A soybean emergence rate identification method based on ridge crown length

Abstract

The application belongs to the technical field of image processing, and discloses a soybean emergence rate identification method based on ridge crown length, which comprises the following steps: step S1, collecting field image through a UAV, and processing the collected image data; step S2, performing feature extraction on the green vegetation area based on the processed field image data, and generating an initial vegetation binary image; and step S3, establishing an emergence rate calculation model to realize the calculation of the emergence rate. The soybean emergence rate identification method based on ridge crown length is used, the length percentage of the missing ridge crown is calculated to inversely deduce the emergence rate, thereby avoiding the dependence on single-plant feature identification, enabling the emergence rate calculation to be stably and low-computing-power realized in the growth stage where plants are blocked before and after and the crown layers overlap, and breaking through the limitations of the prior art on the growth period, image resolution and computing resources.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method for identifying soybean emergence rate based on the length of the canopy layer on the ridge. Background Technology

[0002] As an important oilseed and grain crop in my country, the emergence rate of soybeans directly determines the field population structure, affects subsequent growth and final yield, and is a core indicator that needs to be monitored throughout the entire soybean planting cycle.

[0003] With the rapid development of remote sensing, computer vision, and artificial intelligence technologies, image-based crop growth monitoring methods are gradually replacing traditional manual observation and becoming the mainstream means of intelligent agricultural monitoring. The core of these methods is to accurately identify and analyze crop growth status, growth stage, and key growth indicators by collecting crop images and combining them with relevant algorithms. Currently, soybean cultivation mostly adopts ridge cultivation, with techniques such as dense planting on wide ridges and three-row planting widely used. The growth status of the ridge canopy directly reflects the soybean emergence situation, providing an important entry point for the intelligent identification of soybean emergence rate.

[0004] However, current image-based soybean emergence rate identification and crop growth monitoring technologies still face many bottlenecks in practical field applications. Problems such as the limitation of the growth period window, high dependence on computing resources, and strict requirements for image acquisition make it difficult to meet the needs of large-scale, efficient, and low-cost soybean planting and production. This seriously restricts their application effect in actual soybean planting and production, and fails to adapt to the field characteristics of soybean ridge cultivation, thus limiting the popularization and implementation of precision agriculture technology in the soybean planting field.

[0005] Therefore, developing a technology that can overcome the above-mentioned bottlenecks, adapt to the field scenario of soybean ridge cultivation, and achieve efficient and low-cost soybean emergence rate identification based on the characteristics of the ridge canopy has become an urgent technical problem to be solved in the current field of precision agriculture and the intelligent development of soybean planting. Summary of the Invention

[0006] The purpose of this invention is to provide a soybean emergence rate identification method based on ridge canopy length. The continuous structure of the ridge canopy formed along the planting row direction is used as the emergence rate analysis object. The emergence rate is inferred by calculating the percentage of missing length of the ridge canopy, thereby avoiding dependence on individual plant feature identification. This allows for stable and low-computing emergence rate calculation even during growth stages where plants experience shading and canopy overlap, breaking through the limitations of existing technologies on growth period, image resolution, and computing resources.

[0007] To achieve the above objectives, the present invention provides a method for identifying soybean emergence rate based on ridge canopy length, comprising the following steps: Step S1: Collect field images using drones and process the collected image data; Step S2: Based on the processed field image data, extract features of green vegetation areas to generate an initial binary image of vegetation. Step S3: Establish a seedling emergence rate calculation model to calculate the seedling emergence rate.

[0008] Preferably, in step S1, field images are acquired using a drone, and the acquired image data is processed. The specific process is as follows: Step S11, Field Image Data Acquisition: Using a drone equipped with a visible light sensor, field RGB images are acquired at a flight altitude of 15-80m, with a forward overlap rate of 80% and a lateral overlap rate of 70%. The field RGB images are then stitched together using software to generate field image data for the study area. Step S12, Radiometric Consistency Correction: First, the RGB three channels of the field image data are normalized to grayscale, and the pixel values ​​are linearly mapped to a uniform grayscale range to eliminate radiometric differences between different image blocks. Step S13, Noise Suppression Processing: The corrected image is smoothed by filtering and combined with a bilateral filtering algorithm to smooth high-frequency noise while preserving vegetation edge information to ensure the integrity of the ridge structure. Step S14, Vegetation Enhancement Processing: Based on visible light RGB imagery, construct a visible light vegetation index to distinguish between vegetation and soil background.

[0009] Preferably, the Excess Green index is used to construct vegetation-enhanced images. As shown below: ; R, G, and B represent the red, green, and blue channel values ​​of the image, respectively; the Excess Green index is used to enhance the green vegetation response and suppress the soil background, thereby improving segmentation stability.

[0010] Preferably, in step S2, the green vegetation area is extracted from the processed field image data to generate an initial binary vegetation image. The specific process is as follows: Step S21, Color Space Conversion: Convert the enhanced RGB image to the HSV color space, use the hue components to express the green vegetation, and achieve preliminary separation of vegetation areas by setting the green hue range threshold. Step S22, Adaptive Threshold Segmentation: In the vegetation index image or HSV space, the Otsu algorithm is used for adaptive threshold segmentation to automatically determine the optimal segmentation threshold between the foreground and background, and generate an initial binary vegetation image.

[0011] Preferably, in step S3, the specific process for calculating the emergence rate is as follows: Step S31: Extract the center line of the ridge; Step S32: Quantify the continuity of rows to identify incomplete areas of rows; Step S33: Calculation of the incompleteness ratio; Step S34: Based on the defect ratio and the theoretical germination rate, establish a germination rate calculation model.

[0012] Preferably, in step S31, the specific process for extracting the center line of the ridge is as follows: First, the ridge direction angle is obtained, and the orientation of the binary vegetation image is corrected based on the ridge direction angle to make the ridges parallel to the main axis of the image. Then, a morphological thinning algorithm was used to skeletonize the vegetation area and extract the center line of the ridge with a width of one pixel. Finally, short, noisy skeletons were removed through connected component analysis, leaving only the row structure.

[0013] Preferably, in step S32, the specific process of quantizing the continuity of the row is as follows: A sliding window is established along the row direction to perform continuity analysis on the centerline; let the effective vegetation pixels per unit length be... The theoretical length of the monopoly is The continuity of a monopoly is defined as: ; When the continuity is lower than the set threshold, it is determined to be a missing seedling section, thus realizing the identification of missing areas in the ridge row.

[0014] Preferably, in step S33, the specific process for calculating the incompleteness ratio is as follows: Statistical analysis was performed on all ridges within the study area, and the effective length of the i-th ridge was set as follows: The theoretical total length is The overall incompleteness ratio is: ; in, The number of rows; This indicates the degree of incompleteness of the ridges in the field.

[0015] Preferably, in step S34, a seedling emergence rate calculation model is established based on the defect ratio and the theoretical seedling emergence rate. The specific process is as follows: Assuming a theoretical emergence rate of 100%, the actual emergence rate E is expressed as: .

[0016] Therefore, the present invention employs the above-mentioned method for identifying soybean emergence rate based on ridge canopy length, and the beneficial effects are as follows: (1) Breaking through the limitations of the growth period, applicable to the stage of plant shading: This invention is based on the analysis of the continuity of the ridge surface and the length of the missing parts, without relying on the identification of individual plant boundaries or individual plant features. It can still stably calculate the seedling rate when plants are shaded in front and behind or the canopy overlaps, thus overcoming the defect that the existing technology is only applicable to the early growth stage.

[0017] (2) The calculation process is simple and has low requirements for computing power and image conditions: the seedling rate is calculated by calculating the percentage of missing length of the field surface, avoiding complex feature extraction and deep learning reasoning processes. The amount of calculation is small and the requirements for computing resources and image resolution are low, making it suitable for large-scale field application.

[0018] (3) Highly consistent with agronomic structure, and the results are stable and reliable: This invention makes full use of the information on the ridge structure and planting row direction, and uniformly characterizes different seedling abnormalities such as missed sowing, broken strips, and dead seedlings as ridge continuity disruption. The resulting seedling rate results are in good consistency with the actual field seedling conditions.

[0019] (4) Wide range of applications: The method of the present invention can be implemented based on UAV imaging equipment, has strong adaptability to flight altitude and acquisition conditions, saves time and costs, has a large tolerance for flight altitude, and has good promotion and application value.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] Figure 1 This is a flowchart of a method for identifying soybean emergence rate based on ridge canopy length. Detailed Implementation

[0022] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] like Figure 1 As shown, a method for identifying soybean emergence rate based on ridge canopy length includes the following steps: Step S1: Collect field images using drones and process the collected image data; Step S2: Extract features of green vegetation areas from the processed field image data to generate an initial binary vegetation image. Step S3: Establish a seedling emergence rate calculation model to calculate the seedling emergence rate.

[0024] Example Step S1: Collect field images using drones and process the collected image data.

[0025] Step S11: Data acquisition of Daejeon images.

[0026] Field RGB images were collected using drones equipped with visible light sensors, with a flight altitude of 15-80m (adjusted according to time and cost), a forward overlap of 80%, and a lateral overlap of 70%. Field image data of the study area were then generated by stitching the images together using software.

[0027] Step S12, Radiation Consistency Correction.

[0028] After completing the RGB image stitching in the field and obtaining the orthophoto of the study area, this invention preprocesses the stitched images and extracts vegetation regions to improve the accuracy of vegetation identification. Since the stitched images are derived from multiple original images, there is a problem of inconsistent brightness. First, grayscale normalization is performed on the RGB three channels to linearly map the pixel values ​​to a uniform grayscale range, thereby eliminating the radiation differences between different image blocks.

[0029] Step S13: Noise suppression processing.

[0030] To reduce the impact of sensor noise and complex field background on subsequent identification, the corrected images are smoothed using a smoothing filter. A median filtering algorithm is preferred for noise removal, combined with a bilateral filtering algorithm to smooth high-frequency noise while preserving vegetation edge information, thus ensuring the integrity of the ridge structure.

[0031] Step S14: Vegetation enhancement treatment.

[0032] Since this invention is based on visible light RGB imagery, a visible light vegetation index is constructed to enhance green vegetation by improving the distinction between vegetation and soil background. Preferably, the Excess Green index is used to construct the vegetation-enhanced image. As shown below: ; R, G, and B represent the red, green, and blue channel values ​​of the image, respectively. This index can enhance the response of green vegetation and suppress soil background, thereby improving the stability of subsequent segmentation.

[0033] Step S2: Extract features of green vegetation areas from the processed field image data.

[0034] Step S21: Color space conversion.

[0035] To reduce the impact of lighting variations on the segmentation results, the enhanced RGB image was converted to the HSV color space, and the green vegetation was represented using hue components. By setting a threshold for the green hue range, preliminary separation of the vegetation areas was achieved.

[0036] Step S22: Adaptive threshold segmentation.

[0037] In the vegetation index image or HSV space, the Otsu algorithm is used for adaptive threshold segmentation to automatically determine the optimal segmentation threshold between the foreground and background, generating an initial binary vegetation image.

[0038] Step S3: Establish a seedling emergence rate calculation model to calculate the seedling emergence rate.

[0039] Step S31: Extract the center line of the ridge.

[0040] First, the ridge angle is obtained, and the orientation of the binary vegetation image is corrected based on the ridge angle to make the ridges parallel to the main axis of the image.

[0041] Then, a morphological thinning algorithm (Zhang-Suen thinning algorithm) was used to perform skeletonization processing on the vegetation area, and the center line of the ridge with a width of one pixel was extracted.

[0042] Finally, short, noisy skeletons were removed through connected component analysis, leaving only the main row structure.

[0043] Step S32: Quantify the continuity of rows to identify incomplete areas of rows.

[0044] A sliding window is established along the row direction to perform continuity analysis on the centerline; let the effective vegetation pixels per unit length be... The theoretical length of the monopoly is The continuity of a monopoly is defined as: ; When the continuity is lower than the set threshold, it is determined to be a missing seedling section, thus realizing the identification of missing areas in the ridge row.

[0045] Step S33: Calculate the incompleteness ratio.

[0046] Statistical analysis was performed on all ridges within the entire study area, and the effective length of the i-th ridge was set to 1. The theoretical total length is The overall incompleteness ratio is: ; in, The number of rows; This indicates the degree of incompleteness of the ridges in the field.

[0047] Step S34: Based on the defect ratio and the theoretical germination rate, establish a germination rate calculation model.

[0048] Assuming a theoretical emergence rate of 100%, the actual emergence rate E is expressed as: ; This method replaces individual plant identification with the integrity of the ridge structure, enabling rapid quantitative estimation of seedling emergence rate in large-scale fields.

[0049] Therefore, the present invention adopts the above-mentioned soybean emergence rate identification method based on ridge canopy length. The continuous structure of the ridge canopy formed along the planting row direction is used as the emergence rate analysis object. The emergence rate is inferred by calculating the percentage of missing length of the ridge canopy, thereby avoiding dependence on the identification of individual plant features. This allows for stable and low-computing emergence rate calculation even during the growth stages when plants are shading each other and the canopy overlaps. This breaks through the limitations of existing technologies on growth period, image resolution, and computing resources.

[0050] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for identifying soybean emergence rate based on ridge canopy length, characterized in that, Includes the following steps: Step S1: Collect field images using drones and process the collected image data; Step S2: Based on the processed field image data, extract features of green vegetation areas to generate an initial binary image of vegetation. Step S3: Establish a seedling emergence rate calculation model to calculate the seedling emergence rate.

2. The method for identifying soybean emergence rate based on ridge canopy length according to claim 1, characterized in that, In step S1, field images are collected using a drone, and the collected image data is processed. The specific process is as follows: Step S11, Field Image Data Acquisition: Using a drone equipped with a visible light sensor, field RGB images are acquired at a flight altitude of 15-80m, with a forward overlap rate of 80% and a lateral overlap rate of 70%. The field RGB images are then stitched together using software to generate field image data for the study area. Step S12, Radiometric Consistency Correction: First, the RGB three channels of the field image data are normalized to grayscale, and the pixel values ​​are linearly mapped to a uniform grayscale range to eliminate radiometric differences between different image blocks. Step S13, Noise Suppression Processing: The corrected image is smoothed by filtering and combined with a bilateral filtering algorithm to smooth high-frequency noise while preserving vegetation edge information to ensure the integrity of the ridge structure. Step S14, Vegetation Enhancement Processing: Based on visible light RGB imagery, construct a visible light vegetation index to distinguish between vegetation and soil background.

3. The method for identifying soybean emergence rate based on ridge canopy length according to claim 2, characterized in that, Vegetation-enhanced images were constructed using the Excess Green index. As shown below: ； R, G, and B represent the red, green, and blue channel values ​​of the image, respectively; the Excess Green index is used to enhance the green vegetation response and suppress the soil background, thereby improving segmentation stability.

4. The method for identifying soybean emergence rate based on ridge canopy length according to claim 2, characterized in that, In step S2, features of green vegetation areas are extracted from the processed field image data to generate an initial binary vegetation image. The specific process is as follows: Step S21, Color Space Conversion: Convert the enhanced RGB image to the HSV color space, use the hue components to express the green vegetation, and achieve preliminary separation of vegetation areas by setting the green hue range threshold. Step S22, Adaptive Threshold Segmentation: In the vegetation index image or HSV space, the Otsu algorithm is used for adaptive threshold segmentation to automatically determine the optimal segmentation threshold between the foreground and background, and generate an initial binary vegetation image.

5. The method for identifying soybean emergence rate based on ridge canopy length according to claim 4, characterized in that, In step S3, the specific process for calculating the emergence rate is as follows: Step S31: Extract the center line of the ridge; Step S32: Quantify the continuity of rows to identify incomplete areas of rows; Step S33: Calculation of the incompleteness ratio; Step S34: Based on the defect ratio and the theoretical germination rate, establish a germination rate calculation model.

6. The method for identifying soybean emergence rate based on ridge canopy length according to claim 5, characterized in that, In step S31, the specific process of extracting the center line of the ridge is as follows: First, the ridge direction angle is obtained, and the orientation of the binary vegetation image is corrected based on the ridge direction angle to make the ridges parallel to the main axis of the image. Then, a morphological thinning algorithm was used to skeletonize the vegetation area and extract the center line of the ridge with a width of one pixel. Finally, short, noisy skeletons were removed through connected component analysis, leaving only the row structure.

7. The method for identifying soybean emergence rate based on ridge canopy length according to claim 6, characterized in that, In step S32, the specific process of quantizing the continuity of the row is as follows: A sliding window is established along the row direction to perform continuity analysis on the centerline; let the effective vegetation pixels per unit length be... The theoretical length of the monopoly is The continuity of a monopoly is defined as: ； When the continuity is lower than the set threshold, it is determined to be a missing seedling section, thus realizing the identification of missing areas in the ridge row.

8. The method for identifying soybean emergence rate based on ridge canopy length according to claim 7, characterized in that, In step S33, the specific process for calculating the incompleteness ratio is as follows: Statistical analysis was performed on all ridges within the study area, and the effective length of the i-th ridge was set as follows: The theoretical total length is The overall incompleteness ratio is: ； in, The number of rows; This indicates the degree of incompleteness of the ridges in the field.

9. The method for identifying soybean emergence rate based on ridge canopy length according to claim 8, characterized in that, In step S34, a seedling emergence rate calculation model is established based on the defect ratio and the theoretical emergence rate. The specific process is as follows: Assuming a theoretical emergence rate of 100%, the actual emergence rate E is expressed as: 。

Images