be removed in combination with the building name of the inspected building to improve the overall detection efficiency.

[0053] In another preferred embodiment of the present invention, the process of direction correction is:

[0054] Perform edge detection on the grayscale image, segment the text area and the signboard area, binarize the grayscale image after edge detection to obtain a binary image, select the text area in the binary image, divide the text area into a plurality of character areas, gradually diffuse all the single character areas from the center to the outside until the adjacent single character areas after diffusion contact each other to obtain one or more strip areas, obtain the corresponding straight line equations for the strip areas through median filtering, calculate the average slope of the plurality of straight line equations, perform angle correction on the text area in the initial grayscale image according to the average slope, and convert the text direction to a horizontal direction.

[0055] Due to the shooting angle or shooting conditions of the drone, the resulting sign image may have an abnormal perspective, so it is corrected according to the text area.

[0056] In another preferred embodiment of the present invention, the process of dividing the character area is:

[0057] Scan the corrected image of the sign along the long side of the text area, obtain the grayscale value of each column of pixels in turn, distinguish the pixel grayscale values ​​of the text and the sign, detect the column where the ratio of the number of text pixels to the total number of pixels in the current column is greater than the preset threshold, that is, the starting pixel column A1 of the text, continue to scan the text area until the column where the ratio of the number of text pixels to the total number of pixels in the current column is less than the preset threshold, that is, the ending pixel column A2 of the first character, calculate the pixel width w between A2 and A1, then w is the pixel width of a single character, continue to scan the text area, identify the characters whose ratio of the number of text pixels is greater than the preset threshold and whose pixel width is w, and divide each single character area to obtain several character areas.

[0058] In another preferred embodiment of the present invention, when the building text mark is located on the side of the building, the drone is controlled to fly to the side of the building corresponding to the text mark. If necessary, the drone flies around the building to be inspected and adjusts the shooting angle until the text mark of the building to be inspected is photographed.

[0059] When the building text logo is located at the top of the building, the drone is controlled to descend to a preset height from the top of the building.

[0060] In another preferred embodiment of the present invention, the comparison process between the rectified image and the text mark sample is as follows:

[0061] The corrected image and the text logo sample are displayed at the same image resolution, and the ratios a and b of the pixels corresponding to the characters in the entire image are calculated respectively. A / b is the overlap. When the overlap is lower than a preset threshold, it indicates that the text logo of the building is highly worn. The preset threshold can be determined according to actual needs and is not limited in this embodiment.

[0062] As described above, the technical solution of the present invention provides a trajectory planning method for UAV operations, which can automatically plan the shortest route based on the satellite positioning of the building to be inspected, efficiently collect images of building text signs, and perform accurate image processing and analysis; when there is an error in text recognition, it means that the signboard is severely worn. When the text recognition is correct, further judgment is made by comparing the signboard image with the sample to judge the degree of wear, thereby improving the efficiency and accuracy of building text sign detection, while also improving safety and adaptability, and having significant practical application value for urban building maintenance and planning management.

[0063] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A trajectory planning method for unmanned aerial vehicle operations, characterized in that: The following steps are involved: The drone-mounted positioning module establishes a connection with the positioning satellite to obtain the location information of the buildings to be inspected and the drone, and obtains the shortest route to traverse all the buildings to be inspected based on the location information; Control the drone to fly along the shortest route, fly to the top of the buildings to be inspected in turn, adjust the shooting angle of the drone according to the orientation of the text signs of the buildings to be inspected, obtain the initial front image of the text signs of the buildings to be inspected, and pre-process and correct the direction of the initial front image to obtain a corrected image; Perform text recognition on the corrected image, and verify the recognized text information with the planned text information of the building to be inspected. When there is a recognition error, mark the building to be inspected as requiring text update; When the recognition is correct, the corrected image is compared with the planned text sign sample of the building to be inspected. If the overlap is lower than the preset threshold, the building to be inspected is also marked as requiring updated text signs.

2. A trajectory planning method for unmanned aerial vehicle operation according to claim 1, characterized in that: The generation process of the planning text logo sample is as follows: Obtain the font, color and spacing corresponding to the planning text information, as well as the color of the signboard where the text is located and the ratio of the text to the signboard. Use the signboard as the base map, overlay the text symbols, and generate a planning text sign sample.

3. A trajectory planning method for UAV operation according to claim 1, characterized in that: The pre-processing process is as follows: The initial front image is gray-processed to generate a gray-scale image, a highlight area in the gray-scale image whose gray-scale value is higher than a preset threshold d is identified, the initial front image is obtained, and the layer of the high-brightness area is deleted from the initial front image and replaced with a blank layer.

4. A trajectory planning method for UAV operation according to claim 3, characterized in that: The process of direction correction is: Perform edge detection on the grayscale image, segment the text area and the signboard area, binarize the grayscale image after edge detection to obtain a binary image, select the text area in the binary image, divide the text area into a plurality of character areas, gradually diffuse all the single character areas from the center to the outside until the adjacent single character areas after diffusion contact each other to obtain one or more strip areas, obtain the corresponding straight line equations for the strip areas through median filtering, calculate the average slope of the plurality of straight line equations, perform angle correction on the text area in the initial grayscale image according to the average slope, and convert the text direction to the horizontal direction.

5. The trajectory planning method for UAV operation according to claim 3, characterized in that: The process of dividing character areas is: Scan the corrected image of the sign along the long side of the text area, obtain the grayscale value of each column of pixels in turn, distinguish the pixel grayscale values ​​of the text and the sign, detect the column where the ratio of the number of text pixels to the total number of pixels in the current column is greater than the preset threshold, that is, the starting pixel column A1 of the text, continue to scan the text area until the column where the ratio of the number of text pixels to the total number of pixels in the current column is less than the preset threshold, that is, the ending pixel column A2 of the first character, calculate the pixel width w between A2 and A1, then w is the pixel width of a single character, continue to scan the text area, identify the characters whose ratio of the number of text pixels is greater than the preset threshold and whose pixel width is w, and divide each single character area to obtain several character areas.

6. A trajectory planning method for unmanned aerial vehicle operation according to claim 1, characterized in that: When the building text logo is located on the side of the building, the drone is controlled to fly to the side of the building corresponding to the text logo. When the building text logo is located on the top of the building, the drone is controlled to descend to a preset height from the top of the building.

7. The trajectory planning method for unmanned aerial vehicle operation according to claim 1, characterized in that: The comparison process between the rectified image and the text logo sample is as follows: The corrected image and the text logo sample are displayed at the same image resolution, and the proportions a and b of the pixels corresponding to the characters in the entire image are calculated respectively. a / b is the degree of overlap. When the degree of overlap is lower than the preset threshold, it means that the text logo of the building is highly worn.

Images