A machine vision-based straw coverage recognition method

Abstract

This invention relates to the field of image analysis technology, specifically to a machine vision-based method for identifying straw coverage, comprising the following steps: acquiring a set of polarized images of farmland using a multi-angle polarization imaging device, the set of images including near-infrared band images under multiple polarization angles; calculating a specular reflection suppressed image based on the set of polarized images, calculating the disparity value of the suppressed image, and generating a straw-soil height difference marker map; and calculating the coverage based on the height difference marker map, where coverage = number of straw marker pixels / number of pixels in the exposed soil area × 100%. This invention provides a more stable and adaptable straw coverage calculation result, reducing errors under various interference scenarios (weeds, water film, broken straw), and providing a reliable basis for intelligent agricultural machinery sowing decisions.

Description

Technical Field

[0001] This invention relates to the field of image analysis technology, and in particular to a method for identifying straw coverage based on machine vision. Background Technology

[0002] Straw mulching is one of the important agronomic measures for conservation tillage and reducing soil wind and water erosion. Accurate monitoring of mulch coverage is of great significance for guiding no-till seeding and evaluating the quality of operations. Traditional methods for identifying straw mulch coverage mainly rely on the following technical approaches:

[0003] Manual photography and image annotation method: Operators use ordinary cameras to take images of the farmland surface and manually annotate areas of straw and soil to estimate the coverage. This method is time-consuming, labor-intensive, highly subjective, and difficult to adapt to the automated detection needs of large-scale farmland.

[0004] RGB or multispectral imaging methods: These methods classify and identify straw and soil in images based on color or spectral differences. However, under conditions of strong light changes, weed cover, or straw decay and fading, color and spectral features can easily become confused, leading to a significant decrease in recognition accuracy.

[0005] In addition, some studies have attempted to use polarization imaging to detect farmland surfaces, but most of them use a rotating polarizer and a single camera to acquire images. The image acquisition time difference is large (usually more than 200ms), which leads to serious image misalignment. Especially when acquiring images on mobile agricultural machinery, it is easy to produce blurring and spatial displacement, making it difficult to use for subsequent three-dimensional structure analysis.

[0006] In summary, existing methods generally suffer from the following problems: inconsistent imaging time and space, insufficient utilization of polarization data, difficulty in distinguishing between weed and straw structures, overly coarse definition of soil regions, and lack of a robust coverage calculation model. There is an urgent need for a new straw coverage identification method that integrates image consistency assurance, strong reflection suppression capability, accurate expression of structural features, and clear physical constraints of the calculation model. Summary of the Invention

[0007] This invention provides a machine vision-based method for identifying straw coverage.

[0008] A machine vision-based method for identifying straw cover includes the following steps:

[0009] S1, acquire a group of polarized images of farmland using a multi-angle polarization imaging device, the group of images including near-infrared band images under multiple polarization angles;

[0010] S2, calculate the specular reflection suppressed image based on the polarization image group, calculate the disparity value of the suppressed image, and generate a height difference marker map of straw and soil;

[0011] S3, calculate the coverage based on the height difference marking map, and the coverage = number of straw marking pixels / number of pixels in the soil exposed area × 100%;

[0012] In step S2, regions with disparity values ​​greater than a preset disparity threshold are marked as straw regions.

[0013] Optionally, S1 specifically includes:

[0014] S11 uses a beam splitter to divide the incident light path into three channels, and each channel is equipped with a multi-polarization angle linear polarizer.

[0015] S12, a narrowband filter is set in the near-infrared band;

[0016] S13 drives three CMOS sensors to acquire images simultaneously through a synchronous exposure controller, generating a group of spatially aligned polarization images.

[0017] Optionally, the band of the near-infrared image This enhances the difference in reflection between the straw and the background soil.

[0018] Optionally, multiple polarization angles, including 0°, 45°, and 90° polarization angles, can be used to fully capture surface reflection characteristics and calculate polarization angle differences.

[0019] A cubic beam splitter is used to separate the incident beam into three channels along the optical path. Each channel is equipped with a linear polarizer, and the polarization directions are as follows:

[0020] 0° (parallel polarization direction);

[0021] 45° (oblique polarization direction);

[0022] 90° (vertical polarization direction).

[0023] The beam splitter is coated with a near-infrared anti-reflection film on each surface, and its average transmittance meets the following requirements: To avoid inconsistent brightness in multi-channel images and eliminate the problems of inconsistent field of view and image misalignment caused by traditional rotating polarizers.

[0024] Polarization angles of 0°, 45°, and 90° can comprehensively capture surface reflection characteristics. 0° and 90° represent two orthogonal directions of polarized light, which helps to extract the difference between specular reflection and diffuse reflection. Adding a 45° direction can further capture the intermediate polarization characteristics caused by asymmetric surfaces or tilted structures (such as upright or rolled-up parts of straw). The combination of the three angles can be used to calculate polarization angle difference values ​​(such as maximum, minimum, contrast, etc.), improving the ability to distinguish reflection modes. Specular reflection is strongest at a specific polarization angle and can usually be suppressed by taking the minimum value (min) or degree of polarization (DoP) of images at different polarization angles. Using only two angles, 0° and 90°, may not be sufficient to cover reflections in certain directions. Adding 45° improves the ability to suppress irregular reflective areas, ensuring that "real structures" rather than "spectral artifacts" can be effectively identified even at different incident angles.

[0025] In the near-infrared narrowband filter settings, this includes adding a center wavelength of [missing value] to each channel. A narrowband filter with a wavelength of nm has a filtering range of: ;

[0026] That is, half-wave width nm was used to suppress visible light interference and enhance the difference in reflectance between straw and weeds in the near-infrared region (the reflectance of weeds increased sharply at 870 nm, and the interference was significant).

[0027] The sensor synchronous exposure control also includes driving three CMOS image sensors to expose simultaneously via an external synchronous exposure controller, with the exposure start time error satisfying the following: This ensures that the image acquisition process is almost seamless during operation of the high-speed mobile platform (agricultural machinery).

[0028] For example: vehicle speed km / h, at Displacement in milliseconds only:

[0029] .

[0030] The wavelength range (850nm±15nm) was selected based on the following criteria:

[0031] 1. Avoid visible light interference and improve imaging stability: The visible light band (about 400–700nm) is strongly interfered with by changes in sunlight intensity, leaf surface reflection and color differences, resulting in poor robustness of traditional RGB or broadband imaging in straw and soil identification; the near-infrared band (NIR), especially around 850nm, not only avoids these interferences, but also is suitable for the identification of texture and structural differences because of its relatively small light fluctuations and more uniform reflection under natural conditions.

[0032] 2. The characteristic window of the 850nm band in plant identification: In the spectral characteristics of plants, the main absorption band of chlorophyll is located in the red and blue light regions (approximately 430nm and 660nm), while the reflectance is significantly enhanced in the near-infrared region (700–900nm). However, as dead plant tissue, straw has degraded cell structure, and its reflectance at 850nm is much lower than that of fresh vegetation, while dry soil has stable reflectance in this band. Weeds (especially living leaves) have a sharp increase in reflectance near 870nm (up to more than 30%), and if the filter band is too wide, such interference signals will be introduced.

[0033] 3. Measured spectral data show:

[0034] Soil reflectance: stable at 800–900 nm, with a reflectance of approximately 20–30%;

[0035] Straw reflectance: significantly higher than soil at 850 nm (difference of about 10–20%), but much lower than fresh green plants;

[0036] Weed reflectance: It jumps rapidly at 870–880nm (up to 30% compared to 850nm), which can be significantly misjudged as a high-contrast area of ​​"straw".

[0037] This invention enhances the reflection difference between straw and background (soil) while avoiding interference from the "climbing zone" (860–900nm) of live plant reflection. The high reflectivity zone of live weeds is exactly in the 870–900nm range. This invention controls the bandwidth at 850±15nm, effectively avoiding the introduction of high reflectivity values ​​above 870nm into polarization image calculations, and ensuring that the specular suppression image will not be misjudged due to highly reflective weeds.

[0038] By limiting the wavelength to 850±15nm, the unstructured reflection effects of different ground objects are effectively reduced, the texture consistency between polarization angle images is improved, and the reflection differences are mainly caused by the three-dimensional structure of straw, rather than spectral disturbances.

[0039] Optionally, S2 specifically includes:

[0040] S21. Perform pixel-level registration on near-infrared band images under polarization angles, establish spatial coordinate mapping relationships, and ensure that each polarization angle image reflects the same physical location at the same coordinates.

[0041] S22, calculate the polarization angle difference value for each pixel;

[0042] S23, Construct a specular reflection suppression image: Take the minimum value of the polarization angle difference of all pixels as the benchmark to generate a normalized specular reflection suppression image:

[0043] S24, Perform binocular disparity calculation on the specular reflection suppression image to generate a disparity map;

[0044] S25, convert the disparity map into a height difference marker map;

[0045] S26, mark the continuous area with a height difference > 0.5 mm in the height difference marking map as the straw area.

[0046] Optionally, the binocular parallax calculation includes using a 45° polarized image as a reference view and employing a semi-global matching algorithm to calculate the parallax for each view. and The image is stereo matched to output two disparity maps, which are then averaged and fused to generate the final disparity map.

[0047] Optionally, the polarization angle difference value is calculated as follows:

[0048] ;in, Indicates the polarization angle is Image in coordinates grayscale value at that location This represents a graph showing the difference in polarization angles, used to characterize the heterogeneity of reflection angles.

[0049] Optionally, the generation of the height difference marker map includes converting the parallax values ​​into the actual physical height difference, represented as follows: ,in, The baseline distance is the distance between the centers of the two imaging channels. For the focal length of the imaging lens, This represents the physical height difference at coordinates (x, y) on the height difference map. Let be the disparity value at coordinates (x, y).

[0050] Optionally, S3 specifically includes:

[0051] S31, perform morphological closing operation on the height difference marker map. The structuring element of the closing operation is a circular kernel that connects the broken straw regions.

[0052] S32, extract connected regions with a height difference > 0.5mm as straw regions, and count the total number of pixels. ;

[0053] S33, extract continuous areas with a height difference ≤ 0.2 mm as soil exposure areas;

[0054] S34, Count the total number of pixels in the soil exposed area. ;

[0055] S35, calculate coverage: .

[0056] Optionally, the soil exposure area also satisfies:

[0057] Area ≥ 100cm 2 This is equivalent to an image pixel count of ≥500 pixels;

[0058] The standard deviation of the height difference within the area is ≤0.05mm.

[0059] The beneficial effects of this invention are:

[0060] This invention employs a beam splitter prism + three-channel synchronous exposure architecture, combined with fixed 0° / 45° / 90° polarizers and a narrowband filter with a center wavelength of 850nm±15nm. This effectively avoids the image misalignment problem caused by rotating polarizers in traditional single-sensor systems. At the same time, it filters out the spectral noise caused by the surge in reflection from living weeds at 870nm. Furthermore, the specular reflection suppression image constructed based on the polarization angle difference enhances the representation of optical structural differences between straw and soil, enabling significant structural texture differences to be formed in the straw area without contact, providing ideal input for subsequent height analysis.

[0061] This invention applies a semi-global matching algorithm guided by polarization maps to identify farmland straw. By using a 45° polarization map as a reference view and combining 0° and 90° views for disparity calculation, effective straw disparity information can still be extracted in low-texture, partially occluded, or overlapping areas. Combined with the actual calibrated baseline distance and focal length parameters, the generated height difference marker map has significant distinguishing ability at the thickness level of rotten straw, and its structural texture performance is superior to traditional RGB or NIR depth mapping methods.

[0062] This invention designs a soil exposure area identification mechanism based on three physical constraints: height difference threshold, area, and flatness standard deviation. This effectively avoids the problem of traditional methods misidentifying stones, ruts, and shadows as bare soil. Specifically, it is effective for areas with a height difference ≤ 0.2 mm and an area ≥ 100 cm². 2 Furthermore, areas with a standard deviation ≤0.05mm were selected to ensure that the extracted soil areas have greater surface consistency and physical rationality. The final straw coverage calculation results have stronger stability and field adaptability, reducing errors under various interference scenarios (weeds, water film, broken straw), and providing a reliable basis for intelligent agricultural machinery sowing decisions. Attached Figure Description

[0063] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1This is a schematic diagram of the method flow according to an embodiment of the present invention;

[0065] Figure 2 This is a schematic diagram of the coverage calculation process in an embodiment of the present invention. Detailed Implementation

[0066] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. For some well-known technologies, those skilled in the art may also use other alternative methods to implement the invention. Moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0067] like Figures 1-2 As shown, a machine vision-based method for identifying straw coverage includes the following steps:

[0068] S1, acquires a group of polarized images of farmland using a multi-angle polarization imaging device. The group of images includes near-infrared images under multiple polarization angles.

[0069] S2, calculate the specular reflection suppressed image based on the polarization image group, calculate the disparity value of the suppressed image, and generate a height difference marker map between straw and soil;

[0070] S3. Calculate the coverage based on the height difference marker map. Coverage = (Number of straw marker pixels / Number of pixels in exposed soil area) × 100%;

[0071] In S2, regions with disparity values ​​greater than a preset disparity threshold are marked as straw regions.

[0072] S1 specifically includes:

[0073] S11, Beam-splitting polarization imaging channel design: A cubic beam-splitting prism is used to separate the incident beam into three channels according to the optical path direction. Each channel is equipped with a linear polarizer, and the polarization directions are as follows:

[0074] 0° (parallel polarization direction);

[0075] 45° (oblique polarization direction);

[0076] 90° (vertical polarization direction).

[0077] The beam splitter is coated with a near-infrared anti-reflection film on each surface, and its average transmittance meets the following requirements: To avoid inconsistent brightness in multi-channel images and eliminate the problems of inconsistent field of view and image misalignment caused by traditional rotating polarizers.

[0078] Polarization angles of 0°, 45°, and 90° can comprehensively capture surface reflection characteristics. 0° and 90° represent two orthogonal directions of polarized light, which helps to extract the difference between specular reflection and diffuse reflection. Adding a 45° direction can further capture the intermediate polarization characteristics caused by asymmetric surfaces or tilted structures (such as upright or rolled-up parts of straw). The combination of the three angles can be used to calculate polarization angle difference values ​​(such as maximum, minimum, contrast, etc.), improving the ability to distinguish reflection modes. Specular reflection is strongest at a specific polarization angle and can usually be suppressed by taking the minimum value (min) or degree of polarization (DoP) of images at different polarization angles. Using only two angles, 0° and 90°, may not be sufficient to cover reflections in certain directions. Adding 45° improves the ability to suppress irregular reflective areas, ensuring that "real structures" rather than "spectral artifacts" can be effectively identified even at different incident angles.

[0079] S12, Near-infrared narrowband filter setting: Add a center wavelength of [value missing] to each channel. A narrowband filter with a wavelength of nm has a filtering range of: ;

[0080] That is, half-wave width nm was used to suppress visible light interference and enhance the difference in reflectance between straw and weeds in the near-infrared region (the reflectance of weeds increased sharply at 870 nm, and the interference was significant).

[0081] S13, Sensor Synchronous Exposure Control: Three CMOS image sensors are simultaneously exposed via an external synchronous exposure controller, with the exposure start time error meeting the following requirements: This ensures that the image acquisition process is almost seamless during operation of the high-speed mobile platform (agricultural machinery).

[0082] For example: vehicle speed km / h, at Displacement in milliseconds only:

[0083] .

[0084] The wavelength range (850nm±15nm) was selected based on the following criteria:

[0085] 1. Avoid visible light interference and improve imaging stability: The visible light band (about 400–700nm) is strongly interfered with by changes in sunlight intensity, leaf surface reflection and color differences, resulting in poor robustness of traditional RGB or broadband imaging in straw and soil identification; the near-infrared band (NIR), especially around 850nm, not only avoids these interferences, but also is suitable for the identification of texture and structural differences because of its relatively small light fluctuations and more uniform reflection under natural conditions.

[0086] 2. The characteristic window of the 850nm band in plant identification: In the spectral characteristics of plants, the main absorption band of chlorophyll is located in the red and blue light regions (approximately 430nm and 660nm), while the reflectance is significantly enhanced in the near-infrared region (700–900nm). However, as dead plant tissue, straw has degraded cell structure, and its reflectance at 850nm is much lower than that of fresh vegetation, while dry soil has stable reflectance in this band. Weeds (especially living leaves) have a sharp increase in reflectance near 870nm (up to more than 30%), and if the filter band is too wide, such interference signals will be introduced.

[0087] 3. Measured spectral data show:

[0088] Soil reflectance: stable at 800–900 nm, with a reflectance of approximately 20–30%;

[0089] Straw reflectance: significantly higher than soil at 850 nm (difference of about 10–20%), but much lower than fresh green plants;

[0090] Weed reflectance: It jumps rapidly at 870–880nm (up to 30% compared to 850nm), which can be significantly misjudged as a high-contrast area of ​​"straw".

[0091] This invention enhances the reflection difference between straw and background (soil) while avoiding interference from the "climbing zone" (860–900nm) of live plant reflection. The high reflectivity zone of live weeds is exactly in the 870–900nm range. This invention controls the bandwidth at 850±15nm, effectively avoiding the introduction of high reflectivity values ​​above 870nm into polarization image calculations, and ensuring that the specular suppression image will not be misjudged due to highly reflective weeds.

[0092] By limiting the wavelength to 850±15nm, the unstructured reflection effects of different ground objects are effectively reduced, the texture consistency between polarization angle images is improved, and the reflection differences are mainly caused by the three-dimensional structure of straw, rather than spectral disturbances.

[0093] S2 specifically includes:

[0094] S21, Pixel-level registration processing: Pixel-level registration is performed on near-infrared images at polarization angles of 0°, 45°, and 90° to establish a unified spatial coordinate mapping relationship, ensuring that images at each polarization angle are on the same coordinate system. The locations reflect the same physical location.

[0095] S22, Polarization angle difference calculation: For each pixel, calculate the polarization angle difference value:

[0096] ;in, Indicates the polarization angle is Image in coordinates grayscale value at that location This represents a graph showing the difference in polarization angles, used to characterize the heterogeneity of reflection angles.

[0097] S23, Construction of Specular Reflection Suppression Image: Constructing a normalized specular reflection suppression image based on the polarization angle difference map:

[0098] ;

[0099] in, ; represents the minimum difference value among all pixels in the entire image; , This indicates the grayscale value of the image with specular reflection suppression, ranging from 0 to 255, and the dynamic range. ;

[0100] This is used to eliminate the interference of changes in light intensity under different shooting conditions on image differences and improve the salience of straw structure.

[0101] S24, Binocular disparity calculation: (Based on...) Polarization image as a reference view;

[0102] The semi-global matching (SGM) algorithm was used to respectively... and Stereo matching of images;

[0103] Output two disparity maps , The final disparity map is generated by averaging and fusion. .

[0104] S25, Height Difference Marker Generation: Convert pixel parallax values ​​to actual physical height differences using the following formula: ,in, The baseline distance is the center-to-center distance between the two imaging channels, ranging from 20 to 30 cm. This refers to the focal length of the imaging lens, with a range of 8–12 mm. This represents the physical height difference at coordinates (x, y) on the height difference map, in millimeters. This represents the disparity value at coordinates (x, y), in pixels.

[0105] S26, Straw Area Marking: Mark continuous areas in the height difference map that meet the following conditions as straw areas: If mm Straw area; where the height threshold of 0.5 mm is a robust segmentation standard set based on 382 measured samples of rotten straw thickness, combined with the standard deviation (0.12 mm).

[0106] S3 specifically includes:

[0107] S31, Morphological Closure Operation Processing: Perform morphological closure operation on the height difference marker map to eliminate pinholes and connect broken areas, as follows:

[0108] The structural element core adopts a circular morphological core with a size of 5×5;

[0109] The operation sequence is: expansion → corrosion, to connect the broken straw areas.

[0110] The core size is based on the measured average width of straw breakage of 4.2mm, which corresponds to a pixel diameter of approximately 5 pixels under the condition of adapting spatial resolution.

[0111] S32, Extracting Straw Regions: Extract connected regions from the closing operation results that satisfy the following conditions: The total number of pixels in this area is:

[0112] ;in, This represents the value of a pixel in the height difference map. This represents the set of connected regions with a height difference greater than 0.5 mm, i.e., the straw region. This indicates the number of pixels marked on the straw.

[0113] S33, Extracting Exposed Soil Areas: The criteria for determining soil area extraction are defined as follows:

[0114] ;

[0115] in, Indicates candidate low-altitude regions. Represents the area, in pixels. Indicates the region Standard deviation of internal height difference.

[0116] S34, Count the total number of pixels in the soil exposed area: Sum the total number of pixels in all areas that meet the above three conditions to obtain: ;in, This indicates that the height difference is satisfied. mm, area Pixels, standard deviation mm of region set, This indicates the number of pixels in the exposed soil area.

[0117] S35, the final coverage calculation formula is as follows: .

[0118] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0119] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A machine vision-based method for identifying straw coverage, characterized in that, Includes the following steps: S1, acquire a group of polarized images of farmland using a multi-angle polarization imaging device, the group of images including near-infrared band images under multiple polarization angles; S2, calculate the specular reflection suppressed image based on the polarization image group, calculate the disparity value of the suppressed image, and generate a height difference marker map of straw and soil; S3, calculate the coverage based on the height difference marking map, and the coverage = number of straw marking pixels / number of pixels in the soil exposed area × 100%; In step S2, regions with parallax values ​​greater than a preset difference threshold are marked as straw regions; S2 specifically includes: S21. Perform pixel-level registration on near-infrared band images under polarization angles, establish spatial coordinate mapping relationships, and ensure that each polarization angle image reflects the same physical location at the same coordinates. S22, calculate the polarization angle difference value for each pixel; S23, Construct a specular reflection suppression image: Take the minimum value of the polarization angle difference of all pixels as the benchmark to generate a normalized specular reflection suppression image: S24, Perform binocular disparity calculation on the specular reflection suppression image to generate a disparity map; S25, convert the disparity map into a height difference marker map; S26, mark the continuous area with a height difference > 0.5 mm in the height difference marking map as the straw area.

2. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, S1 specifically includes: S11 uses a beam splitter to divide the incident light path into three channels, and each channel is equipped with a multi-polarization angle linear polarizer. S12, a narrowband filter is set in the near-infrared band; S13 drives three CMOS sensors to acquire images simultaneously through a synchronous exposure controller, generating a group of spatially aligned polarization images.

3. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, The band of the near-infrared image This enhances the difference in reflection between the straw and the background soil.

4. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, The multiple polarization angles include 0°, 45°, and 90° polarization angles, in order to comprehensively obtain the surface reflection characteristics and to calculate the polarization angle difference value.

5. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, The binocular parallax calculation includes using a 45° polarized image as the reference view and employing a semi-global matching algorithm to calculate the binocular parallax separately. and The image is stereo matched to output two disparity maps, which are then averaged and fused to generate the final disparity map.

6. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, The polarization angle difference value is calculated as follows: ;in, Indicates the polarization angle is Image in coordinates grayscale value at that location This represents a graph showing the difference in polarization angles, used to characterize the heterogeneity of reflection angles.

7. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, The generation of the height difference marker map includes converting the parallax values ​​into the actual physical height difference, which is represented as follows: ,in, The baseline distance is the distance between the centers of the two imaging channels. For the focal length of the imaging lens, This represents the physical height difference at coordinates (x, y) on the height difference map. Let be the disparity value at coordinates (x, y).

8. The method for identifying straw coverage based on machine vision according to claim 1, characterized in that, S3 specifically includes: S31, perform morphological closing operation on the height difference marker map. The structuring element of the closing operation is a circular kernel that connects the broken straw regions. S32, extract connected regions with a height difference > 0.5mm as straw regions, and count the total number of pixels. ; S33, extract continuous areas with a height difference ≤ 0.2 mm as soil exposure areas; S34, Count the total number of pixels in the soil exposed area. ; S35, calculate coverage: .

9. The method for identifying straw coverage based on machine vision according to claim 8, characterized in that, The soil exposure area also meets the following requirements: The area is ≥100cm², which is equivalent to an image pixel count ≥500 pixels; The standard deviation of the height difference within the area is ≤0.05mm.

Images