Plant phenotype detection system and method based on multi-wavelength Shanghai laser radar

Abstract

The invention provides a plant phenotype detection system and method based on a multi-wavelength Shanghai laser radar. The system comprises a laser emitting assembly and a receiving imaging assembly, the laser emitting assembly is used for emitting multi-wavelength coaxial sheet laser to irradiate a plant target, and different wavelengths correspond to reflection characteristic differences of plant tissues in different spectral bands; the receiving imaging assembly comprises a receiving lens and an image sensor, and the included angle between the optical axis of the receiving lens and the imaging plane of the image sensor is configured to meet the Saborne imaging condition, so that object points in the irradiation range of the sheet-shaped laser can be clearly imaged on the image sensor. The method comprises the steps of performing imaging acquisition on multi-wavelength laser reflection signals, performing response correction and decoupling on imaging signals based on multi-channel spectral response characteristics of an image sensor to obtain multi-wavelength reflection intensity information, and realizing plant structure identification and phenotype parameter extraction by combining three-dimensional point cloud data and multi-band spectral characteristics.

Description

Technical Field

[0001] This invention relates to the fields of optical measurement, plant phenotyping and multi-wavelength laser imaging, and particularly to a plant phenotyping system and method based on multi-wavelength Saxony lidar. Background Technology

[0002] Plant phenotype refers to the externally measurable characteristics exhibited by plants under the combined influence of specific genotypes and environmental conditions, including plant structure, leaf morphology, photosynthetic characteristics, and physiological state. Rapid and accurate acquisition of plant phenotypic parameters is crucial in agricultural breeding, crop monitoring, and precision agriculture management. Traditional manual measurement methods rely on individual manual measurements, resulting in low efficiency, large errors, and strong subjectivity, making it difficult to meet the demands of modern agriculture for high-throughput, automated phenotypic analysis.

[0003] With the development of optical measurement technology, non-contact sensing methods such as RGB imaging, hyperspectral imaging, and 3D LiDAR are increasingly being applied to the extraction of plant structure and physiological parameters. Among these, conventional LiDAR, based on the Time-of-Flight (TOF) principle, can acquire high-precision 3D structural information, but it typically only possesses a single-wavelength laser, making it unable to simultaneously measure the multi-band spectral reflectance characteristics of plants. This limits its use in distinguishing different plant tissues (such as leaves and branches) or analyzing leaf health. Furthermore, traditional LiDAR suffers from limited depth of field; influenced by factors such as spot size and focus setting, it faces limitations in imaging area during close-range, high-precision phenotypic scanning.

[0004] While multispectral imaging technology can acquire spectral reflectance information of plants in different wavelength bands, it lacks vertical depth structure and cannot provide accurate three-dimensional morphological data, especially in scenarios with complex canopies or overlapping leaves, making it difficult to accurately recover structural details. Existing image reconstruction methods, such as multi-view stereo vision (MVS) or structured light methods, are sensitive to lighting conditions and are prone to reconstruction errors on low-texture plant surfaces.

[0005] In recent years, the Scheimpflug imaging principle has been applied to the field of large depth-of-field fringe measurement. By tilting the image plane, it achieves full-field focal plane imaging, enabling high signal-to-noise ratio laser fringe acquisition at close range. However, existing Scheimpflug lidar systems are mostly used in industrial inspection and smoke / aerosol monitoring, lacking multi-wavelength, multi-modal extension designs for plant phenotyping. Therefore, it is necessary to design a multi-wavelength Scheimpflug lidar system that is simple in structure, low in cost, possesses multi-wavelength spectral response capabilities, and can achieve high-precision 3D reconstruction. This would meet the need for simultaneous measurement of geometric structure and spectral characteristics in plant phenotyping and improve the robustness and accuracy of phenotyping extraction in complex plant scenarios. Summary of the Invention

[0006] This invention aims to address the shortcomings of existing close-range 3D plant measurement devices, such as insufficient depth-of-field imaging, lack of multi-wavelength spectral acquisition, limited 3D accuracy, and high system complexity. It provides a plant phenotyping system and method based on a multi-wavelength Sabine lidar. This system achieves a novel lidar architecture with simultaneous acquisition of structure and spectrum in close-range scenes, high 3D measurement accuracy, and full-field depth-of-field imaging through multi-wavelength laser beam combining, light sheet shaping, Sabine imaging structure design, and calibration mechanisms.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0008] Firstly, a plant phenotyping system based on a multi-wavelength Saxony lidar includes a laser emitting component, a receiving and imaging component, a data processing component, and a control component for controlling system operation; the laser emitting component emits a multi-wavelength coaxial sheet laser to irradiate the plant target; the laser emitting component includes:

[0009] Multiple lasers are used to emit lasers of different wavelengths; each wavelength corresponds to the typical reflection characteristics of plant tissues in different bands.

[0010] A beam-combining optical element used to combine laser beams from multiple lasers into a coaxial multi-wavelength laser beam;

[0011] Optical shaping element used to shape a coaxial multi-wavelength laser beam into a sheet laser;

[0012] The receiving imaging component includes an image sensor and a receiving lens that receives backscattered signals from plant targets and images the backscattered signals onto the image sensor.

[0013] Preferably, the laser includes a first laser, a second laser, and a third laser; the beam combining optical element includes a first dichroic mirror and a second dichroic mirror; the first dichroic mirror is configured to enable the first laser and the second laser to achieve primary beam combining, and the primary beam combining is then incident on the second dichroic mirror to further combine with the laser emitted by the third laser, thereby forming a multi-wavelength coaxial laser beam.

[0014] Preferably, the optical shaping element includes a Powell lens or a cylindrical lens group; the multi-wavelength coaxial laser beam is shaped by the optical shaping element to form a sheet laser with uniform energy distribution.

[0015] Preferably, the image sensor is a CMOS image sensor; the angle between the image sensor and the optical axis of the receiving lens is determined by the Saxophone imaging conditions.

[0016] Preferably, the receiving imaging assembly also includes a filter installed at the front end of the receiving lens or between the receiving lens and the image sensor. The filter is a multi-channel bandpass filter, with the center wavelength of each channel corresponding to the laser wavelength, in order to suppress ambient stray light and enhance selectivity for the laser wavelength.

[0017] Preferably, the laser emitting component is fixed above the receiving imaging component; the receiving imaging component is positioned by a mounting bracket fixed on the protective housing; the control component includes a servo motor, which is fixed to the mounting bracket by a thread, and the output shaft of the servo motor is connected to the receiving imaging component to drive the receiving imaging component to rotate around a horizontal axis to achieve dynamic scanning or field of view adjustment.

[0018] Preferably, the system also includes an attitude sensing component for acquiring attitude information of the system in real time during the scanning process; the attitude angle data collected by the attitude sensing component is used to perform rotation correction on the spatial coordinates corresponding to the scanning angle to compensate for attitude changes during the scanning process, and the attitude sensing component communicates with the data processing component to realize the real-time transmission of attitude angle data and participate in the reconstruction of the three-dimensional point cloud.

[0019] Secondly, the plant phenotypic detection method based on multi-wavelength Sabine lidar includes the following steps:

[0020] The laser emitting component sequentially combines lasers of different wavelengths emitted by multiple lasers and shapes them into a sheet-like laser beam to irradiate the plant target using an optical shaping element.

[0021] The reflected light from the plant target is collected using a receiving imaging assembly that includes an image sensor and a receiving lens; the sheet laser forms clear, slanted stripes on the image sensor under the conditions of Saxophone imaging.

[0022] The data processing component constructs a 3D point cloud of the plant target by determining the relationship between the pixel position of the tilted stripes and the spatial depth of the plant body;

[0023] By normalizing and fusing multi-wavelength reflection signals, the spectral characteristics of different plant tissues are obtained;

[0024] By jointly modeling the spatial geometric information of three-dimensional point cloud and multi-band spectral features, the identification and segmentation of plant target structures are realized, and the corresponding plant phenotypic parameters are output.

[0025] The plant phenotyping method based on multi-wavelength Sabin lidar is used to implement the plant phenotyping system based on multi-wavelength Sabin lidar as described in the first aspect.

[0026] As a preferred option, the method also includes:

[0027] Construct a mapping model between the pixel positions of the tilted stripes and their physical depth:

[0028] A flat, white cardboard with a known depth is selected as the calibration target and placed perpendicular to the plane of the sheet laser. When the sheet laser shines on the calibration target, a standard planar target image with clear, slanted stripes will be formed.

[0029] Standard planar target images are acquired at different distances, and the vertical pixel coordinates of the laser stripes in the imaging plane are extracted to establish a nonlinear mapping relationship between pixels and distance.

[0030] Preferably, the method also includes a step of correcting the multi-channel response of the image sensor. By establishing a multi-channel spectral response model, the original imaging signal is decoupled and corrected to obtain the actual reflection intensity information corresponding to each wavelength of laser.

[0031] The image sensor has N channels, and different color channels have different spectral response characteristics to different wavelengths of light. For any color channel, its output signal intensity is related to the spectral response characteristics of the corresponding channel of the image sensor, the spectral reflectance characteristics of the plant surface to incident light, and the energy distribution of the incident laser at each wavelength. Based on the multi-channel spectral response relationship, a mapping relationship between the output signal of the color channel and the actual reflection intensity of the multi-wavelength laser is constructed, thereby realizing the accurate acquisition of multi-wavelength reflection information.

[0032] Compared with the prior art, the beneficial effects of the present invention are reflected in:

[0033] 1. Unlike traditional multi-band laser detection systems that typically employ multiple detectors or imaging channels to acquire signals of different wavelengths, resulting in complex system structures and high debugging costs, this invention utilizes a single image sensor to achieve simultaneous acquisition of multi-band laser signals. This approach ensures multi-wavelength information acquisition while reducing the number of imaging devices, simplifying the system hardware structure, and lowering the complexity of system integration and debugging, thereby effectively reducing overall manufacturing and maintenance costs. The accuracy of the spectral signals is ensured through the selection of multi-channel filters and multi-channel response correction of the image sensor.

[0034] 2. Unlike traditional multi-wavelength laser imaging systems that suffer from limited depth of field at close range, poor consistency of imaging across different wavelength laser stripes, and difficulty in supporting accurate reflection intensity calculations, this invention combines the principles of Saxony imaging to design a receiving imaging structure. By using coaxial beam combining of multi-wavelength lasers and sheet-like optical shaping, it ensures that laser stripes of different wavelengths maintain a clear imaging state within a relatively short distance range. This improves the signal-to-noise ratio and spatial consistency of multi-band laser reflection signals, providing a stable and reliable imaging foundation for multi-wavelength decoupled measurements and subsequent extraction of plant phenotypic parameters. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the structure of the laser emitting component in Embodiment 1 of the present invention (taking a three-channel example).

[0036] Figure 2 This is a system assembly diagram of Embodiment 1 of the present invention;

[0037] Figure 3 This is a schematic diagram of the depth measurement model calibration in Embodiment 2 of the present invention;

[0038] Figure 4 This is a spectral response curve of the camera in Embodiment 2 of the present invention;

[0039] Figure 5 This is a laser intensity distribution diagram for each wavelength in Embodiment 1 of the present invention;

[0040] Figure 6 This is a graph showing the spectral transmission characteristics of the filter in Embodiment 1 of the present invention.

[0041] Among them, 1. Powell lens; 2. First laser; 3. Second laser; 4. Third laser; 5. Upper mounting bracket; 6. Laser base; 7a. First dichroic mirror; 7b. Second dichroic mirror; 8. Servo motor; 9. Receiving lens; 10. Lower left mounting bracket; 11. Filter; 12. Protective housing; 13. Microcontroller; 14. Inertial measurement unit; 15. Camera mounting housing; 16. Image sensor; 17. Lower right mounting bracket. Detailed Implementation

[0042] To make the technical means, inventive features, objectives, and effects of the invention readily understandable, the invention is further described below with reference to specific illustrations. However, the invention is not limited to the embodiments described below.

[0043] It should be noted that the structures, proportions, sizes, etc., illustrated in the accompanying drawings of this specification are only used to complement the content disclosed in the specification for those skilled in the art to understand and read, and are not intended to limit the conditions under which the present invention can be implemented. Therefore, they have no substantial technical significance. Any modifications to the structure, changes in the proportions, or adjustments to the size, without affecting the effects and objectives that the present invention can produce, should still fall within the scope of the technical content disclosed in the present invention.

[0044] Example 1:

[0045] This invention provides a multi-wavelength Sabine lidar system for plant phenotyping. Based on the Sabine imaging principle, the system illuminates the plant with multi-wavelength sheet lasers and receives the reflected and backscattered signals to obtain the plant's structural features and spectral reflectance characteristics, ultimately achieving three-dimensional point cloud reconstruction and phenotypic parameter extraction.

[0046] like Figure 1 , Figure 2 The multi-wavelength Saxony lidar system for plant phenotyping, shown, includes: a laser emitting component, a receiving imaging component, an attitude sensing component, a control component, and a data processing component.

[0047] The laser emitting assembly includes a Powell lens 1, multiple lasers emitting lasers of different wavelengths, multiple dichroic mirrors, and an upper mounting bracket 5; the number of dichroic mirrors is one less than the number of lasers.

[0048] In this embodiment, the laser includes a first laser 2, a second laser 3, and a third laser 4. The three lasers are mounted on the upper mounting bracket 5 through positioning holes or positioning slots. Their specific spatial positions on the bracket can be set according to the optical path layout requirements and are not limited to a unique installation order or fixed arrangement. The three lasers are used to emit lasers of different wavelengths, and their emission wavelengths match the transmission or reflection bands of the dichroic mirrors that are subsequently installed.

[0049] The number of dichroic mirrors is one less than the number of lasers, used to achieve step-by-step beam combining of laser beams emitted from multiple lasers. The dichroic mirrors include a first dichroic mirror 7a and a second dichroic mirror 7b. The two dichroic mirrors (7a, 7b) are fixed to the upper mounting bracket 5 via positioning slots and positioned in the laser propagation optical path to step-by-step combine laser beams from different lasers according to a predetermined optical path sequence. In this embodiment, the first dichroic mirror 7a is configured to achieve primary beam combining between the first laser 2 and the second laser 3. The combined beam is then incident on the second dichroic mirror 7b, where it is further combined with the laser emitted by the third laser 4, thereby forming a multi-wavelength coaxial laser beam. The dichroic mirrors have high transmittance for lasers whose wavelengths match those emitted by their corresponding lasers, and reflect or cut off other wavelengths of laser light, ensuring effective beam combining of different wavelengths of laser light under different incident sequences and spatial positions.

[0050] After being combined, the multi-wavelength laser beam is shaped by a Powell lens 1 to form a sheet-like laser with uniform energy distribution, which is used to illuminate the target under test, thereby improving the uniformity of light intensity and measurement stability within the irradiated area. To ensure the uniformity of the light sheet illumination, this invention uses a Powell lens to expand and homogenize the multi-wavelength laser, so that the output beam forms a fan-shaped light sheet with a constant energy distribution, reducing the energy concentration and edge brightness attenuation problems caused by traditional cylindrical lenses.

[0051] In this embodiment, the upper mounting bracket 5 is made of metal material through machining to improve the heat dissipation performance of the laser and the overall structural stability.

[0052] The receiving imaging assembly includes a receiving lens 9, a filter 11, an image sensor 16, and a camera mounting housing 15.

[0053] The receiving lens 9 is a fixed-focus lens with a focal length of 16mm. The emitted laser light sheet is absorbed and reflected by the object under test, and then the backscattered signal is collected by the receiving lens 9 and transmitted to the image sensor 16.

[0054] The filter 11 is mounted at the front end of the receiving lens 9, or between the receiving lens 9 and the image sensor 16, to suppress ambient stray light and enhance selectivity for laser wavelength. The filter 11 is a multi-channel bandpass filter, and its transmission curve is shown below. Figure 6 As shown, the laser output spectrum curve ( Figure 5 Corresponding to ).

[0055] According to the Shapiro imaging conditions, the angle between the image plane and the object plane can be expressed as:

[0056]

[0057] Where α is the angle between the principal plane of the receiving lens and the laser plane, L is the distance from the center point of the lens to the laser plane, and f is the focal length of the receiving lens.

[0058] In this embodiment, the angle α between the main plane of the receiving lens and the laser plane is set to be nearly perpendicular, and the distance L from the center point of the lens to the laser plane is set according to the system structure parameters. Combining the above parameters and calculating based on the Saxophone imaging conditions, the angle β between the image plane and the object plane can be determined to be approximately 9°. Accordingly, the mounting angle of the image sensor 16 relative to the optical axis of the receiving lens 9 is preferably set to approximately 80° to satisfy the Saxophone imaging conditions and obtain an extended depth-of-field imaging effect.

[0059] In this embodiment, the image sensor is an RGB three-channel CMOS image sensor, with each color channel being a red channel, a green channel, and a blue channel. The digital image data output by the image sensor 16 is transmitted to the data processing component via a data interface for subsequent multi-wavelength laser reflection intensity decoupling and three-dimensional reconstruction processing.

[0060] The camera mounting housing 15 is connected to the laser base 6, the receiving lens 9 and the image sensor 16 by threads, respectively, to ensure the optical axis consistency and mechanical stability of each component inside the receiving imaging assembly.

[0061] The receiving imaging component is positioned via the lower left mounting bracket 10 and the lower right mounting bracket 17, both of which are threaded onto the protective housing 12. A servo motor 8 is threaded to both the lower left mounting bracket 10 and the camera mounting housing 15, driving the camera mounting housing 15 to rotate around a horizontal axis for dynamic scanning or field-of-view adjustment. The camera mounting housing 15 and the lower right mounting bracket 17 are connected via a shaft hole, providing precise rotational support and positioning.

[0062] The attitude sensing component is an inertial measurement unit (IMU) 14, which is used to acquire the system's attitude information in real time during the scanning process. This attitude information is used to perform rotational correction on the spatial coordinates corresponding to the scanning angle to compensate for attitude changes during the scanning process. The IMU 14 communicates with the data processing component wirelessly via Bluetooth to achieve real-time transmission of attitude angle data, which is then used for subsequent 3D point cloud reconstruction.

[0063] The control components include a laser drive circuit, a servo motor control module, and a system main control unit. This embodiment uses a single-chip microcomputer (ESP32) as the main control unit, which has advantages such as compact structure, low power consumption, and strong wireless communication capabilities. The scanning execution mechanism (laser emitting component and receiving imaging component) of this invention is driven by a servo motor, enabling continuous scanning of the light sheet position or the plant under test to achieve stripe data acquisition. In conjunction with the attitude sensing module, the system can acquire the tilt angle information of the device in real time during the scanning process and use it for compensation of the angle in 3D point cloud reconstruction, avoiding depth errors caused by mechanical vibration or angle deviation.

[0064] The control components and power supply components are housed together within the protective housing 12 to achieve an integrated system design.

[0065] The data processing component can be a portable laptop, giving the system good mobility and deployment flexibility. The laptop communicates with the image sensor 16, inertial measurement unit 14, and main control unit 13 via USB, Bluetooth, and Wi-Fi to perform tasks such as image preprocessing, 3D point cloud reconstruction, and phenotypic parameter extraction. Furthermore, the laptop also provides human-machine interaction, offering parameter settings, acquisition control, operating status display, and result visualization through a graphical interface. This configuration avoids the problems of large size and high cost associated with traditional industrial PCs, making this system more suitable for field plant phenotypic detection scenarios.

[0066] This invention employs a modular hardware structure, featuring simple structure, compact optical path, stable imaging, low cost, insensitivity to ambient light, and the ability to obtain high-quality three-band fringe data even under complex plant geometries. Its output three-dimensional structure and multi-band reflectivity information can be directly applied to various plant phenotypic detection tasks, making it suitable for applications such as smart agriculture, high-throughput breeding, horticultural research, and growth monitoring.

[0067] Example 2:

[0068] Multiwavelength Saxophone for plant phenotyping

[0069] After the system's hardware and software are built, the system needs to be calibrated and corrected to ensure the accuracy of the 3D reconstruction results and the stability of the multispectral measurements. This includes calibration of the laser intensity at each wavelength, correction of the imaging channel response, and calibration of the depth measurement model.

[0070] Laser intensity calibration was performed using a spectrometer. The specific method is as follows: A white diffuse reflector was placed at the system's working distance (e.g., 1 m). 447 nm, 523 nm, and 640 nm lasers (2, 3, 4) were turned on, and the intensity values ​​of the three channels were recorded using a spectrometer (with the spectrometer probe position and sampling parameters fixed) or an illuminometer. Based on the measured relative spectral energy relationship of the three wavelengths of laser light, an intensity correction coefficient can be further calculated, thereby ensuring that the multi-wavelength reflection intensities received by the camera (image sensor) are compared under the same energy reference system.

[0071] Since this system uses the Saxophone imaging principle, its depth information and the pixel positions of the laser stripes in the image have a monotonic mapping relationship. Therefore, it is necessary to construct a mapping model between pixel positions and physical depth. Figure 3 As shown, a flat, white cardboard with a known depth is selected as the calibration target and placed perpendicular to the laser beam plane. When the laser shines on the target, a clear laser stripe is formed on its surface. Image processing is used to extract the vertical pixel position of this stripe in the image, and the corresponding actual physical depth is recorded. Finally, based on the Saxophone imaging principle, a mapping relationship between pixel position and physical depth can be established, thereby calibrating the depth measurement model. To achieve high-precision 3D structure reconstruction, this invention constructs a spatial depth calibration model based on Saxophone imaging. By acquiring standard planar target images at different distances, the vertical pixel coordinates of the laser stripe in the imaging plane are extracted, and a nonlinear mapping relationship between pixels and object distance is established, achieving an accurate solution from the laser stripe position to the actual spatial depth.

[0072] This invention further considers the cross-response characteristics of different color channels of an image sensor to multi-wavelength lasers. By establishing a multi-channel response model, the original imaging signal is decoupled and corrected to obtain the reflection intensity information corresponding to each wavelength of laser. Since image sensors have different relative spectral response sensitivities under different wavelength conditions, to ensure the comparability of multi-wavelength laser reflection intensity data, it is necessary to correct the response of the camera's color channels. Based on the camera's spectral response curve (… Figure 4The relative response intensities of the RGB three color channels are obtained under multiple dominant wavelength conditions, and the response intensities are expressed as percentages. For each dominant wavelength, the response values ​​of the RGB three channels are summed, and the sum is used to normalize the response of each channel, thereby eliminating the gain differences between different color channels of the image sensor and enabling the reflection intensity data obtained by lasers at different wavelengths to be expressed under a unified scale system.

[0073] Furthermore, considering the cross-response relationship between the color channels of a CMOS image sensor and different wavelengths of laser light, the original output signals of the RGB three channels are essentially a linear superposition of multi-wavelength laser reflection signals. Let the original output signals of the red, green, and blue channels be respectively... , and The actual reflected light intensities of the corresponding wavelength lasers are respectively , and Then its linear relationship can be expressed as:

[0074]

[0075] in, This represents the relative spectral response coefficient of color channel i of the image sensor to laser light of wavelength j.

[0076] For simplicity, let's denote it as:

[0077]

[0078] Then we have:

[0079]

[0080] When the response matrix When the matrix is ​​invertible, the original output signals of the RGB channels can be decoupled and corrected by matrix inversion, thereby obtaining the actual reflected light intensity corresponding to each wavelength of laser:

[0081]

[0082] Right now:

[0083]

[0084] In this embodiment, the response matrix Each element The response matrix is ​​determined by the camera's spectral response curve, the selected laser wavelength, and the transmittance of the filter in the corresponding band, and is obtained through pre-calibration. The specific values ​​of the response matrix may vary under different system configurations, but this does not affect the implementation of the decoupling correction method described in this invention.

[0085] The working principle of a multi-wavelength Saxophone lidar is as follows:

[0086] First, the laser emitting assembly sequentially combines the laser beams emitted by 447 nm, 523 nm, and 640 nm laser diodes through a dichroic mirror and shapes them into a sheet-like laser beam through a Powell lens to irradiate the plant. The sheet-like laser generates reflected and backscattered signals on the plant surface, and the differences in the spectral response of different wavelengths of laser light on plant tissue can provide spectral characteristic information for subsequent phenotypic analysis.

[0087] The receiving imaging component collects the reflected light, and the sheet laser forms clear, tilted stripes on the image sensor under the conditions of Sabin's imaging. The pixel positions of the stripes have a definite relationship with the spatial depth of the plant, which can be used for the reconstruction of the 3D point cloud. The image data is transmitted to the data processing component via USB, and after background suppression, stripe center extraction, and geometric backprojection, the 3D point cloud data of the plant is obtained.

[0088] By combining attitude compensation information provided by the inertial measurement unit, attitude compensation and coordinate correction are performed on the reconstructed point cloud data, thereby eliminating the impact of attitude changes during the scanning process on the accuracy of the point cloud. Subsequently, through the normalization and fusion of multi-wavelength reflection signals, the spectral characteristics of different plant tissues can be obtained. By modeling the point cloud spectrum and spatial coordinates, the identification and extraction of structures such as leaves and branches can be achieved, and finally, phenotypic parameters such as plant height, leaf area, and leaf angle are output.

Claims

1. A plant phenotyping system based on multi-wavelength Sabin lidar, characterized in that, It includes a laser emitting assembly, a receiving and imaging assembly, a data processing assembly, and a control assembly for controlling the operation of the system; the laser emitting assembly emits a multi-wavelength coaxial sheet laser to irradiate the plant target; the laser emitting assembly includes: Multiple lasers are used to emit lasers of different wavelengths; each wavelength corresponds to the typical reflection characteristics of plant tissues in different bands. A beam-combining optical element used to combine laser beams from multiple lasers into a coaxial multi-wavelength laser beam; Optical shaping element used to shape a coaxial multi-wavelength laser beam into a sheet laser; The receiving imaging component includes an image sensor (16) and a receiving lens (9) that receives backscattered signals from plant targets and images the backscattered signals onto the image sensor.

2. The plant phenotyping system based on multi-wavelength Sabin lidar according to claim 1, characterized in that, The laser includes a first laser (2), a second laser (3) and a third laser (4); the beam combining optical element includes a first dichroic mirror (7a) and a second dichroic mirror (7b); the first dichroic mirror (7a) is configured to enable the first laser (2) and the second laser (3) to achieve a first-order beam combining, and the first-order beam combining is then incident on the second dichroic mirror (7b) to further combine with the laser emitted by the third laser (4), thereby forming a multi-wavelength coaxial laser beam.

3. The plant phenotyping system based on multi-wavelength Sabine lidar according to claim 2, characterized in that, The optical shaping element includes a Powell lens (1); after being shaped by the Powell lens (1), the multi-wavelength coaxial laser beam forms a sheet laser with uniform energy distribution.

4. The plant phenotyping system based on multi-wavelength Sabine lidar according to claim 3, characterized in that, The angle between the image sensor (16) and the optical axis of the receiving lens (9) is set according to the Saxophone imaging conditions.

5. The plant phenotyping system based on multi-wavelength Sabin lidar according to claim 1, characterized in that, The receiving imaging assembly also includes a multi-channel bandpass filter (11) mounted at the front end of the receiving lens (9) or between the receiving lens (9) and the image sensor (16), the center wavelength of which corresponds to the wavelength of the laser used, in order to suppress ambient stray light and enhance selectivity for the laser wavelength.

6. The plant phenotyping system based on multi-wavelength Sabin lidar according to claim 1, characterized in that, The laser emitting component is fixed above the receiving imaging component; the receiving imaging component is positioned by a mounting bracket fixed on the protective housing (12); the control component includes a servo motor (8); the servo motor (8) is threadedly connected to the mounting bracket and the receiving imaging component respectively, and is used to drive the receiving imaging component to rotate around the horizontal axis to achieve dynamic scanning or field of view adjustment.

7. The plant phenotyping system based on multi-wavelength Sabin lidar according to claim 1, characterized in that, The system also includes an attitude sensing component for acquiring attitude information of the system in real time during the scanning process; the attitude angle data collected by the attitude sensing component is used to perform rotation correction on the spatial coordinates corresponding to the scanning angle to compensate for attitude changes during the scanning process, and the attitude sensing component communicates with the data processing component to realize the real-time transmission of attitude angle data and participate in the reconstruction of the three-dimensional point cloud.

8. A plant phenotypic detection method based on multi-wavelength Sabin lidar, characterized in that, Includes the following steps: The laser emitting component sequentially combines lasers of different wavelengths emitted by multiple lasers and shapes them into a sheet-like laser beam to irradiate the plant target using an optical shaping element. The reflected light from the plant target is collected using a receiving imaging assembly that includes an image sensor and a receiving lens; the sheet laser forms clear, slanted stripes on the image sensor under the conditions of Saxophone imaging. The data processing component constructs a 3D point cloud of the plant target by determining the relationship between the pixel position of the tilted stripes and the spatial depth of the plant body; By normalizing and fusing multi-wavelength reflection signals, the spectral characteristics of different plant tissues are obtained; By jointly modeling the spatial geometric information of three-dimensional point cloud and multi-band spectral features, the identification and segmentation of plant target structures are realized, and the corresponding plant phenotypic parameters are output. The plant phenotyping method based on multi-wavelength Sabin lidar is used to implement the plant phenotyping system based on multi-wavelength Sabin lidar as described in claim 1.

9. The plant phenotypic detection method based on multi-wavelength Sabin lidar according to claim 8, characterized in that, Also includes: Construct a mapping model between the pixel positions of the tilted stripes and their physical depth: A flat, white cardboard with a known depth is selected as the calibration target and placed perpendicular to the plane of the sheet laser. When the sheet laser shines on the calibration target, a standard planar target image with clear, slanted stripes will be formed. Standard planar target images are acquired at different distances, and the vertical pixel coordinates of the laser stripes in the imaging plane are extracted to establish a nonlinear mapping relationship between pixels and distance.

10. The plant phenotypic detection method based on multi-wavelength Sabin lidar according to claim 8, characterized in that, Also includes: Based on the differences in the spectral response of different color channels of an image sensor to different wavelengths of light, the original imaging signals of multiple color channels are jointly analyzed and normalized to construct a mapping relationship between the output signals of the color channels and the reflection intensity of lasers at each wavelength, thereby obtaining the actual reflection intensity information corresponding to each wavelength of laser.

Images