A method of measuring cavitation in plant xylem

Abstract

This invention discloses a method for measuring cavitation in plant xylem, comprising: calculating image parallax in an image sample set and recovering scene depth information; extracting the center line of the main stem based on the segmentation results of the main stem; determining the two-dimensional pixel coordinates of the target vibration measurement point above the plant roots along the direction of the center line of the main stem; and calculating the three-dimensional spatial position of the target vibration measurement point by combining the scene depth information corresponding to the target vibration measurement point; automatically coarsely aligning the target vibration measurement point above the plant roots; obtaining the optimal alignment posture of the laser Doppler vibrometer; collecting vibration signals from the target vibration measurement point above the plant roots; and measuring the cavitation in the plant xylem using the collected vibration signals. This invention addresses the problems of manual reliance, low efficiency, poor repeatability, and unstable signal quality caused by on-site interference in existing technologies for measurement point identification and LDV alignment, and achieves automatic positioning, automatic alignment, and cavitation event measurement of the position above the plant roots.

Description

Technical Field

[0001] This invention relates to the field of non-destructive testing of plant physiological state and optical vibration measurement technology, and particularly to an automatic alignment device for a laser Doppler vibrometer (LDV) integrating depth vision recognition and multi-axis gimbal control, as well as a method for measuring cavitation of plant xylem using the device. Background Technology

[0002] Water is a crucial ecological factor governing the growth and metabolism of terrestrial plants. Under drought or high evaporation demands, according to the cohesive-tension theory, plants primarily transport water from the roots to the aboveground parts through negative pressure in the xylem. Because this process relies on maintaining a continuous water column in a metastable state, the plant's hydraulic system is inherently fragile. As drought intensifies, when xylem tension exceeds the limits of its anatomical structure, it can cause water column rupture and embolism, a process known as xylem cavitation. Xylem cavitation significantly reduces hydraulic conductivity, inhibits photosynthesis and carbon assimilation, and may lead to hydraulic failure or even plant death. Therefore, accurately characterizing and monitoring xylem cavitation is essential for assessing plant drought resistance and predicting agricultural and ecosystem responses under extreme drought conditions.

[0003] Traditional methods for measuring plant xylem vulnerability (such as natural dehydration and centrifugation) are mostly destructive operations performed in vitro, easily introducing open vessel artifacts and thus overestimating the plant's vulnerability. To achieve in-situ monitoring, acoustic emission technology has been introduced to capture the ultrasonic mechanical waves during gas embolism. However, existing commercially available contact piezoelectric ceramic sensors have significant limitations in continuous in vivo monitoring: their own weight can easily cause mechanical damage to the plant's vulnerable cortex; at the same time, during long-term monitoring, the acoustic coupling grease often dries out and is lost due to bark roughness and water loss. This coupling degradation leads to significant attenuation of high-frequency signals, making it difficult to guarantee the long-term consistency and fidelity of continuous monitoring data.

[0004] LDV (Laser-Driven Vibration) is suitable for non-contact acquisition of micro-vibrations on plant surfaces, but stable measurements near the root collar, i.e., above the roots, still face significant limitations. The location of measurement points in this area is difficult to standardize and is often accompanied by obstructions and complex backgrounds, making rapid and accurate identification of measurement points challenging. Furthermore, LDV requires high precision in laser beam alignment; relying solely on manual operation is not only inefficient but also prone to introducing human error. In addition, wind-induced vibrations, environmental noise, and stray light in the field environment further degrade signal quality, while assembly errors and temperature drift errors between the camera, gimbal, and LDV also affect system alignment accuracy. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method for measuring cavitation in plant xylem, thereby resolving the issues of manual reliance on measurement point identification and LDV alignment, low efficiency, poor repeatability, and unstable signal quality caused by on-site interference in the prior art. This method enables automatic positioning, automatic alignment, and cavitation event measurement of the plant's root and above locations.

[0006] A method for measuring cavitation in plant xylem, characterized by comprising the following steps: Step 1: Place the plant to be tested on the vibration isolation platform so that the plant is in the common measurable space of the binocular depth camera and the laser Doppler vibration meter. Adjust the position of the vibration isolation platform and the plant posture to ensure that the target area above the plant roots is unobstructed. Step 2: Use a binocular depth camera to acquire color and depth images of the plant to be tested, and construct an image sample set containing information about the main stem region and background of the plant using the color and depth images; Step 3: Train the plant main stem recognition model using the image sample set from Step 2 to obtain the segmentation result of the main stem of the plant to be tested. Then, extract the geometric morphological features of the main stem from the segmentation result of the main stem of the plant to be tested, and determine the two-dimensional pixel coordinates of the reference point of the plant root according to the geometric rules. Step 4: Calculate the image disparity in the image sample set of Step 2 and restore the scene depth information. Based on the segmentation result of the main stem of the plant, extract the center line of the main stem. According to the two-dimensional pixel coordinates of the reference point of the plant root, determine the two-dimensional pixel coordinates of the target vibration point above the plant root along the direction of the center line of the main stem. Combined with the scene depth information corresponding to the target vibration point, calculate the three-dimensional spatial position of the target vibration point in the camera coordinate system. Step 5: Use the Perspective-n-Point algorithm to jointly calibrate the laser Doppler vibrometer and the binocular depth camera to obtain the transformation matrix from the binocular depth camera coordinate system to the laser Doppler vibrometer coordinate system. Step 6: Using the transformation matrix obtained in Step 5, the three-dimensional spatial position of the target vibration measurement point obtained in Step 4 is transformed from the binocular depth camera coordinate system to the laser Doppler vibration meter coordinate system. The mapping relationship between the three-dimensional spatial position of the target vibration measurement point and the horizontal rotation angle and pitch angle of the laser Doppler vibration meter is established. Combined with image parallax control, the laser Doppler vibration meter is rotated to automatically coarsely align the target vibration measurement point above the plant roots. Step 7: After the laser Doppler vibrometer completes automatic coarse alignment, the horizontal and pitch angles of the laser Doppler vibrometer are finely adjusted using the intensity of the laser Doppler vibration signal to obtain the optimal alignment posture of the laser Doppler vibrometer. Under the optimal alignment posture, the laser Doppler vibrometer collects vibration signals from the target vibration measurement points above the plant roots, and measures the cavitation of the plant xylem using the collected vibration signals.

[0007] The cavitation of plant xylem is measured by collecting vibration signals, specifically including the following process: After the laser Doppler vibrometer is in its optimal alignment position, continuous vibration signals are acquired from the target vibration measurement points above the plant roots to obtain the vibration time-domain signals of the target vibration measurement points within a set sampling period. These vibration time-domain signals reflect the velocity signals of the micro-vibration response on the plant tissue surface. The acquired vibration time-domain signals are preprocessed, including background noise suppression, abnormal peak removal, baseline correction, and bandpass filtering, to reduce the impact of environmental mechanical vibration, system drift, and low-frequency disturbances on subsequent identification results, obtaining effective vibration signals for subsequent analysis.

[0008] After obtaining the effective vibration signal, a joint time-domain and frequency-domain analysis is performed on the effective vibration signal to extract signal feature parameters characterizing the cavitation activity of plant xylem. These signal feature parameters include instantaneous vibration amplitude, peak value, root mean square value, pulse width, rate of change of rise time, short-time energy, dominant frequency, spectral peak value, characteristic frequency band energy, and transient pulse event count. Since xylem cavitation typically manifests as short-duration, relatively concentrated transient vibration events, pulse characteristics and local energy abrupt change characteristics in the signal can be extracted to characterize the cavitation response.

[0009] Based on the extracted signal feature parameters, candidate transient events in the effective vibration signal are identified, and candidate transient events that meet preset judgment conditions are determined to be xylem cavitation events. The preset judgment conditions include amplitude threshold conditions, energy threshold conditions, and duration. When the effective vibration signal shows a sudden increase in amplitude and a short-term energy rise within a preset time window, and the corresponding spectral energy is concentrated in a preset characteristic frequency band, it can be determined that a xylem cavitation event has occurred within the preset time window.

[0010] After identifying xylem cavitation events, the results were statistically analyzed to obtain characterizing parameters such as the event occurrence time, duration, intensity, cumulative occurrence count, number of events per unit time, and energy distribution at the target vibration measurement point. Event intensity was characterized by peak amplitude and characteristic frequency band energy; the number of events per unit time was used to characterize the degree of xylem cavitation activity in the tested plant during the current observation phase; and the event occurrence sequence and intensity variation trend were used to characterize the development process of xylem cavitation.

[0011] Based on the frequency of events, the number of events per unit time, the event intensity, and their temporal distribution, the cavitation state of the xylem in the tested plant is quantitatively measured. The xylem cavitation measurement results include whether cavitation occurred, the onset time of cavitation, the activity level of the cavitation event, the cavitation development trend, and the changes in cavitation response within different measurement periods. Comparative analysis can also be performed on measurement results from multiple target vibration points or multiple measurement periods to obtain the differences in xylem cavitation characteristics of the tested plant at different spatial locations or different time stages.

[0012] To achieve the above objectives, the LDV automatic alignment device provided by the present invention is implemented as follows: the device includes a soundproof box, a vibration isolation table, a two-axis rotating gimbal, an LDV mounted on the rotating gimbal, a depth camera for plant and measurement point identification, and sound-insulating cotton material lining the interior.

[0013] Specifically, the soundproof enclosure mainly consists of two layers of sound-insulating material and supporting plywood. The inner layer is lined with 4 cm thick sound-insulating cotton, and the outer layer is lined with 6 cm thick sound-insulating stone slabs, forming a relatively enclosed space with optical windows. This reduces interference from external wind, ambient noise, and stray light on target identification and vibration measurement, while also providing laser safety protection. Two vibration isolation platforms are installed inside the enclosure, one for the plant under test and the other for the measuring equipment. A two-axis rotating gimbal is located inside the enclosure and mechanically connected to the vibration isolation platforms, used to drive the LDV to adjust its direction in at least two rotational degrees of freedom. The LDV is fixedly mounted on the gimbal and outputs electrical or digital signals related to the vibration of the target surface. The depth camera uses an Intel RealSense D415 to acquire image and depth information of the target plant and the area above its roots. The device also includes a control and processing unit, which is used to identify measurement points and estimate their three-dimensional coordinates based on depth camera data, convert the measurement point coordinates to the coordinate system of the gimbal or laser Doppler vibrometer, calculate and control the gimbal rotation angle so that the LDV laser beam is automatically aligned with the target measurement point, and complete the acquisition, processing and output of the LDV output signal.

[0014] Specifically, a method for measuring cavitation in plant xylem using the above-mentioned device includes the following steps: Step 1: Place the plant to be tested on the vibration isolation platform inside the soundproof box, so that the plant is in the common measurable space of the binocular depth camera and the laser Doppler vibration meter; adjust the position of the vibration isolation platform and the plant posture to ensure that the target area above the plant roots is unobstructed, and seal the soundproof box after the position adjustment is completed to suppress the interference of external vibration and environmental noise on the measurement process.

[0015] Step 2: Use a binocular depth camera mounted on a gimbal to acquire color and depth images of the plant to be tested. Filter, label, and enhance the acquired images to construct an image sample set containing information about the plant's main stem region and background, providing a data foundation for subsequent training of the plant main stem segmentation model.

[0016] Step 3: Train the plant main stem recognition model based on the constructed dataset to obtain the segmentation result of the main stem of the plant to be tested; then extract the geometric morphological features of the main stem based on the main stem segmentation result, and determine the two-dimensional pixel coordinates of the plant root reference point according to the preset geometric rules.

[0017] Step 4: Calculate image parallax and recover scene depth information based on the principle of binocular stereo vision. Extract the center line of the main stem based on the segmentation results of the plant main stem. Based on the two-dimensional pixel coordinates of the root reference point, determine the two-dimensional pixel coordinates of the target vibration measurement point at a specified distance above the plant root along the direction of the center line of the main stem. Combine the depth information corresponding to the target vibration measurement point to calculate the three-dimensional spatial position of the target vibration measurement point in the camera coordinate system.

[0018] Step 5: Use the Perspective-n-Point (PnP) algorithm to jointly calibrate the LDV and the stereo depth camera to obtain the transformation matrix from the stereo depth camera coordinate system to the LDV coordinate system.

[0019] Step 6: Using the transformation matrix obtained in Step 5, the spatial position of the target vibration measurement point obtained in Step 4 is transformed from the coordinate system of the binocular depth camera to the coordinate system of the laser Doppler vibration meter. A mapping relationship is established between the spatial position of the target vibration measurement point and the horizontal and vertical rotation angles of the laser Doppler vibration meter. The laser Doppler vibration meter is rotated by controlling the pan-tilt unit to achieve automatic coarse alignment of the vibration measurement point at a specified distance above the plant roots.

[0020] Step 7: After the laser Doppler vibrometer completes the coarse alignment, the visual positioning error and the signal intensity returned by the laser Doppler vibrometer are used as the joint feedback quantity to perform closed-loop fine alignment of the horizontal and pitch angles of the two-dimensional rotating gimbal, obtain the optimal alignment posture, and complete the acquisition of vibration signals of the target vibration measurement point to obtain vibration response data related to the cavitation process of plant xylem, thereby realizing the precise positioning and alignment control of plant xylem cavitation.

[0021] In step 1, the plant to be tested is preferably a plant that can undergo xylem cavitation and produce a corresponding acoustic vibration response.

[0022] In step 2, the plant images preferably cover the imaging of the main stem of the plant under different lighting conditions, different plant postures, different growth stages, and different measurement distances.

[0023] In step 2, the collected dataset is divided into training set, validation set and test set in a ratio of 7:2:1.

[0024] In step 3, the plant main stem recognition model is the YOLOv8-seg model, which is used to perform instance segmentation on the main stem region in the input plant image and output a binary segmentation mask of the plant main stem.

[0025] In step 3, extracting the geometric morphological features of the main stem involves filtering the connected components of the main stem segmentation mask, retaining the effective connected regions corresponding to the main stem, and extracting the set of main stem pixels from these connected regions. Geometric center Main spindle direction Centerline And local width information. Among them, the set of main stem pixels. Represented as: (1) Geometric center of the main stem region Represented as: (2) Main stem axis direction The covariance matrix of the main stem pixel set The largest eigenvalue corresponds to the eigenvector, which is determined as follows: (3) (4) Furthermore, the connected regions of the main stem are subjected to skeletal processing to obtain the point set of the main stem centerline. : (5) in, Indicates the first Two-dimensional coordinates of each main stem pixel. and They represent the first The horizontal and vertical coordinates of each main stem pixel This represents the total number of pixels within the main stem region. Indicates the geometric center of the main stem region. The covariance matrix representing the distribution of pixels on the main stem. Indicates the first The offset vector of each main stem pixel relative to the geometric center, superscript This indicates the transpose operation. This represents the principal axis direction vector of the main stem in the image plane. , and represent the components of the principal axis in the horizontal and vertical directions, respectively. This represents the eigenvector corresponding to the largest eigenvalue of the covariance matrix. Represents the set of points along the center line of the main stem. Indicates the first Two-dimensional coordinates of the centerline point and They represent the first The horizontal and vertical coordinates of each centerline point This represents the total number of points on the center line.

[0026] In step 3, the preset geometric rule is based on the direction of the main stem axis. Identify candidate regions at the base of the main stem, and then within these candidate regions, follow the direction normal to the main stem axis. Search the two sides of the main stem and use the center point of the section with the most stable local cross-sectional width as the reference point for the plant root.

[0027] The normal direction The calculation formula is: (6) In the formula, , and represent the components of the principal axis in the horizontal and vertical directions, respectively; It is the normal direction vector perpendicular to the direction of the main stem axis.

[0028] For candidate points on the center line Its projection value along the principal axis direction The calculation is as follows: (7) In the formula, Indicates the first The coordinates of the centerline points; This represents the projection value of the point along the principal axis. Represents the principal axis direction vector Transpose of; This indicates the geometric center of the main stem region.

[0029] Candidate area at the lower part of the main stem Defined as: (8) In the formula, This represents the set of centerline points within the candidate region at the lower part of the main stem; This represents the maximum value among all the projected values ​​of the centerline points; This indicates the preset candidate range length.

[0030] For candidate regions Any candidate point within The distance parameters to the two boundaries of the main stem can be obtained by searching along the normal direction. and The width of the local section corresponding to the candidate point. for: (9) Corresponding cross-sectional center point The calculation is as follows: (10) In the formula, Indicates the first Local cross-sectional width at each candidate point; and These represent the distance parameters to the boundary of the main stem along the positive and negative normal directions, respectively. Indicates the first The center point of each candidate cross section.

[0031] Median of local cross-sectional width within the candidate region Represented as: (11) Width stability evaluation function Build as: (12) In the formula, This represents the median width of a local section within the candidate region; Indicates the median operation; Indicates the first The width stability evaluation value of each candidate point.

[0032] Finally, the evaluation function is selected. The smallest candidate cross-section center point is used as the reference point for the plant root. ,Right now: (13) In the formula, Two-dimensional pixel coordinates representing the reference point of the plant root; This represents the center point of the candidate section with the smallest evaluation value; This indicates that the evaluation function The candidate point number that yields the minimum value.

[0033] In step 4, the binocular stereo vision technology includes: Step 4.1: Solve for the internal parameters, external parameters, and distortion coefficients of the binocular camera using the checkerboard calibration method; Step 4.2: Based on the external parameters obtained in the calibration process, the Bouguet algorithm is applied to correct the binocular images to achieve coplanarity of the left and right image planes and horizontal alignment of the epipolar lines. Step 4.3: Use a semi-global stereo matching algorithm to obtain the disparity data of the left and right views, and then calculate and extract the three-dimensional depth information of the target scene. The three-dimensional depth information includes: image disparity and scene depth information.

[0034] In step 4, the main stem centerline is the main stem centerline point set mentioned above. .

[0035] In step 4, based on the depth information corresponding to the centerline of the main stem, the target vibration measurement point is converted from image coordinates to three-dimensional spatial coordinates in the camera coordinate system. .

[0036] (14) In the formula, This represents the three-dimensional spatial coordinates of the target vibration measurement point in the camera coordinate system; , and These represent the target vibration measurement points in the camera coordinate system. axis, shaft and Coordinate components along the axis; and These represent the horizontal and vertical pixel coordinates of the target vibration measurement point, respectively. This indicates the depth value corresponding to the target vibration measurement point; and These represent the equivalent focal length of the camera in the horizontal and vertical directions, respectively. and These represent the horizontal and vertical coordinates of the camera's principal point, respectively.

[0037] In step 6, the joint calibration coordinate transformation relationship based on the PnP algorithm is expressed as follows: (15) In the formula, and These represent the three-dimensional coordinates of the target point in the binocular depth camera coordinate system and the laser Doppler vibration meter coordinate system, respectively. and These represent the rotation matrix and translation vector, respectively, used to transform the coordinate system from the binocular depth camera coordinate system to the laser Doppler vibration meter coordinate system.

[0038] Compared with the prior art, the advantages of the present invention are as follows: This invention combines binocular depth vision perception with automatic alignment of a laser Doppler vibration meter. By segmenting the main stem and using geometric rules governing the stability of local cross-sectional width, it precisely locates the root reference point, achieving both low system complexity and high stability without requiring the construction of additional key point positioning models. Based on this reference, the system can automatically select vibration measurement points at a specified distance above the root and drive the instrument to complete alignment, effectively overcoming the shortcomings of traditional manual adjustments that rely on experience, have strong subjectivity in measurement point selection, and are difficult to repeat, significantly improving the automation level of measurement. Furthermore, this method has good adaptability to varying lighting conditions, plant morphology, and measurement distances. Combined with the low-noise, non-contact measurement environment provided by the soundproof box and vibration isolation table, it minimizes interference from external environmental vibrations and human intervention, making it highly suitable for standardized acquisition, continuous in-situ monitoring, and high-throughput batch testing of plant xylem cavitation acoustic vibration responses. Attached Figure Description

[0039] The present invention will be further described in detail below with reference to the accompanying drawings.

[0040] Figure 1 This is a 3D model diagram of the present invention; Figure 2 This is a model diagram of the two-dimensional rotating gimbal in this invention.

[0041] Figure 3 This is a flowchart illustrating a specific implementation of the present invention; Detailed Implementation

[0042] like Figure 1 and Figure 2 As shown, the automatic alignment device for a laser Doppler vibration meter integrating depth vision recognition and multi-axis gimbal control mainly includes: a soundproof box 1, a laser Doppler vibration meter 2, a two-dimensional rotating gimbal 3, a binocular depth camera 4, a vibration isolation table 5, a plant to be tested 6, a movable door 7, and sound-absorbing cotton 8. The laser Doppler vibration meter 2 is fixed on the two-dimensional rotating gimbal 3. The two-dimensional rotating gimbal 3 and the plant to be tested 6 are placed on different vibration isolation tables 5. The sound-absorbing cotton 8 is attached inside the soundproof box 1. All components are placed inside the soundproof box 1. Before the experiment begins, the movable door 7 is closed.

[0043] More specifically, such as Figure 2 As shown, the specific structure of the two-dimensional gimbal mainly includes: load object 1, longitudinal rotation axis 2, longitudinal rotation drive motor 3, lateral rotation drive motor 4, base 5, and lateral rotation axis 6.

[0044] like Figure 3 As shown, a method for cavitation positioning and alignment control of plant xylem is provided, the method mainly includes the following steps: Step 1: Place the plant to be tested on the vibration isolation platform inside the soundproof box, ensuring the plant is within the shared measurable space of the binocular depth camera and the laser Doppler vibration meter. The plant to be tested is preferably one capable of xylem cavitation and generating a corresponding acoustic vibration response. Adjust the position of the vibration isolation platform and the plant's posture to ensure the area above the plant roots is unobstructed, and match the measurement direction of the laser Doppler vibration meter with the field of view of the binocular depth camera. After completing the position adjustment, seal the soundproof box to suppress interference from external vibrations and environmental noise in the subsequent measurement process.

[0045] Step 2: Acquire color and depth images of the plant under test using a binocular depth camera mounted on a 2D rotating platform. Filter, label, and enhance the acquired images to construct an image sample set containing information about the plant's main stem region and background. The image samples preferably cover the main stem imaging under different lighting conditions, plant postures, growth stages, and measurement distances to improve the generalization ability of the subsequent model. The constructed dataset is preferably divided into a training set, a validation set, and a test set in a 7:2:1 ratio for training, validating, and testing the plant main stem recognition model.

[0046] Step 3: Train a plant main stem recognition model based on the dataset constructed in Step 2. The plant main stem recognition model is preferably a YOLOv8-seg instance segmentation model, used to segment the main stem region in the input plant image and output a binary segmentation mask of the plant main stem. After obtaining the main stem segmentation result, the segmentation mask is filtered for connected components, retaining the effective connected regions corresponding to the plant main stem. Subsequently, the set of main stem pixels, geometric center, principal axis direction, centerline, and local width information are extracted from these connected regions. The principal axis direction can be determined by the eigenvector corresponding to the largest eigenvalue of the main stem pixel distribution covariance matrix, and the centerline can be obtained by skeletonizing the main stem connected regions.

[0047] Furthermore, candidate regions at the lower part of the main stem are determined according to the main axis direction, and a two-sided boundary search is performed on the center line points within the candidate regions along the normal direction perpendicular to the main axis direction to obtain the local cross-sectional width corresponding to each candidate point; the center point of the cross-section with the most stable local width within the candidate regions is used as the reference point of the plant root, thereby determining the two-dimensional pixel coordinates of the reference point of the plant root.

[0048] In other words, this step completes the automatic calculation from "main stem segmentation result" to "two-dimensional coordinates of root reference point", providing geometric reference for subsequent target vibration measurement point positioning.

[0049] Step 4: Recover the depth information of the target scene based on the principle of binocular stereo vision, and determine the position of the target vibration measurement point by combining it with the center line of the main stem. Specifically, firstly, the intrinsic, extrinsic, and distortion parameters of the binocular camera are solved using the checkerboard calibration method; then, the binocular images are corrected based on the calibration results to achieve coplanarity of the left and right image planes and horizontal alignment of the epipolar lines; then, a semi-global stereo matching algorithm is used to obtain the disparity data of the left and right views, and the scene depth information is obtained by converting the disparity. After obtaining the center line of the main stem and the root reference point, the two-dimensional pixel coordinates of the target vibration measurement point at a preset distance above the plant root are determined along the direction of the center line of the main stem, using the root reference point as the starting point; combined with the depth value corresponding to the target vibration measurement point, it is converted from image coordinates to three-dimensional spatial coordinates in the camera coordinate system. The three-dimensional spatial coordinates of the target vibration measurement point in the camera coordinate system can be expressed as: In the formula, This represents the three-dimensional spatial coordinates of the target vibration measurement point in the camera coordinate system; , and These represent the target vibration measurement points in the camera coordinate system. axis, shaft and Coordinate components along the axis; and These represent the horizontal and vertical pixel coordinates of the target vibration measurement point, respectively. This indicates the depth value corresponding to the target vibration measurement point; and These represent the equivalent focal length of the camera in the horizontal and vertical directions, respectively. and These represent the horizontal and vertical coordinates of the camera's principal point, respectively.

[0050] Step 5: The PnP algorithm is used to jointly calibrate the binocular depth camera and the laser Doppler vibrometer, solving the rigid transformation relationship between the binocular depth camera coordinate system and the laser Doppler vibrometer coordinate system to obtain the rotation matrix and displacement vector, thereby establishing the spatial mapping between the two coordinate systems. The coordinate transformation relationship can be expressed as: In the formula, and These represent the three-dimensional coordinates of the target point in the binocular depth camera coordinate system and the laser Doppler vibration meter coordinate system, respectively. and These represent the rotation matrix and translation vector, respectively, used to transform the coordinate system from the binocular depth camera coordinate system to the laser Doppler vibration meter coordinate system.

[0051] Step 6: Using the coordinate transformation relationship obtained in Step 5, the three-dimensional spatial position of the target vibration measurement point obtained in Step 4 is transformed from the binocular depth camera coordinate system to the laser Doppler vibrometer coordinate system. Then, based on the spatial position of the target vibration measurement point in the laser Doppler vibrometer coordinate system, the horizontal and vertical rotation angles of the corresponding two-dimensional rotating gimbal are calculated, and a mapping relationship between the spatial position of the target point and the gimbal rotation angles is established. Subsequently, the two-dimensional rotating gimbal is controlled to drive the laser Doppler vibrometer to rotate, so that the measurement optical path of the laser Doppler vibrometer automatically points to the target vibration measurement point at a preset distance above the plant roots, achieving automatic coarse alignment of the target vibration measurement point.

[0052] Step 7: After the laser Doppler vibrometer completes coarse alignment, the horizontal and vertical angles of the two-dimensional rotating gimbal are finely adjusted using the visual positioning error and the signal intensity returned by the laser Doppler vibrometer as joint feedback quantities. This allows the laser measurement spot to further approach and stably lock onto the target vibration measurement point, obtaining the optimal alignment posture. After completing the fine alignment, the laser Doppler vibrometer is activated to collect vibration signals from the target vibration measurement point, obtaining vibration response data related to the cavitation process of the plant xylem, thereby achieving precise positioning and alignment control of plant xylem cavitation.

Claims

1. A method for measuring cavitation in plant xylem, characterized in that, Includes the following steps: Step 1: Place the plant to be tested on the vibration isolation platform so that the plant is in the common measurable space of the binocular depth camera and the laser Doppler vibration meter. Adjust the position of the vibration isolation platform and the plant posture to ensure that the target area above the plant roots is unobstructed. Step 2: Use a binocular depth camera to acquire color and depth images of the plant to be tested, and construct an image sample set containing information about the main stem region and background of the plant using the color and depth images; Step 3: Train the plant main stem recognition model using the image sample set from Step 2 to obtain the segmentation result of the main stem of the plant to be tested. Then, extract the geometric morphological features of the main stem from the segmentation result of the main stem of the plant to be tested, and determine the two-dimensional pixel coordinates of the reference point of the plant root according to the geometric rules. Step 4: Calculate the image disparity in the image sample set of Step 2 and restore the scene depth information. Based on the segmentation result of the main stem of the plant, extract the center line of the main stem. According to the two-dimensional pixel coordinates of the reference point of the plant root, determine the two-dimensional pixel coordinates of the target vibration point above the plant root along the direction of the center line of the main stem. Combined with the scene depth information corresponding to the target vibration point, calculate the three-dimensional spatial position of the target vibration point in the camera coordinate system. Step 5: Use the Perspective-n-Point algorithm to jointly calibrate the laser Doppler vibrometer and the binocular depth camera to obtain the transformation matrix from the binocular depth camera coordinate system to the laser Doppler vibrometer coordinate system. Step 6: Using the transformation matrix obtained in Step 5, the three-dimensional spatial position of the target vibration measurement point obtained in Step 4 is transformed from the binocular depth camera coordinate system to the laser Doppler vibration meter coordinate system. The mapping relationship between the three-dimensional spatial position of the target vibration measurement point and the horizontal rotation angle and pitch angle of the laser Doppler vibration meter is established. Combined with image parallax control, the laser Doppler vibration meter is rotated to automatically coarsely align the target vibration measurement point above the plant roots. Step 7: After the laser Doppler vibrometer completes automatic coarse alignment, the horizontal and pitch angles of the laser Doppler vibrometer are finely adjusted using the intensity of the laser Doppler vibration signal to obtain the optimal alignment posture of the laser Doppler vibrometer. Under the optimal alignment posture, the laser Doppler vibrometer collects vibration signals from the target vibration measurement points above the plant roots, and measures the cavitation of the plant xylem using the collected vibration signals.

2. The method for measuring cavitation in plant xylem according to claim 1, characterized in that, In step 1, the vibration isolation table is arranged inside the soundproof box.

3. The method for measuring cavitation in plant xylem according to claim 1, characterized in that, In step 3, the plant main stem recognition model is the YOLOv8-seg model, which is used to perform instance segmentation on the main stem region in the input plant image and output a binary segmentation mask of the plant main stem.

4. The method for measuring cavitation in plant xylem according to claim 1, characterized in that, In step 3, the geometric rule is based on the direction of the main stem axis. Identify candidate regions at the base of the main stem, and then within these candidate regions, follow the direction normal to the main stem axis. Search the two sides of the main stem and use the center point of the section with the most stable local cross-sectional width as the reference point for the plant root.

5. The method for measuring cavitation in plant xylem according to claim 1, characterized in that, In step 4, binocular stereo vision technology is used to calculate the image parallax in the image sample set from step 2 and recover the scene depth information.

6. The method for measuring cavitation in plant xylem according to claim 5, characterized in that, Step 4, the binocular stereo vision technology, specifically includes: Step 4.1: Solve for the internal parameters, external parameters, and distortion coefficients of the binocular camera using the checkerboard calibration method; Step 4.2: Based on the external parameters obtained in the calibration process, the Bouguet algorithm is applied to correct the binocular images, and the coplanarity of the left and right image planes is aligned with the epipolar lines. Step 4.3: Use a semi-global stereo matching algorithm to obtain the disparity data of the left and right views, and then calculate and extract the three-dimensional depth information of the target scene.

7. The method for measuring cavitation in plant xylem according to claim 6, characterized in that, In step 4, the three-dimensional depth information includes: image parallax and scene depth information.

Images