Unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method, system and equipment suitable for complex mountainous area and medium

Abstract

The invention relates to the technical field of visible light multi-angle radiation tea garden gap separation, in particular to an unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method, system and device suitable for a complex mountainous area and a medium, and the method comprises the steps that a tea garden true color image is obtained, data is preprocessed, and a visible light orthoimage is obtained; drawing a section sample line according to the visible light orthoimage, and extracting RGB spectrum curve graphs of each ground feature in different time periods; according to the wave band curve graphs of the ground features on the sample lines at different time periods, analyzing the wave spectrum characteristics of the three wave bands of red, green and blue of the ground features; based on an analysis result, determining characteristics and differences of red, green and blue spectrum curves of each ground feature in different time periods, and designing a separation algorithm formula; and substituting a separation algorithm into the tea garden image data for calculation, and performing threshold segmentation on a single-band image result after calculation to separate tea garden gap ground features so as to extract tea trees.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of visible light multi-angle radiation tea garden gap separation, and particularly relates to a method, system, device and medium for unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation suitable for complex mountainous areas. BACKGROUND

[0002] As the largest tea producing country in the world, China's tea garden area and yield rank first in the world. Accurate acquisition of tea garden planting area and actual picking area data is not only of great significance to agricultural planning, yield estimation and ecological management, but also provides a basis for the government to formulate more reasonable agricultural policies, supports tea farmers to improve yield, and promotes the sustainable development of the tea industry. Accurate area data can effectively guide resource allocation and management decision-making, thus promoting the process of agricultural modernization.

[0003] In recent years, with the rapid development of unmanned aerial vehicle remote sensing technology, its application in the field of agriculture is increasingly widespread. The remote sensing equipment carried by unmanned aerial vehicles can quickly acquire high-resolution agricultural images of a large area, providing efficient and convenient technical support for crop monitoring, pest detection, yield prediction, etc. In particular, in tea garden management, unmanned aerial vehicle remote sensing technology has become an important means to replace traditional manual monitoring due to its efficiency and convenience. This technology significantly reduces labor costs and improves management efficiency, providing new possibilities for the scale and fine management of tea gardens.

[0004] However, the existing technology still has obvious deficiencies in the processing and analysis of tea garden remote sensing images. Current research and application focuses on the extraction of the overall area of the tea garden, while the fine separation of the gap area of the tea garden is not adequately addressed. Existing methods often rely on multispectral or radar data for data acquisition, which, although rich in information, is costly and difficult to adapt to low-cost visible light unmanned aerial vehicle images. At the algorithm level, the limitations of the technology are also significant: while deep learning models have high potential, they require a large amount of labeled data for training, which is costly and time-consuming; while traditional threshold segmentation methods are simple and easy to use, they lack precision and are difficult to meet the needs of tea garden image processing under complex terrain and lighting conditions. These problems make it difficult to accurately calculate the effective planting area of the tea garden, and cannot meet the management needs of high precision and efficiency.

[0005] In addition, current research on tea garden gap separation is still relatively limited. Due to the inability to effectively identify and exclude the gap area, the calculation of the actual picking area of the tea garden is often not accurate enough, directly affecting the scientificity of agricultural planning and management decisions. Therefore, it is urgent to develop a high-precision tea garden gap separation method to overcome the limitations of existing technology and improve the automation and intelligence level of tea garden management, providing technical support for the sustainable development of the tea industry. SUMMARY

[0006] In view of the above existing problems, the present application is proposed.

[0007] The present application aims to solve the problem of few methods and low precision in extracting complex mountainous tea garden gaps based on remote sensing information at present, and proposes a UAV visible light multi-angle radiation tea garden gap separation method, system, equipment and medium suitable for complex mountainous areas. The method can adapt to the situation of complex mountainous tea gardens under different sunlight conditions, effectively improving the inaccurate problem of current actual tea garden picking area extraction. The method principle of the present application is simple and clear, the implementation process is relatively simple, and has strong practicality.

[0008] To solve the above technical problems, the present application provides the following technical solutions:

[0009] In the first aspect, the present application provides a UAV visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, comprising:

[0010] Obtain a true color image of a tea garden, pre-process the data, and obtain a visible light orthographic image;

[0011] According to the visible light orthographic image, a profile sample line is drawn, and the visualization data graph of each ground object at different time periods is extracted;

[0012] According to the visualization data graph on the sample line of each ground object at different time periods, the wave characteristics of the ground object wave band are analyzed;

[0013] Based on the analysis result, the characteristics and differences of the spectral curve of each ground object at different time periods are determined, and a separation algorithm is designed;

[0014] The separation algorithm is brought into the tea garden image data for calculation, and the single-band image result after operation is separated through image segmentation, and the tea tree is extracted.

[0015] As an optimal solution of the UAV visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, wherein: the true color image of the tea garden is obtained, the data is pre-processed, and the visible light orthographic image is obtained, comprising:

[0016] Obtain the tea garden UAV image data of the research area at noon and in the afternoon;

[0017] Based on the visible light probe carried by the UAV, the image data of the research area is obtained;

[0018] The data is pre-processed by using a processing software to obtain the visible light orthographic image of the research area.

[0019] As a preferred scheme of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, according to the visible light orthographic image, a profile sample line is drawn, and a visualization data graph of each ground object at different time periods is extracted, including:

[0020] Based on the obtained visible light orthographic image, profile sample lines are drawn on tea trees and tea garden gaps and other ground objects, respectively;

[0021] According to the profile sample line, a wave band curve graph of each ground object at different time periods is extracted.

[0022] As a preferred scheme of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, according to the visualization data graph on the sample line of each ground object at different time periods, the wave band characteristics of the ground object are analyzed, including:

[0023] By analyzing the wave band curve graph of the ground object on a specific profile sample line, a spectral feature mode is extracted;

[0024] Considering the influence of the time period on the spectral value size, the feature of each ground object is extracted by comparing the feature mode.

[0025] As a preferred scheme of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, based on the analysis result, the characteristics and differences of the spectral curve of each ground object at different time periods are determined, and a separation algorithm is designed, including:

[0026] According to the characteristics of the tea garden ground object, the ground object information is judged;

[0027] According to the judgment result of the ground object information, a separation algorithm is designed, and a single wave band image result is calculated.

[0028] As a preferred scheme of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, based on the analysis result, the characteristics and differences of the spectral curve of each ground object at different time periods are determined, and a separation algorithm is designed, including:

[0029]

[0030] T2 = min (b1, b2) - b3

[0031] T = T1 * T2

[0032] Wherein, T is the final single wave band image result of tea tree and tea garden gap separation calculation, T1 represents the calculation formula of highlighting tea trees and suppressing tea garden gaps, T2 represents the calculation formula of separating similar tea trees and tea garden gaps, b1 represents the red wave band, b2 represents the green wave band, b3 represents the blue wave band, and k represents a very small constant.

[0033] As a preferred scheme of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas, wherein: the separation algorithm is brought into the tea garden image data for calculation, and the single-band image result after operation is separated through image segmentation to extract tea trees from the tea garden gap features, comprising:

[0034] The calculation formula is brought into the original image to calculate the single-band image with the result of T1, and threshold segmentation is performed. According to the test separation effect, when T1>3.725, the separation degree of tea trees and other features is the best, that is, T1>3.725 is tea trees, and T1≤3.725 represents a tea garden gap. The value of T1≤3.725 is set to 0.

[0035] The calculation formula is brought into the original image to calculate the single-band image with the result of T2. Finally, the result after T1 threshold segmentation is multiplied by T2 to obtain the single-band image of T. Threshold segmentation is used again, T>235.882 is set as tea trees, and T≤235.882 is set as a tea garden gap to obtain the actual tea garden picking area.

[0036] The beneficial effects of the preferred technical scheme are: the technology analyzes the spectral characteristic differences of different features in the red, green and blue (RGB) bands, and designs an innovative separation algorithm formula. The formula uses a specific combination of red, green and blue bands (b1, b2, b3) and a minimum constant k to generate two key intermediate results: T1 (highlighting tea trees and suppressing tea garden gaps) and T2 (separating similar tea trees and tea garden gaps). When implemented, the formula is first applied to the original image data to generate single-band images of T1 and T2. Then, threshold segmentation is performed on T1. When T1>3.725, it is determined as tea trees, otherwise it is determined as a tea garden gap and set to 0. Then, the segmentation result of T1 is multiplied by T2 to obtain the final single-band image T, and secondary threshold segmentation is performed. T>235.882 is determined as tea trees, and otherwise it is determined as a tea garden gap, thereby accurately extracting the tea tree area. The beneficial effects of the preferred scheme are: through multi-stage calculation and threshold segmentation, high-precision separation of tea trees and tea garden gap features is achieved, the robustness of the algorithm is significantly improved, and the light changes and shadow interference in complex mountainous environments are effectively dealt with. At the same time, only visible light data is required, avoiding the dependence on high-cost equipment and deep learning training, reducing the technical threshold, and finally providing reliable technical support for accurate calculation and fine management of tea garden picking area, solving the area calculation deviation problem caused by gap feature interference in traditional methods.

[0037] In the second aspect, the present application provides a unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation system suitable for complex mountainous areas, comprising: a collection module for obtaining a true color image of a tea garden, pre-processing the data, and obtaining a visible light orthographic image;

[0038] an extraction module, which draws a profile sample line according to the visible light orthographic image, and extracts the RGB spectrum curve of each ground object at different time periods;

[0039] an analysis module, which analyzes the wave characteristics of the red, green and blue bands of each ground object at different time periods according to the band curve on the sample line;

[0040] a calculation module, which determines the characteristics and differences of the red, green and blue spectrum curves of each ground object at different time periods based on the analysis results, and designs a separation algorithm formula;

[0041] an output module, which brings the separation algorithm into tea garden image data for calculation, and separates the ground objects in the tea garden gaps by threshold segmentation after the operation of the single-band image results, and then extracts the tea trees.

[0042] In a third aspect, the present application provides an electronic device, comprising:

[0043] a memory and a processor;

[0044] The memory is used to store computer executable instructions, and the processor is used to execute the computer executable instructions, which realize the steps of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas.

[0045] In a fourth aspect, the present application provides a computer readable storage medium, which stores computer executable instructions, which realize the steps of the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas when executed by a processor.

[0046] Compared with the prior art, the present application has the following beneficial effects: from the data level, the present application uses unmanned aerial vehicle visible light band images, which has low data acquisition cost and high flexibility. Visible light cameras are common and low-cost choices for unmanned aerial vehicle equipment, and do not require additional multi-spectral sensors or satellite data acquisition permissions, especially for complex mountainous areas where satellite remote sensing may be blocked or have unstable data quality. At the same time, unmanned aerial vehicles can flexibly adjust the shooting time and angle according to actual needs, such as through multi-angle acquisition at noon and in the afternoon, which can quickly complete data acquisition in the same area, greatly improving the efficiency and timeliness of data acquisition; from the core processing method, the present application is based on the spectrum curve characteristics of visible light bands, and realizes ground object separation by designing specific band calculation formulas, which eliminates the dependence on elevation information and multi-spectral vegetation index, has a more concise processing flow, and has wider applicability; for complex mountainous areas, the present application has unique advantages. Using multi-angle radiation data to enhance the distinguishing ability, effectively overcoming the influence of mountainous terrain undulations and complex lighting conditions on image analysis, significantly improving the accuracy and stability of ground object identification in complex mountainous environments.

[0047] In summary, the present application has obvious advantages in data acquisition cost, processing efficiency, environmental adaptability and application promotion, and provides an efficient, economical and accurate solution for ground object identification in complex terrain. BRIEF DESCRIPTION OF DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0049] Figure 1 The technical flowchart of the embodiment of the present application.

[0050] Figure 2 The afternoon period grassland, road and tea tree profile spectrum graph of the embodiment of the present application.

[0051] Figure 3 The afternoon period tea garden gap grassland and tea tree profile spectrum graph of the embodiment of the present application.

[0052] Figure 4 The afternoon period tea garden gap grassland, tea tree and bare soil profile spectrum graph of the embodiment of the present application.

[0053] Figure 5 The afternoon period tea garden gap grassland and tea tree spectrum characteristic confusion profile spectrum graph of the embodiment of the present application.

[0054] Figure 6 The noon period tea garden gap grassland and tea tree profile spectrum graph of the embodiment of the present application.

[0055] Figure 7 The noon period tea garden gap bare soil, grassland and tea tree profile spectrum graph of the embodiment of the present application.

[0056] Figure 8 The noon and afternoon period original image of the embodiment of the present application.

[0057] Figure 9 The tea garden gap elimination result of the embodiment of the present application. DETAILED DESCRIPTION

[0058] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor should belong to the protection scope of the present application.

[0059] Embodiment 1, refer to Figure 1 For an embodiment of the present application, a UAV visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas is provided, comprising:

[0060] S1: Obtain a true color image of a tea garden, pre-process the data, and obtain a visible light orthographic image;

[0061] S2: According to the visible light orthographic image, draw a profile sample line, and extract the visualization data graph of each ground object at different time periods;

[0062] S3: According to the visualization data graph on the sample line of each ground object at different time periods, analyze the wave characteristics of the ground object wave band;

[0063] S4: Based on the analysis result, determine the characteristics and differences of the spectral curve of each ground object at different time periods, and design a separation algorithm;

[0064] S5: Bring the separation algorithm into the tea garden image data for calculation, and separate the tea trees by image segmentation through the single-band image result after operation.

[0065] It should be noted that the visible light image obtained by the low-cost UAV is pre-processed and the RGB spectrum curve is extracted by the profile sample line, which effectively avoids the limitation of relying on high-cost multi-spectral or radar data. Subsequently, the visible light wave band characteristics are analyzed and a separation algorithm without a large amount of labeled data is designed, which not only solves the problems of high training cost and long time consumption of deep learning model, but also overcomes the defects of insufficient precision of traditional threshold method. Finally, the algorithm can accurately separate the tea garden gap ground object and extract the tea trees, solve the problem that the existing technology pays insufficient attention to the tea garden gap area, leading to inaccurate calculation of the actual picking area, thereby providing strong technical support for high-precision tea garden area statistics and fine management, and meeting the scientific needs of agricultural planning and management.

[0066] Embodiment 2, refer to Figures 1-9 For an embodiment of the present application, based on the above embodiment, a UAV visible light multi-angle radiation tea garden gap separation method suitable for complex mountainous areas is provided.

[0067] In the embodiments of the present application, the tea garden true color image obtained in step S1 is preprocessed to obtain a visible light orthographic image, including steps A1-A3:

[0068] A1: Obtain tea garden unmanned aerial vehicle image data of the study area at noon and in the afternoon

[0069] A2: Obtain image data of the study area based on the visible light probe carried by the unmanned aerial vehicle

[0070] A3: Preprocess the data using processing software to obtain a visible light orthographic image of the study area.

[0071] It should be noted that the specific implementation of steps A1-A3 is:

[0072] The unmanned aerial vehicle carries a visible light camera to take aerial photographs of the mountainous tea garden at noon and in the afternoon, respectively, to obtain a tea garden RGB true color image, where R, G, and B represent the abbreviations of red, green, and blue bands, respectively. The present application obtains tea garden image data based on DJI Jingling 4 unmanned aerial vehicle, sets the flight height to 100 meters, and ensures that the ground resolution is not less than 0.05 meters.

[0073] In the embodiments of the present application, the data is preprocessed using processing software in step A3 to obtain a visible light orthographic image of the study area. The software used for preprocessing the data is DJI mapping and ENVI, etc., which includes image stitching and cropping to obtain a visible light orthographic image of the study area.

[0074] In an optional embodiment, the preprocessing can also use image radiation correction and light balance processing to uniformly process the brightness of images collected at different flight times and under different light conditions, so as to reduce the image brightness difference caused by the change of solar elevation angle and improve the accuracy of subsequent analysis.

[0075] In another optional embodiment, the preprocessing can also use image geometric correction and geographic registration processing to match and correct the coordinates of the obtained images, so that they are accurately aligned with the true geographical position, ensuring that subsequent ground object identification, change detection and other processing have spatial consistency and comparability.

[0076] In the embodiments of the present application, step S2 includes drawing a profile sample line according to the visible light orthographic image and extracting a visualization data graph of each ground object at different time periods, including steps B1-B2:

[0077] B1: Draw a profile sample line on the tea tree and the gap between the tea garden and the ground object based on the obtained visible light orthographic image;

[0078] B2: Extract a band curve graph of each ground object at different time periods according to the profile sample line.

[0079] In the embodiments of the present application, the visual data graph is an RGB band curve graph.

[0080] It should be noted that the extraction of the visual data graph of each ground object at different time periods is based on the visible light band orthographic image of the tea garden of the unmanned aerial vehicle, and the profile sample lines are drawn on the tea trees and the gaps between the tea gardens and other ground objects, and the RGB band curve graph of each ground object at different time periods is extracted.

[0081] In an optional embodiment, the visual data graph can be a normalized vegetation index (NDVI) curve graph. If the unmanned aerial vehicle data contains a near-infrared band, the NDVI can be calculated, and a curve graph of its change over time can be drawn.

[0082] In another optional embodiment, the visual data graph can be a multi-spectral / hyper-spectral band curve graph. If the unmanned aerial vehicle is equipped with a multi-spectral or hyper-spectral sensor (which usually also contains a visible light band), the visual data graph can not only be the RGB three bands, but also contain more bands (such as blue edge, green light, red edge, near-infrared, etc.) curve graph. In this way, the spectral characteristics of the ground object can be more comprehensively reflected.

[0083] In the embodiments of the present application, the step S3 of analyzing the spectral characteristics of the R, G and B bands of the ground object according to the band curve graph on the sample line of each ground object at different time periods includes steps C1-C2

[0084] C1: Extract the spectral feature mode by analyzing the band curve graph of the ground object on a specific profile sample line

[0085] C2: Considering the influence of time period on the size of spectral value, distinguish each ground object by comparing the feature mode.

[0086] It should be noted that the specific implementation of steps C1-C2 is as follows:

[0087] According to the obtained band curve graph on the sample line of each ground object at different time periods, the spectral characteristics of the R, G and B bands of the ground object are analyzed. Specifically, Figures 2 to 7 The GRB profile spectrum graph of the profile sample line of each ground object at different time periods is shown, and the line segment and arrow correspond to the profile sample line area and the indicated spectrum area, respectively. The color of the spectrum corresponds to the corresponding red, green and blue bands, respectively. The horizontal axis value is the cross-sectional position of the sample line, and the vertical axis value is the spectrum size. The yellow sample line and arrow represent grassland, the orange sample line represents road and bare land, and the black sample line represents tea tree. Figure 2 As shown in the profile spectrum graph of the grassland, road and tea tree at the afternoon period, the spectral characteristics of the grassland are that the green band is greater than the red band, the red band is greater than the blue band, and the difference between the adjacent bands is approximately equal. The spectral characteristics of the road are that the red band is greater than the green band, the green band is greater than the blue band, and the difference between the adjacent bands is also approximately equal. The spectral characteristics of the tea tree are that the red band and the green band are similar in size, the bands are intertwined with each other, and the blue band is the smallest;Figure 3 The profile spectrum of the grassland and tea tree in the afternoon is shown. The spectral characteristics of the grassland are that the green band is greater than the red band, the red band is greater than the blue band, and the difference between the adjacent bands is approximately arithmetic progression. The spectral characteristics of the tea tree are that the red band and the green band are similar in size, the bands are intertwined with each other, and the blue band is the smallest; Figure 4 The profile spectrum of the grassland, tea tree and bare soil in the afternoon is shown. The spectral characteristics of the grassland are that the green band is greater than the red band, and the red band is greater than the blue band. The spectral characteristics of the tea tree are that the green band is greater than the red band, the red band is greater than the blue band, and the difference between the red and green bands is very small. The spectral characteristics of the bare soil are that the red band is greater than the green band, and the green band is greater than the blue band. The difference between the adjacent bands is also approximately arithmetic progression, which is similar to the spectral characteristics of the road; Figure 5 The profile spectrum of the grassland and tea tree in the afternoon is shown. The spectral characteristics of the tea tree are that the red band and the green band are similar in size, the bands are intertwined with each other, and the blue band is the smallest. The grassland and the bare soil in the figure are mixed, which appears brown, and the spectral characteristics are that the red and green bands are similar in size, and the blue band is the smallest, which is similar to the characteristics of the tea tree. The difference between the red and green bands and the blue band is that the tea tree is greater than the grassland; Figure 6 The profile spectrum of the grassland and tea tree in the afternoon is shown. Due to the poor growth of the tea trees in the area, the grass is lush, and the ground object structure is also complex. The spectral characteristics of the grassland are that the green band is greater than the red band, and the red band is greater than the blue band. The spectral characteristics of the tea tree are that the red band and the green band are similar in size, the bands are intertwined with each other, and the blue band is the smallest. Due to the direct sunlight at noon, the spectral values of the grassland and the tea tree are not significantly different. Figure 7 The profile spectrum of the grassland, tea tree and bare soil in the afternoon is shown. The spectral characteristics of the tea tree are that the red band and the green band are similar in size, the bands are intertwined with each other, and the blue band is the smallest. The spectral characteristics of the grassland are that the green band is greater than the red band, and the red band is greater than the blue band, and the difference between the red and green bands is smaller than that in the afternoon. The spectral characteristics of the bare soil are that the red band is greater than the green band, and the green band is greater than the blue band. The difference between the adjacent bands is also approximately arithmetic progression.

[0088] In summary, the spectrum of each object in the tea garden under different solar radiation angles will be different, but also retains the inherent characteristics. In general, the spectral characteristics of tea plants are that the difference between red and green bands is small, and the difference between red and green bands and blue band is large. The spectral characteristics of grassland in the afternoon period are that the green band is greater than the red band, the red band is greater than the blue band, and the difference between green and red bands and the difference between red and blue bands are comparable. Due to the influence of shadow, the values of red, green and blue bands are relatively low. The spectral characteristics of grassland at noon are similar to those in the afternoon period, except that the difference between green and red bands is small and the difference between red and blue bands is large. Due to the vertical radiation of the sun at noon, the values of red, green and blue bands are comparable to the spectral values of tea plants at the same period. The spectral characteristics of bare soil in the afternoon period are that the spectral value of the red band is greater than that of the green band, the spectral value of the green band is greater than that of the blue band, and the difference between red and green bands is comparable to the difference between green and blue bands. Due to the influence of shadow, the spectral values of red, green and blue are small. The spectral characteristics of bare soil at noon are similar to those in the afternoon period, but due to the direct sunlight, the spectral values of red, green and blue are large, which are similar to the spectrum of the road.

[0089] In the embodiments of the present application, based on the results of the analysis in step S4, the characteristics and differences of the red, green and blue spectral curves of each object at different periods are determined, and a separation algorithm formula is designed. The specific implementation is as follows:

[0090] It should be noted that based on the summarized spectral information characteristics of various objects in tea garden at different time periods, a band calculation formula is designed to distinguish tea trees from other objects. For intuitive and simple expression, we use R, G and B to represent red, green and blue bands respectively. The main feature of tea trees is that R and G are very close, while they are quite different from B. Bare soil, roads and grasslands have more uniform differences, but grasslands are sometimes similar to tea trees, only the difference between R and G and B is smaller, and the overall spectral value is smaller in the afternoon period. The spectrum of bare soil and roads is similar to that of grassland, except that the overall spectral value of bare soil and roads is larger, and R is greater than B, which is opposite to that of grassland. According to the characteristics of tea garden objects, tea trees: R and G band values are close (R≈G), sometimes R>G, and sometimes G>R, but R and G are much larger than B (i.e. B is the smallest). The difference between R and G is small, while the difference between them and B is large. Bare soil and roads: R>G>B, the difference between the three bands is considerable (i.e. R-G≈G-B). Grassland: G>R>B, the difference between the three bands is considerable (G-R≈R-B). But in some areas, the spectrum of grassland is similar to that of tea trees, that is, R and G are close, but the difference between R and G and B is smaller than that of tea trees. The closeness of R and G of tea trees, i.e. |R-G|, gets the minimum of tea trees, and the other objects are larger. Combined with the distance between R and G and B of tea trees, i.e. even taking the smaller one of red and green, and then subtracting the blue band, the value obtained is still larger, i.e. min(R,G)-B gets the maximum of tea trees, and the other objects are smaller. Combine the two formulas min(R,G)-B / |R-G| to further enlarge the gap between tea trees and other objects. For grassland, although R and G are sometimes close, the difference between R and G and B is also relatively small, and min(R,G)-B can not only suppress grassland, but also get a relatively small value for bare soil and roads.

[0091] In the embodiments of the present application, the formula is designed as:

[0092]

[0093] T2 = min(b1, b2) - b3 (2)

[0094] T = T1 x T2 (3)

[0095] In the formula, T represents the final single-band image result of the tea tree and the tea garden gap separation calculation, T1 represents the calculation formula of highlighting the tea tree and suppressing the tea garden gap, T2 represents the calculation formula of separating the similarity of the tea tree and the tea garden gap, b1 represents the red band, b2 represents the green band, b3 represents the blue band, k represents a very small constant, and the denominator is prevented from being zero. The formula is designed by using the characteristics that the tea tree is close in the red and green bands and the minimum band is still greatly different from the blue band, and the characteristics that the grassland is close in the red and green bands but the minimum band is small. T1 and T2 are multiplied to further expand the difference between the tea tree and the grassland, and the absolute difference value of the red and green bands in the denominator strengthens the selection of the red and green close objects. Therefore, the formula is targeted in mathematics and can effectively extract the tea tree and suppress the grassland and bare soil with similar spectra.

[0096] In an optional embodiment, the formula design can be based on the difference between the average value of the red and green bands and the blue band. First, the average value of the red and green bands is calculated, and then the difference with the blue band is calculated. Since the tea tree is close in the red and green bands, the average value can more accurately reflect the overall reflection characteristics, and the blue band is obviously lower. Through the difference, the difference of the tea tree in the overall reflection characteristics can be further highlighted, and other objects such as grassland, bare soil and road can be effectively suppressed.

[0097] In another optional embodiment, the formula design can introduce the maximum band value for normalization to enhance the separation effect. Based on the analysis of the red, green and blue bands, the maximum value is selected as the reference standard of overall brightness, and the result is normalized. This method has stronger stability under different light conditions, can appropriately reduce the influence of objects (such as bare soil and road) with high brightness but not meeting the structural characteristics of tea trees, while maintaining the difference of tea trees, thereby enhancing the separation effect.

[0098] In the embodiment of the application, the separation algorithm is brought into the tea garden image data for calculation in step S5, and the single-band image result after operation is separated by image segmentation to extract the tea tree from the tea garden gap object. The specific implementation manner is as follows:

[0099] In the embodiment of the application, the image segmentation adopts threshold segmentation, and the specific implementation manner is as follows:

[0100] Based on the designed formula, first, the formula (1) is brought into the original image to calculate the single-band image with the result of T1, and threshold segmentation is performed. According to the separation effect of the test, when T1>3.725, the separation degree of tea trees and other ground objects is the best, that is, T1>3.725 is tea trees, and T1≤3.725 represents tea garden gaps. In order to simplify the separation effect for subsequent calculation, the value of T1≤3.725 is set to 0. The formula (2) is brought into the original image to calculate the single-band image with the result of T2. Finally, the result of the T1 threshold segmentation is multiplied by T2 to obtain the single-band image of T. Finally, threshold segmentation is used again, when T>235.882, the extraction effect of tea trees is the best, that is, T>235.882 is tea trees, and T≤235.882 is tea garden gaps, that is, the actual tea garden picking area is obtained, as shown in FIG. 6. The green part is tea trees, and the black part is other ground objects in the tea garden gaps. Figure 9

[0101] In an optional embodiment, image segmentation can be edge detection method. First, an edge detection operator (such as Sobel operator) is applied to the single-band image T1 or T2 calculated in step S5 to identify the edges with sharp changes in brightness in the image. These edges are usually located at the junction of tea trees and tea garden gap ground objects. Then, by connecting these edge lines, the approximate outline of the tea trees can be outlined, so as to separate the tea tree area from the background area. This method focuses on capturing the boundaries between objects.

[0102] It should be noted that although the edge detection method can capture the approximate outline of the object, its robustness and accuracy may not be as good as the threshold segmentation method based on specific features (T1, T2) in the scene of tea garden which may have noise, irregular shape and low contrast. Threshold segmentation directly utilizes the statistical difference of tea trees and background in T1, T2 value, and can more stably and directly realize region division.

[0103] In an optional embodiment, image segmentation can be clustering analysis method. The single-band images T1 and T2 calculated in step S5 are taken as input features (for example, each pixel point can have two feature values: T1 value and T2 value), and then K-means algorithm is applied. K is set to 2 in advance (assuming there are only two categories of tea trees and background), and the algorithm will automatically classify the pixels into two categories according to their feature value similarity. Usually, tea tree pixels will be clustered into one category due to their unique spectral characteristics (reflected in T1 and T2 values), and tea garden gap ground objects will be clustered into another category. In this way, the segmentation result can be directly obtained.

[0104] It should be noted that in the specific application of tea tree recognition, because of large calculation amount, sensitivity to parameters, lack of direct use of specific tea tree features, and possible production of not sharp enough boundaries, the threshold segmentation method is more effective and convenient. ​

[0105] It should be noted that the above examples are only used to illustrate the technical solutions of the present application and not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.

[0106] In order to verify the accuracy of the tea tree extraction, 100 sample points are randomly generated by arcgis, and the true values of the sample points are assigned by visual interpretation with the true color image as a reference. Then, the sample points are intersected with the classified image, and the confusion matrix of the sample points and the classified image is counted to evaluate the accuracy, as shown in Table 1. The overall accuracy is 93%, and the Kappa coefficient is 0.8453. From the accuracy evaluation, the classification accuracy is high.

[0107] Table 1

[0108]

[0109] It should be noted that the above examples are only used to illustrate the technical solutions of the present application and not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present application, and all of them should be covered in the scope of the claims of the present application.

[0110] In the above, it is a schematic scheme of a method for separating tea garden gaps by unmanned aerial vehicle visible light multi-angle radiation suitable for complex mountainous areas. It should be noted that the technical scheme of the system for separating tea garden gaps by unmanned aerial vehicle visible light multi-angle radiation suitable for complex mountainous areas and the technical scheme of the method for separating tea garden gaps by unmanned aerial vehicle visible light multi-angle radiation suitable for complex mountainous areas described above belong to the same concept. The technical scheme of the system for separating tea garden gaps by unmanned aerial vehicle visible light multi-angle radiation suitable for complex mountainous areas in the present embodiment is not described in detail, and can be referred to the description of the technical scheme of the method for separating tea garden gaps by unmanned aerial vehicle visible light multi-angle radiation suitable for complex mountainous areas.

[0111] The present embodiment also provides a system for separating tea garden gaps by unmanned aerial vehicle visible light multi-angle radiation suitable for complex mountainous areas, comprising:

[0112] The collection module acquires a true color image of a tea garden, pre-processes data, and obtains a visible light orthographic image.

[0113] The extraction module draws a profile sample line according to the visible light orthographic image, and extracts an RGB spectrum curve of each ground object at different time periods.

[0114] The analysis module analyzes the wave characteristics of the red, green and blue bands of each ground object at different time periods according to the wave band curve on the sample line.

[0115] The calculation module determines the characteristics and differences of the red, green and blue spectrum curves of each ground object at different time periods based on the analysis results, and designs a separation algorithm formula.

[0116] The output module brings the separation algorithm into tea garden image data for calculation, and separates the tea garden gap ground object through threshold segmentation after the operation of the single-band image result, and then extracts the tea tree.

[0117] The embodiment also provides an electronic device suitable for the unmanned aerial vehicle visible light multi-angle radiation tea garden gap separation in complex mountainous areas, which comprises a memory and a processor.

[0118] The embodiment also provides a storage medium having a computer program stored thereon, and the program is executed by a processor to realize the method for separating the tea garden gap in complex mountainous areas by the unmanned aerial vehicle visible light multi-angle radiation.

[0119] The storage medium provided by the embodiment belongs to the same inventive concept as the method for separating the tea garden gap in complex mountainous areas by the unmanned aerial vehicle visible light multi-angle radiation, and the technical details not described in the embodiment can be referred to the above embodiment, and the embodiment has the same beneficial effects as the above embodiment.

[0120] Those skilled in the art can clearly understand the present application by the above description of the embodiments, and the present application can be realized by software and necessary general hardware, and of course, can also be realized by hardware. Based on such understanding, the technical solutions of the present application or the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a floppy disk, a read-only memory (ROM), a random access memory (RAM), a FLASH, a hard disk, or an optical disc, and includes a number of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods of various embodiments of the present application.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application but not limit the present application, and although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and all should be covered in the scope of the claims of the present application.

Claims

1. A method for separating gaps in tea gardens using visible light multi-angle radiation from unmanned aerial vehicles (UAVs) adapted to complex mountainous areas, characterized in that... include: Acquire true-color images of the tea garden, preprocess the data, and obtain visible light orthophotos; Based on visible light orthophotos, profile lines are drawn to extract visualization data maps of various features at different times. Based on the visualized data maps of ground features at different time periods, the spectral characteristics of ground feature bands are analyzed; Based on the analysis results, the characteristics and differences of the spectral curves of various objects at different times were determined, and a separation algorithm was designed. The separation algorithm was applied to the tea garden image data for calculation, and the single-band image results were then processed by image segmentation to separate the ground features in the tea garden and extract the tea trees.

2. The method for separating gaps in tea gardens using visible light multi-angle radiation from unmanned aerial vehicles (UAVs) adapted to complex mountainous areas, as described in claim 1, is characterized in that... The process of acquiring a true-color image of the tea garden and preprocessing the data to obtain a visible light orthophoto image includes: Acquire drone imagery data of tea gardens in the study area during two time periods: noon and afternoon; Image data of the study area were acquired using a visible light sensor mounted on a drone; The data was preprocessed using processing software to obtain visible light orthophotos of the study area.

3. The method for separating gaps in tea gardens using visible light multi-angle radiation from unmanned aerial vehicles (UAVs) adapted to complex mountainous areas, as described in claim 2, is characterized in that... The process of drawing profile lines based on visible light orthophotos and extracting visualization data maps of various landforms at different times includes: Based on the obtained visible light orthophotos, profile lines were drawn on the gaps between tea trees and tea gardens; Wavelength curves of different time periods were extracted from the profile transects.

4. The method for separating gaps in tea gardens using visible light multi-angle radiation from unmanned aerial vehicles (UAVs) adapted to complex mountainous areas, as described in claim 3, is characterized in that... The analysis of the spectral characteristics of ground features based on visualized data maps of transects at different time periods includes: Spectral feature patterns are extracted by analyzing the band curves of ground features along specific profile lines; Considering the influence of time period on spectral values, feature extraction is performed on features of various geographical features by comparing feature patterns.

5. The method for separating gaps in tea gardens using visible light multi-angle radiation from a UAV adapted to complex mountainous areas, as described in claim 4, is characterized in that... Based on the analysis results, the characteristics and differences of the spectral curves of various land cover at different time periods are determined, and a separation algorithm is designed, including: Determine the information about the land features based on the characteristics of the tea garden. Based on the results of the ground feature information judgment, a separation algorithm is designed to calculate the single-band image results.

6. The method for separating gaps in tea gardens using visible light multi-angle radiation from unmanned aerial vehicles (UAVs) adapted to complex mountainous areas, as described in claim 5, is characterized in that... Based on the analysis results, the characteristics and differences of the spectral curves of various land cover at different time periods are determined, and a separation algorithm is designed, including: T2 = min(b1, b2) - b3 T = T1 × T2 In this table, T represents the final single-band image result of the tea tree and tea garden gap separation calculation, T1 represents the calculation formula that highlights the tea tree and suppresses the tea garden gap, T2 represents the calculation formula for separating similar tea tree and tea garden gap, b1 represents the red band, b2 represents the green band, b3 represents the blue band, and k represents a very small constant.

7. The method for separating gaps in tea gardens using visible light multi-angle radiation from unmanned aerial vehicles (UAVs) adapted to complex mountainous areas, as described in claim 6, is characterized in that... The step of applying the separation algorithm to tea garden image data for calculation, and then using image segmentation to separate interstitial features in the tea garden and extract tea trees, includes: The calculation formula was substituted into the original image to calculate the single-band image with the result T1, and threshold segmentation was performed. According to the separation effect of the test, the tea tree was best separated from other land features when T1>3.725, that is, T1>3.725 represents tea tree, and T1≤3.725 represents gaps in tea garden. The value of T1≤3.725 was set to 0. The calculation formula is substituted into the original image to calculate the single-band image T2. Finally, the result of threshold segmentation of T1 is multiplied by T2 to obtain the single-band image T. Threshold segmentation is applied again. T>235.882 is set as tea trees, and T≤235.882 is set as tea garden gaps to obtain the actual tea garden harvesting area.

8. A UAV visible light multi-angle radiation tea garden gap separation system adapted to complex mountainous areas, using the method described in any one of claims 1-7, characterized in that, include: The acquisition module obtains true-color images of the tea garden, preprocesses the data, and obtains visible light orthophotos. The extraction module draws profile lines based on visible light orthophotos and extracts RGB spectrum curves of various objects at different times. The analysis module analyzes the spectral characteristics of ground features in the red, green, and blue bands based on the band curves of the ground features at different times. The calculation module, based on the analysis results, determines the characteristics and differences of the red, green and blue spectral curves of various objects at different times, and designs separation algorithm formulas; The output module takes the separation algorithm into the tea garden image data for calculation, and then uses threshold segmentation to separate the ground features in the gaps of the tea garden and extract the tea trees.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the method for separating gaps in tea gardens using visible light multi-angle radiation from a UAV adapted to complex mountainous areas as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the method for separating gaps in tea gardens using visible light multi-angle radiation from a UAV adapted to complex mountainous areas, as described in any one of claims 1 to 7.

Images