Corncob position height measuring method and device based on depth camera and laser radar three-dimensional fusion

Abstract

The invention discloses a corn ear height measuring method and device based on three-dimensional fusion of a depth camera and a laser radar. The measuring method comprises the following steps: acquiring an RGB image and a depth image containing a measured corn ear through a depth camera; the reference height is acquired through the reference acquisition module. A pitch angle and a roll angle of the depth camera are acquired through the inertial measurement unit. And according to the position of the height identification point in the image, obtaining a vertical direction included angle of the height identification point relative to the depth camera. And obtaining the relative height of the corn ear according to the included angle in the vertical direction and the depth value of the corresponding height identification point in the depth image. And fusing the corrected relative height with the reference height acquired by the reference acquisition module to obtain the corn ear height. The pitch angle and the roll angle of the depth camera are collected, and the depth value and the vertical direction included angle corresponding to the height identification point are corrected by using the pitch angle and the roll angle, so that the accuracy of remote measurement of the relative height of the corn ear is further improved.

Description

Technical Field

[0001] The present invention relates to the field of agricultural intelligent technology, and in particular to a method and device for measuring the height of corn ears based on the stereo fusion of a depth camera and a laser radar. Background Art

[0002] Corn is one of the most widely grown food crops worldwide. Its growth status, yield, and quality are not only directly related to agricultural economic benefits, but also affect food security and the stability of the supply chain. Corn ear height refers to the vertical distance from the ear on the plant to the ground. It is an important indicator for measuring corn growth status and yield potential. In terms of variety selection and genetic improvement, ear height directly affects the plant's lodging resistance. Excessively high ear height will cause the plant's center of gravity to shift upward, increasing the risk of lodging, which in turn affects yield. Therefore, in the breeding process, measuring and analyzing ear height can help screen varieties with strong lodging resistance and optimize plant type to give them stronger stem support and better root development to adapt to dense planting and high-yield planting patterns. In terms of yield prediction and model construction, ear height is often used as an important parameter to measure plant biomass and photosynthetic capacity, and is widely used in crop growth models and phenotypic research. Under different ecological conditions, ear height also affects planting strategies. For example, windy areas are suitable for promoting low-ear varieties to enhance lodging resistance, while areas with sufficient sunlight can choose higher-ear varieties to improve photosynthetic efficiency and yield.

[0003] Currently, there are many methods for measuring corn ear height. Traditional manual measurement methods require using tools such as rulers and tape measures to measure and manually record ear height along the corn plant. While this method is simple, intuitive, and easy to implement, it has disadvantages such as high time cost, labor intensity, and unsuitability for large-scale application. Laser rangefinders are commonly used for rapid measurement of crops across large fields and offer high accuracy, but they are significantly affected by the environment and crop density. Image technology has also been applied to measuring ear height. This method captures an image of a graduated measuring rod and uses image recognition to automatically calculate ear height. While efficient, this method requires the rod to be placed vertically, and rod offset can introduce errors. These methods all consider the vertical direction from the ground. Similarly, many methods calculate ear height by acquiring horizontal depth information. For example, a binocular camera is used to capture images from different angles, calculating parallax to obtain depth information, and then estimating the height of the measured object from the ground. However, the farther the object is from the camera, the smaller the parallax, and the closer the object, the larger the parallax. Although it can calculate the height of the measured object, it requires high camera calibration. In areas with dense vegetation, depth confusion may occur, resulting in large errors in depth calculation, which in turn affects the height estimation of the measured object. Emerging 3D reconstruction technology reconstructs the scene from a 3D point cloud and uses the 3D coordinates from the ground to the measured object to obtain the height. However, this method requires ground reference points for 3D point cloud reconstruction. In complex environments, such as the interior of corn fields, it is difficult to capture the ground. Therefore, this technology is mainly applicable to the outer areas of corn fields and has limitations in measuring the height of corn ears in the inner areas.

[0004] With the rapid development of intelligent agricultural technology, sensor technologies such as depth cameras and lidar are gradually being applied to crop monitoring. In order to solve these problems, the present invention proposes a method for measuring the height of corn ears based on the stereo fusion of depth cameras and lidar. By stereo fusion of the ranging height in the vertical direction of the ground and the depth information in the horizontal direction, the advantages of the two sensors are combined to overcome the limitations of existing measurement methods and realize the automation and accurate measurement of the height of corn ears. At the same time, the method has strong adaptability and can work stably under different environmental conditions. It is suitable for measuring the ear position of the inner and outer layers of corn planting areas, meets the needs of modern agriculture for efficient and accurate measurement technology, and provides important support for large-scale corn production management. Summary of the Invention

[0005] The present invention provides a method and device for measuring the height of corn ears based on the stereo fusion of a depth camera and a lidar. The method aims to solve the technical difficulties of existing measurement methods, such as being time-consuming, labor-intensive, inefficient, difficult to measure inside corn-growing areas, and easily affected by factors such as lighting and occlusion in complex field environments, making them incapable of adapting to large-scale corn ear measurement. The method thus realizes rapid, accurate, and automated measurement of the height of corn ears.

[0006] In a first aspect, the present invention provides a method for measuring the height of corn ears based on the stereo fusion of a depth camera and a laser radar; the process is as follows: The depth camera captures RGB and depth images of the corn ear under test, the reference height is acquired through the reference acquisition module, and the pitch and roll angles of the depth camera are acquired through the inertial measurement unit.

[0007] The position of the measured corn ear in the RGB image is identified by the corn ear recognition model as a height recognition point.

[0008] According to the position of the height recognition point in the RGB image, the vertical angle of the height recognition point relative to the depth camera is obtained.

[0009] The relative height of the corn ear is obtained according to the vertical angle α and the depth value of the corresponding height identification point in the depth image.

[0010] The pitch and roll angles are used to correct the relative height of the corn ear to eliminate the error caused by the depth camera posture.

[0011] The corrected relative height is fused with the reference height collected by the reference collection module to obtain the height of the corn ear.

[0012] As a preference, the relative height of the corn ear before correction H s The expression is: H s =dis tanα Where dis is the vertical angle of the depth camera, and α is the vertical angle of the height recognition point relative to the depth camera.

[0013] Preferably, the expression of the vertical angle α is as follows: α=arctan( ) in, is the ratio of the vertical component; VFOV is the vertical field of view of the camera.

[0014] Vertical component ratio The expression is: ; in, The ordinate of the target recognition point in the pixel coordinate system with the center point of the RGB image as the origin; W The pixel width of the RGB image.

[0015] As a preferred method, the process of correcting the relative height of the corn ear is as follows: The pitch angle collected by the inertial measurement unit is used to correct the depth value of the height identification point.

[0016] The vertical angle is corrected using the roll angle collected by the inertial measurement unit.

[0017] The relative height of the corn ear is recalculated using the corrected depth value of the height identification point and the vertical angle.

[0018] Preferably, the expression for correcting the depth value of the height identification point is as follows: ; Where dis and dis' are the depth values ​​before and after correction, respectively; θ is the pitch angle collected by the inertial measurement unit.

[0019] As a preferred method, the process of correcting the vertical angle is as follows: Establish the corrected vertical component ratio The expression is as follows: ; in, a is the horizontal coordinate of the target recognition point in the pixel coordinate system with the center point of the RGB image as the origin; Roll angle collected by the inertial measurement unit.

[0020] Using the corrected vertical component ratio Calculate the corrected vertical angle.

[0021] Preferably, the fused height of the corn ear is obtained by superimposing the relative height of the corn ear, the reference height collected by the reference collection module, and the height difference between the reference collection module and the depth camera.

[0022] In a second aspect, the present invention provides a device for measuring the height of an ear of corn based on the stereoscopic fusion of a depth camera and a laser radar, which is used in the aforementioned method. The device comprises a housing, a depth camera, a reference acquisition module, a control module, a display module, and a power supply module. The housing comprises a main body and a bottom extension. The depth camera, control module, and display module are all mounted on the main body. The reference acquisition module is mounted on the bottom extension and is arranged in a vertical orientation.

[0023] Preferably, the reference acquisition module uses a laser radar to obtain the reference height by collecting the distance from itself to the ground.

[0024] Preferably, the control module is provided with a corn ear recognition model constructed based on the YOLOv5 model.

[0025] The present invention has the following beneficial effects.

[0026] 1. Data Fusion Improves Measurement Accuracy: This system combines a depth camera with a LiDAR sensor, leveraging the strengths of both. The depth camera provides highly accurate horizontal depth information, while the LiDAR provides highly accurate vertical reference height information. This fusion of the two enables the system to maintain high measurement accuracy even in complex environments, avoiding the errors inherent in a single sensor due to environmental fluctuations.

[0027] 2. Image acquisition error correction: The present invention collects the pitch angle and roll angle of the depth camera, and uses the pitch angle and roll angle to correct the depth value and vertical angle corresponding to the height identification point, thereby further improving the accuracy of long-distance measurement of the relative height of the corn ear.

[0028] 3. Strong Adaptability: This system operates stably in diverse environmental conditions and requires no ground reference points. Whether measuring in the outer reaches of corn fields or within denser, inner corn fields, the system provides accurate and stable measurements, effectively addressing common measurement challenges in agricultural production.

[0029] 4. Automated Measurement: This invention automates the measurement of corn ear height, eliminating the need for traditional manual measurement and reducing the complexity of manual operation. This automated process not only improves work efficiency but also reduces the potential for errors caused by manual operation, significantly enhancing measurement accuracy and reliability.

[0030] 5. Portability: The measurement device provided by this invention features a compact design, lightweight, long battery life, and excellent portability. Users can conveniently carry it to various measurement sites, making it particularly suitable for applications in large-scale farmland and complex terrain. Its portable design allows for flexible use in various locations, enhancing its ease of use in agricultural production. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a structural block diagram of embodiment 1 of the present invention.

[0032] Figure 2 This is a schematic structural diagram of the housing in Example 1 of the present invention.

[0033] Figure 3 This is a flow chart of Example 2 of the present invention.

[0034] Figure 4 This is a schematic diagram of the principle of calculating height recognition points in Example 2 of the present invention.

[0035] Figure 5 This is a schematic diagram showing the principle of height correction of the pitch angle and roll angle in Example 2 of the present invention. DETAILED DESCRIPTION

[0036] The present invention will be further described below with reference to the accompanying drawings.

[0037] Example 1 like Figure 1 and Figure 2 As shown, a corn ear height measurement device based on the stereo fusion of a depth camera and a laser radar includes a housing, a depth camera, a reference acquisition module, a control module, a display module, and a power supply module. The housing includes a main body 1 and a bottom extension 2.

[0038] The main body of the housing is provided with the following structure: Battery fixing area: A 10000mAH battery is installed as a power supply module to provide battery life support for the Raspberry Pi and ensure that the device can work stably for a long time in the field.

[0039] Raspberry Pi fixed area: A Raspberry Pi, serving as the control module, is installed. The Raspberry Pi is connected to the display module to form the core control and display system, responsible for coordinating various components, data processing, and YOLOv5 model deployment, performing depth camera data fusion, target detection, and measurement calculation tasks.

[0040] Charging port: connected to the power supply module for charging.

[0041] Data transmission interface: connected to the control module for data transmission.

[0042] Control button: The user can manually trigger two functions through the control button: (1) collect RGB depth image data, and (2) start corn ear position recognition and height measurement.

[0043] Camera Installation Location: A depth camera is installed. The depth camera is positioned horizontally. The depth camera is used to capture real-time color and depth images of the corn ear, which are used for YoloV5 model recognition and horizontal data acquisition for stereo fusion, respectively. In this embodiment, the depth camera uses an Intel RealSense Depth Camera D435i.

[0044] The Intel RealSense Depth Camera D435i is a powerful depth camera consisting of an RGB camera, two infrared cameras, and an infrared emitter. Its depth imaging principle is based on active stereo infrared imaging technology. This imaging method uses infrared light emitted by the infrared emitter to illuminate an object. The two infrared cameras capture the infrared light signals reflected from the object's surface and calculate the depth information of each pixel. Unlike traditional monocular cameras, the RGB and infrared cameras are physically located at different positions, resulting in parallax between the color image and the depth image. This means that the position of the same object may differ between the color and depth images. To accurately align the depth information with the color image, image stream alignment is required. By precisely spatially aligning the RGB and depth image streams, each pixel in the depth image corresponds to the corresponding pixel in the color image. This process is implemented using the camera's built-in alignedFs function. The alignment method aligns depth with color, leaving the color image unchanged and transforming the depth image. This ensures that the depth information is correctly matched to the object features in the color image, providing accurate data support for subsequent image analysis and object detection.

[0045] The Intel RealSense Depth Camera D435i has a 42° vertical field of view and a 69° horizontal field of view, and can store both color and depth images. The D435i stores RGB frames at a resolution of 1280×720, and each pixel in the stored depth image contains a depth value in meters. The depth value of a depth camera is defined as the perpendicular distance from a point in space to the camera's optical axis plane. This refers to the coordinate value of the point in the camera's coordinate system along the optical axis (Z axis), which reflects the perpendicular distance from the object's surface to the camera's plane. The depth camera is equipped with an inertial measurement unit capable of detecting posture information.

[0046] The reference acquisition module is mounted at the bottom of the bottom extension, with the detection portion facing vertically downward. In this embodiment, the reference acquisition module utilizes a laser radar, which is connected to a Raspberry Pi via a TTL-to-USB interface. The laser radar measures the height of the corn ear height measurement device from the ground, serving as vertical reference data for stereo fusion.

[0047] In this embodiment, the control module uses a Raspberry Pi 4B motherboard; the display module uses a Weishi Electronics 7-inch display; and the power supply module uses a Raspberry Pi-specific battery. The Raspberry Pi 4B serves as the control core, and power is provided by a Raspberry Pi-specific battery. The depth camera is connected to the Raspberry Pi via a USB interface and can capture color and depth images of corn ears in real time; the display is connected to the Raspberry Pi via DuPont cables and a flat cable for real-time display of collected data and system status. Two buttons are connected to the Raspberry Pi for manually triggering data acquisition and manually triggering corn ear recognition and height measurement. The lidar communicates with the Raspberry Pi via a serial port.

[0048] The LiDAR communicates with the Raspberry Pi via a TTL-to-USB interface, fully adhering to a specific communication protocol. When a button is pressed, a high-level signal is triggered, and the Raspberry Pi then sends a single measurement command, "ADDR0602CS," to the LiDAR rangefinder's port. Upon receiving the command, the LiDAR rangefinder performs a range measurement and returns the result. The Raspberry Pi continues sending this command until it receives a valid return code to ensure successful data acquisition. The LiDAR rangefinder's serial communication is configured at a baud rate of 9600 bps, 8 data bits, and returns data using standard ASCII encoding. For example, the raw data returned by the rangefinder is "3132332E343536," which represents ASCII characters in hexadecimal format. Every two hexadecimal characters correspond to one ASCII character. Specifically, "31" corresponds to the character '1,' "32" corresponds to the character '2,' "2E" corresponds to '.', and so on. The Raspberry Pi parses this hexadecimal data and converts it into the corresponding ASCII string, "123.456." Next, the Raspberry Pi converts the string into a floating-point number using the programming language's parsing function. This process ultimately yields the measurement value of 123.456 meters.

[0049] Example 2 like Figure 3 As shown, a method for measuring the height of corn ears based on the stereo fusion of a depth camera and a lidar uses the corn ear height measuring device provided in Example 1.

[0050] The method for measuring the height of corn ears comprises the following steps: Step 1: Establish a corn ear recognition model for identifying the location of corn ears in an RGB image. In this embodiment, the corn ear recognition model is built and trained based on the YOLOv5 model. The corn ear recognition model returns an array containing multiple target boxes for the RGB image. Each target box in the array consists of four values; these four values ​​represent the coordinates of the top-left and bottom-right corners of the recognition box in the image's two-dimensional coordinate system, where the image center is the origin.

[0051] The training process of the corn ear recognition model is as follows: First, we constructed a dataset: we divided the collected RGB images into training, test, and validation sets in an 8:1:1 ratio. We then used the labelimg tool to annotate the corn ear images, manually marking the location and category of the corn ears in the images and generating the corresponding label files.

[0052] Next, train the model: Based on the dataset configuration file data.yaml, select yolov5s as the infrastructure for the corn ear recognition model. Set the training hyperparameters and train the model 250 times. During training, the YOLOv5 model gradually improves its corn ear recognition accuracy through continuous iteration and optimization. After 250 training iterations, the trained best weight file, best.pt, is generated. This file contains the optimal parameters for the corn ear recognition model. During training, the model continuously updates the weights to minimize the loss function and improve recognition accuracy. After training is complete, the generated best.pt model can be used in practical corn ear recognition tasks.

[0053] Step 2: Install and configure the depth camera's built-in SDK and write code by calling the API to implement real-time acquisition of RGB and depth images and alignment of the RGB and depth images. Each acquisition operation is triggered by a keystroke, and the acquired image is saved locally using OpenCV's imwrite() function. The image is saved at a resolution of 1280×720 and in JPG format. Each time it is saved, the system automatically records the current time as part of the file name. For example, the saved file name format is "current acquisition time_rgb.jpg" and "current acquisition time_depth.jpg," ensuring that each acquired image is uniquely identified. The depth camera integrates a high-precision IMU module, which can be obtained through the API and used to calculate the roll and pitch angles.

[0054] Keep the device level and press the first button. The laser radar will shoot laser towards the ground and measure the distance value based on the TOF principle as the reference height. At the same time, the depth camera collects RGB image and depth image data of the target area. The Raspberry Pi saves the RGB image and depth image, and the RGB image is displayed on the display. The inertial measurement unit in the depth camera collects the pitch angle and roll angle of the depth camera relative to the horizontal plane.

[0055] Step 3: Use the depth camera lens position as the coordinate origin to establish the spatial rectangular coordinate system OX c Y c Z c ;X cAxis parallel to the horizontal direction; Y c Axis parallel to the vertical direction; Z c The axis is perpendicular to the camera plane. The camera plane is a plane passing through the camera lens position and perpendicular to the camera lens axis.

[0056] Step 4: Press the second button. The RGB image and depth image are input to the trained corn ear recognition model. The corn ear recognition model identifies and selects the corn ear target in the RGB image. For each identification box output by the corn ear recognition model, the center point coordinates are detected and used as the corn ear coordinates (a, b) in the 2D image coordinates. Based on the depth map data corresponding to the RGB image, the depth value dis corresponding to each corn ear coordinate (a, b) is collected. The depth value dis represents the vertical distance from the corn ear to the camera plane. The 2D coordinates (a, b) are converted to the corresponding 3D coordinates (a, b, c) in the 3D pixel space.

[0057] The coordinate position of the corn ear (a, b) in the actual three-dimensional space is used as the height recognition point; the line connecting the corn ear and the camera lens is used as the vertical plane (Y c -OZ c The included angle between the projection on the X plane and the axis of the camera lens is taken as the vertical angle α of the height recognition point; the line connecting the corn ear and the camera lens is taken as the vertical angle α of the height recognition point. c -OZ c The included angle between the projection on the plane and the axis of the camera lens is taken as the horizontal included angle β of the height recognition point; wherein, -21.5°<α<21.5°, -34.5°<β<34.5°.

[0058] The expressions for the vertical angle α and the horizontal angle β are as follows: α=arctan( )β=arctan( ) = = ; in, is the vertical component ratio; is the proportion of the horizontal component; L and W are the pixel width and pixel height of the image respectively; VFOV and HFOV represent the vertical field of view of the camera (43° in this embodiment) and the horizontal field of view (69° in this embodiment). In this embodiment, the resolution of the image is 1280 720, so the pixel width is 1280 and the pixel height is 720.

[0059] Step 5: Figure 4As shown in the figure, the vertical angle α, the horizontal angle β, the depth value dis, and the height measured by the lidar are combined with trigonometric functions to establish the basic expression of the coordinates Ø(x, y, z) of the height recognition point in the three-dimensional space coordinates as follows: Ø(x,y,z)=(dis tanβ,dis tanα,dis)=(dis ,dis ,dis) Step 6: In actual use, since the device is not absolutely level, the depth camera has a certain pitch angle and roll angle relative to the target installation direction, so it needs to be corrected.

[0060] like Figure 5 As shown in the figure, when the camera has a non-zero pitch angle θ, the measured depth value is lower than the true depth value. When the camera has a roll angle Ɛ less than 0, the left image is raised and the right image is lowered. Similarly, when the roll angle Ɛ is greater than 0, the left image is lowered and the right image is raised. This can lead to incorrect estimation of height recognition points.

[0061] Therefore, it is necessary to combine trigonometric functions and projection relationships to correct the basic expression of the coordinates Ø(x, y, z) and obtain the calibration expression of the coordinates Ø(x, y, z) as follows: Ø(x,y,z)=( , , ) = ; Where dis is the depth value obtained by the depth camera; Ɛ and θ are the roll and pitch angles of the camera; is the vertical component ratio; is the horizontal component ratio, and The sign of is the same as that of b and a respectively.

[0062] Step 7. Calculate the height identification point Ø(x,y,z) based on the calibration expression and the reference height measured by the lidar , the expression for spike height is as follows: Height = + + ; in, is the height difference between the depth camera and the lidar.

[0063] The ear height of corn ears was detected by the method of this example.

[0064] This embodiment is applicable to corn planting, yield prediction and other links in agricultural production, especially for the measurement of corn ear height in large-scale farmland. It can greatly improve measurement efficiency, reduce labor costs, and provide technical support for intelligent agricultural management.

[0065] The height of the corn ear was measured using a tape measure and the corn ear height measurement method provided in this embodiment. The ear height measured manually was used as the true value and evaluated using the root mean square error (RMS) error. The RMS error calculation formula is as follows: RMSE = ; in, Indicates the ear height measured by the device, It represents the manually measured ear height, and n is the number of valid data.

[0066] The measurement results are shown in Table 1 below.

[0067] Table 1 Comparison between automatic measurement and manual measurement

[0068] The root mean square error (RMSE) between corn ear height measured using this device and manual measurements was 0.0235, indicating a small difference and high measurement accuracy. This demonstrates the device's reliability and stability in practical applications. Therefore, the device can accurately measure actual corn ear height and is suitable for use in scenarios requiring high measurement accuracy.

Claims

1. A method for measuring the height of corn ears based on the stereo fusion of a depth camera and a laser radar; characterized by: The depth camera collects RGB images and depth images of the corn ear under test; the reference height is collected by the reference acquisition module; and the pitch angle and roll angle of the depth camera are collected by the inertial measurement unit. Identify the position of the measured corn ear in the RGB image using the corn ear recognition model as a height recognition point; According to the position of the height recognition point in the RGB image, obtain the vertical angle of the height recognition point relative to the depth camera; Obtain the relative height of the corn ear according to the vertical angle α and the depth value of the corresponding height identification point in the depth image; The relative height of corn ear is corrected using pitch angle and roll angle; The corrected relative height is fused with the reference height collected by the reference collection module to obtain the height of the corn ear.

2. The method according to claim 1, wherein: The relative height of the corn ear before correction H s The expression is: H s =dis tanα Where dis is the vertical angle of the depth camera, and α is the vertical angle of the height recognition point relative to the depth camera.

3. The method according to claim 1, wherein: The expression of the vertical angle α is as follows: α=arctan( ) in, is the vertical component ratio; VFOV is the vertical field of view of the camera; Vertical component ratio The expression is: in, The ordinate of the target recognition point in the pixel coordinate system with the center point of the RGB image as the origin; W The pixel width of the RGB image.

4. The method according to claim 3, wherein: The process of correcting the relative height of corn ears is as follows: The pitch angle collected by the inertial measurement unit is used to correct the depth value of the height identification point; The vertical angle is corrected using the roll angle collected by the inertial measurement unit; The relative height of the corn ear is recalculated using the corrected depth value of the height identification point and the vertical angle.

5. The method according to claim 4, characterized in that: The expression for correcting the depth value of the height recognition point is as follows: Where dis and dis' are the depth values ​​before and after correction, respectively; θ is the pitch angle collected by the inertial measurement unit.

6. The method according to claim 4, characterized in that: The process of correcting the vertical angle is: Establish the corrected vertical component ratio The expression is as follows: in, a is the horizontal coordinate of the target recognition point in the pixel coordinate system with the center point of the RGB image as the origin; The roll angle collected by the inertial measurement unit; Using the corrected vertical component ratio Calculate the corrected vertical angle.

7. The method according to claim 1, wherein: The fused corn ear height is obtained by superimposing the relative height of the corn ear, the benchmark height collected by the benchmark collection module, and the height difference between the benchmark collection module and the depth camera.

8. A device for measuring the height of corn ears based on the stereo fusion of a depth camera and a laser radar, characterized by: Used to execute the method as described in any one of claims 1 to 7; the corn ear height measuring device comprises a shell, a depth camera, a reference acquisition module, a control module, a display module and a power supply module; the shell comprises a main body part (1) and a bottom extension part (2); the depth camera, the control module and the display module are all installed on the main body part (1); the reference acquisition module is installed on the bottom extension part (2) and is arranged in a vertical direction.

9. The corn ear height measuring device according to claim 8, characterized in that: The reference acquisition module uses a laser radar to obtain the reference height by collecting the distance from itself to the ground.

10. The corn ear height measuring device according to claim 8, characterized in that: The control module is provided with a corn ear recognition model constructed based on the YOLOv5 model.

Images