Mycorrhiza-root system trimodality synergistic imaging system and method

Abstract

The present application relates to the technical field of biological tissue detection, and provides a mycorrhiza-root system three-mode cooperative imaging system and method.The mycorrhiza-root system three-mode cooperative imaging system comprises a linearly polarized laser, a sample stage, a first semi-transparent half mirror, a detection unit, an activity acquisition mechanism and a processor.The first semi-transparent half mirror is used for reflecting S-polarized light to the mycorrhiza-root system and transmitting mixed light reflected by the mycorrhiza-root system.The detection unit is used for acquiring S-polarized light, P-polarized light and theta-polarized light.The activity acquisition mechanism is used for acquiring the heat flow density of the mycorrhiza-root system.The processor is used for establishing a quantitative correlation model of specific heat capacity, phase delay and peak polarization angle to obtain a pseudo-color image of the mycorrhiza-root system.The mycorrhiza-root system three-mode cooperative imaging system realizes three-dimensional synchronous analysis of the structure, activity and composition of the mycorrhiza-root system, and provides multi-dimensional quantitative data for the study of the mycorrhiza-root system symbiotic mechanism.

Description

Technical Field

[0001] This invention relates to the field of biological tissue detection technology, and in particular to a mycorrhizal-root three-modal synergistic imaging system and method. Background Technology

[0002] The symbiotic relationship between mycorrhizae and roots is crucial for plant nutrient absorption and resistance to adversity. However, due to their high similarity in morphology (overlapping fine structures) and spectrum (similar absorption characteristics), traditional single-modal techniques have significant limitations. For example, polarization imaging is advantageous in capturing microscopic structural differences such as hyphal diameter and root cell wall arrangement, but it struggles to distinguish regions with similar structures but different physiological activities (e.g., active and dormant hyphae). Specific heat capacity measurement can identify physiologically active regions (e.g., root vascular tissue and mycorrhizal metabolic hotspots) through differences in thermal response, but it cannot accurately locate the morphological boundaries of these regions. Polarization angle spectroscopy can distinguish spectrally similar substances based on polarization "fingerprints," but when used alone, it does not provide information on the physiological activity dimension, making it difficult to determine the actual functional state of mycorrhizae.

[0003] Therefore, developing an imaging system capable of simultaneously identifying mycorrhizal-root boundaries, activity, and components has become a critical issue that urgently needs to be addressed in this field. Summary of the Invention

[0004] This invention provides a mycorrhizal-root three-modal collaborative imaging system and method to overcome the shortcomings of existing single-modal technologies that cannot simultaneously identify the structure, activity, and composition of mycorrhizal-root systems.

[0005] This invention provides a mycorrhizal-root system three-modal collaborative imaging system, comprising: a linearly polarized laser for emitting S-polarized light; a movable sample stage for placing mycorrhizal-root systems; a first semi-transparent mirror disposed opposite to the linearly polarized laser, the first semi-transparent mirror being used to reflect the S-polarized light to the mycorrhizal-root system and transmit mixed light reflected by the mycorrhizal-root system, the mixed light including S-polarized light, P-polarized light, and θ-polarized light; a detection unit for acquiring the S-polarized light, the P-polarized light, and the θ-polarized light; and an activity acquisition mechanism for acquiring the heat flux density of the mycorrhizal-root system. The processor is configured to calculate the phase delay of the P-polarized light and the S-polarized light, and obtain the specific heat capacity of the mycorrhizal-root system based on the heat flux density and the heating area. The processor is also configured to generate a polarization angle spectrum based on the light intensity of the θ-polarized light and the polarization angle of the detection unit, and obtain the peak polarization angle and spectral width based on the polarization angle spectrum. Furthermore, the processor is configured to establish a quantitative correlation model between specific heat capacity, phase delay, and peak polarization angle, employing a weighted voting + deep learning fusion strategy to weight the specific heat capacity, phase delay, and peak polarization angle, and input the weighted data into the MobileNetV2 network to obtain a fused feature vector. The processor is also configured to compare the fused feature vector with the mycorrhizal-root system feature database to obtain a pseudo-color image of the mycorrhizal-root system.

[0006] According to the present invention, a mycorrhizal-root three-modal collaborative imaging system is provided, wherein the detection unit comprises: a first polarizer, movably disposed above a first semi-transparent mirror, the first polarizer being used to filter out S-polarized light in the mixed light; and a first detector, disposed above the first polarizer, the first detector being used to collect P-polarized light when the first polarizer is stationary, the first detector being also used to collect S-polarized light when the first polarizer is removed, and the first detector being also used to collect θ-polarized light when the first polarizer is rotated.

[0007] According to the present invention, a mycorrhizal-root trimodal imaging system is provided, wherein the detection unit comprises: a second semi-transparent mirror disposed above a first semi-transparent mirror; a first detection component disposed above the second semi-transparent mirror, the first detection component being used to collect the P-polarized light and the S-polarized light; and a second detection component disposed on one side of the second semi-transparent mirror, the second detection component being used to collect the θ-polarized light.

[0008] According to the present invention, a mycorrhizal-root three-modal collaborative imaging system is provided, wherein the first detection component includes: a second polarizer, movably disposed above the second semi-transparent mirror, the second polarizer being used to filter out S-polarized light; and a second detector, disposed above the second polarizer, the second detector being used to collect the P-polarized light and the S-polarized light.

[0009] According to the present invention, a mycorrhizal-root three-modal collaborative imaging system is provided, wherein the second detection component includes: a third polarizer, rotatably disposed on one side of the second semi-transparent and semi-reflective mirror, the third polarizer being used to filter out S-polarized light; and a third detector, disposed on one side of the third polarizer, the third detector being used to collect the θ-polarized light.

[0010] According to the mycorrhizal-root trimodal collaborative imaging system provided by the present invention, it further includes: a temperature sensor for detecting ambient temperature, the temperature sensor being electrically connected to the processor; a humidity sensor for detecting ambient humidity, the humidity sensor being electrically connected to the processor; the processor is further configured to compensate for the specific heat capacity when the difference between the current temperature and the initial temperature of the environment is greater than a first preset value, and to compensate for the spectral width when the difference between the current humidity and the initial humidity of the environment is greater than a second preset value.

[0011] According to the present invention, a mycorrhizal-root system trimodal collaborative imaging system includes an activity acquisition mechanism comprising: multiple thermostatic heaters surrounding the sample stage, wherein the thermostatic heaters are used to heat the mycorrhizal-root system after their own temperature reaches the target temperature; multiple radiators, each of which is used to dissipate heat from one of the thermostatic heaters, and the radiators are turned on after the thermostatic heaters are turned off; and a thermal infrared camera disposed above and opposite the sample stage, wherein the thermal infrared camera is used to acquire the heat flux density of the mycorrhizal-root system, and the thermal infrared camera is electrically connected to the processor.

[0012] According to the present invention, a mycorrhizal-root system trimodal imaging system is provided, wherein the constant temperature heater includes: a housing, wherein the air outlet of the radiator faces the housing; louvers are disposed in the housing, wherein when the louvers are opened, multiple airflow channels are formed, the airflow channels are connected to the interior of the housing and face the mycorrhizal-root system; and a heating element is disposed inside the housing.

[0013] This invention also provides an imaging method based on the mycorrhizal-root trimodal collaborative imaging system described above, comprising: acquiring the phase delay of P-polarized light and S-polarized light, the heat flux density of the mycorrhizal-root system, the light intensity of θ-polarized light, and the polarization angle of the detection unit; obtaining the specific heat capacity of the mycorrhizal-root system based on the heat flux density and the heating area; generating a polarization angle spectrum based on the light intensity of θ-polarized light and the polarization angle of the detection unit; and obtaining the peak polarization angle and spectral width based on the polarization angle spectrum; establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle; employing a weighted voting + deep learning fusion strategy to weight the heat capacity, phase delay, and peak polarization angle; inputting the weighted data into a MobileNetV2 network to obtain a fused feature vector; and comparing the fused feature vector with a mycorrhizal-root feature database to obtain a pseudo-color image of the mycorrhizal-root system.

[0014] According to an imaging method provided by the present invention, before the step of establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle, the imaging method further includes: acquiring the current temperature of the environment; comparing the current temperature with an initial temperature; and compensating for the specific heat capacity when the difference between the current temperature and the initial temperature is greater than a first preset value.

[0015] According to an imaging method provided by the present invention, before the step of establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle, the imaging method further includes: acquiring the current humidity of the environment; comparing the current humidity with the initial humidity; and compensating the spectral width when the difference between the current humidity and the initial humidity is greater than a second preset value.

[0016] The mycorrhizal-root three-modal synergistic imaging system provided in this invention, by setting up a linearly polarized laser, a first semi-transparent and semi-reflective mirror, a detection unit, an activity acquisition mechanism, and a processor, achieves three-dimensional synchronous analysis of the structure, activity, and composition of mycorrhizal-root systems, providing multi-dimensional quantitative data for the study of mycorrhizal-root symbiotic mechanisms. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the mycorrhizal-root three-modal collaborative imaging system provided by the present invention.

[0019] Figure 2 yes Figure 1 The diagram shows the structure of the detection unit.

[0020] Figure 3 yes Figure 1 The diagram shows the structure of the active acquisition mechanism.

[0021] Figure 4 This is a comparison chart of the recognition accuracy of the mycorrhizal-root three-modal collaborative imaging system provided in this embodiment of the invention with that of existing technologies.

[0022] Figure label: 10. Linearly polarized laser; 20. Sample stage; 30. First semi-transparent mirror; 40. Detection unit; 41. Second semi-transparent mirror; 42. Second polarizer; 43. Third polarizer; 50. Active collection mechanism; 51. Thermostatic heater; 511. Housing; 512. Louver; 52. Heat sink; 53. Thermal infrared camera; 60. Optical support. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0024] The following is combined Figures 1-4 The mycorrhizal-root three-modal synergistic imaging system and method of the present invention are described.

[0025] like Figure 1 As shown, in an embodiment of the present invention, the mycorrhizal-root three-modal collaborative imaging system includes: a linearly polarized laser 10, a sample stage 20, a first semi-transparent and semi-reflective mirror 30, a detection unit 40, an activity acquisition mechanism 50, and a processor.

[0026] A linearly polarized laser 10 is used to emit S-polarized light, and a sample stage 20 is used to place the mycorrhizal-root system. In this embodiment, the sample stage 20 is movable so that the field of view of the detection unit 40 and the activity acquisition mechanism 50 can cover the sample stage 20. A first semi-transparent mirror 30 is disposed opposite to the linearly polarized laser 10. The first semi-transparent mirror 30 is used to reflect the S-polarized light to the mycorrhizal-root system and transmit the mixed light reflected by the mycorrhizal-root system, wherein the mixed light includes S-polarized light, P-polarized light, and θ-polarized light. The activity acquisition mechanism is used to acquire the heat flux density of the mycorrhizal-root system.

[0027] The processor has a data processing module and a data fusion module. The data processing module calculates the phase delay of P-polarized and S-polarized light, which distinguishes the structures of mycorrhizae and roots, ensuring clear boundaries during imaging. The data processing module also generates a specific heat capacity spectrum of the mycorrhizae-root system based on heat flux density and heating area. Based on the specific heat capacity C, the activity of the mycorrhizae-root system is classified into three levels: C ≥ 3.9 J / (g·℃) indicates high activity, 2.6 J / (g·℃) ≤ C < 3.9 J / (g·℃) indicates medium activity, and C < 2.6 J / (g·℃) indicates low activity, thus distinguishing between active and dormant root systems. The data processing module also generates a polarization angle spectrum based on the intensity of the θ-polarized light and the polarization angle of the detection unit 40, and obtains the peak polarization angle and spectral width based on the polarization angle spectrum. The components of the mycorrhizae-root system can be distinguished based on the peak polarization angle and spectral width.

[0028] The data fusion module is used to establish a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle. It adopts a weighted voting + deep learning fusion strategy to weight specific heat capacity, phase delay, and peak polarization angle, and inputs the weighted data into the MobileNetV2 network to obtain a fused feature vector. The data fusion module is also used to compare the fused feature vector with the mycorrhizal-root feature database to obtain a pseudo-color image of mycorrhizal-root system.

[0029] Specifically, in this embodiment, after S-polarized light strikes the mycorrhizal-root system, the microstructure of the mycorrhizal-root system exhibits optical anisotropy, which converts part of the S-polarized light into other polarization states. The unconverted portion remains S-polarized light. Therefore, the mixed light reflected by the mycorrhizal-root system contains S-polarized, P-polarized, and θ-polarized light. The mixed light is transmitted through the first semi-transparent mirror 30 and then enters the detection unit 40. Optionally, the detection unit 40 may include a polarizer and a detector. The polarizer filters out the S-polarized light, at which point the light signal collected by the detector is pure P-polarized light. When the polarizer is removed, the light collected by the detector consists of P-polarized and S-polarized light; at this point, the detector does not collect θ-polarized light. Subtracting the pure P-polarized light from the P-polarized and S-polarized light yields the pure S-polarized light. The processor can then calculate the phase delay of the P-polarized and S-polarized light. By establishing a linear correlation model between the phase delay of P-polarized light and the mycorrhizal sample database, the mycorrhizal diameter can be obtained. Similarly, by establishing a linear correlation model between the phase delay of S-polarized light and the root system sample database, the root system diameter can be obtained. Thus, the polarization data spectrum of mycorrhizae and roots can be obtained, and mycorrhizae and roots can be distinguished structurally.

[0030] When the polarizer is rotated, it filters out S-polarized light, and the light signal collected by the detector is θ-polarized light. The processor generates a polarization angle spectrum data map based on the intensity of the θ-polarized light and the polarization angle of the detection unit 40, and obtains the peak polarization angle and spectral width based on the polarization angle spectrum data map. From the above discussion, it can be seen that when collecting P-polarized and S-polarized light, the detector does not collect θ-polarized light. Although the mixed light theoretically contains θ-polarized light, in this embodiment, it can be considered that the mixed light contains only S-polarized light + P-polarized light. Conversely, when collecting θ-polarized light, although the mixed light contains P-polarized light, the detector does not collect P-polarized light, and it can also be considered that the mixed light contains only θ-polarized light + S-polarized light. In the above description, the polarization data map, specific heat capacity data map, and polarization angle spectrum data map are all global data maps of the mycorrhizal-root system. Samples ≤ 2cm × 2cm are defined as small samples, and data maps of the core area are directly collected; samples > 2cm × 2cm are defined as large samples, and are divided into minimum regions of 5mm × 5mm and then stitched together to ensure full coverage without omissions. Specifically, the mycorrhizal-root system can be divided into multiple regions. During information collection, the sample stage 20 is controlled to move along a preset path, and the detection unit and activity collection mechanism 50 sequentially collect P-polarized light, S-polarized light, specific heat capacity, and θ-polarized light of each region and send them to the processor. The processor processes the information of each region and stitches them together to obtain a full-domain data map of the mycorrhizal-root system.

[0031] Taking the stitching of polarization data maps of various regions as an example, a scanner is installed on the linear module of the sample stage 20. As the sample stage 20 moves a certain distance along a preset path, the scanner scans the spatial coordinates of that region. The scanner sends the spatial coordinate numbers of each region to the processor. The detector acquires the P-polarized light signal and mixed light signal of the mycorrhizae in each region and sends them to the processor. The processor, based on the phase delay and the linear correlation model of mycorrhizae / root diameter established in the sample database, can obtain the mycorrhizae diameter and root diameter, that is, obtain the polarization data map of mycorrhizae and root fibers in each region. After gradual movement, multiple polarization data maps of multiple regions can be obtained. The processor stitches together the multiple polarization data maps to obtain the global polarization data map of mycorrhizae and roots. Specifically, the sample stage 20 has a built-in position sensor module that can detect the displacement of the sample stage 20 in real time, assign a unique spatial coordinate number to each region, and send the spatial coordinate number to the processor in real time. The detector collects the P-polarized light signal and mixed light signal of that region, and simultaneously binds and stores the collected light signal with the spatial coordinate number. After all regions have been acquired, the processor invokes a stitching algorithm, using spatial coordinate numbers as a reference, to establish a global coordinate mapping matrix. Based on this matrix, the processor arranges, registers, and weights the polarization data spectrum of each region according to its original coordinate position, ultimately forming a global polarization data spectrum without misalignment or stitching gaps. The stitching of the global polarization angle spectrum data spectrum is similar and will not be described in detail here.

[0032] For specific heat capacity data maps, after determining the scanning path, the heat capacity map of each collection point on the scanning path is obtained. Image preprocessing, image registration, grayscale correction and boundary smoothing fusion are performed on the heat capacity map of each collection point to obtain the specific heat capacity data map of mycorrhiza-root system.

[0033] After obtaining the global polarization data spectrum, global specific heat capacity data spectrum, and global polarization angle spectrum of the mycorrhizal-root system (for small samples, the core region data spectrum is directly obtained; for large samples, the global spectrum is formed by region division, local modeling, and then stitching together), the data fusion module first performs environmental adaptive calibration and normalization denoising on the three types of spectra, and then performs quantitative correlation model (C=0.28δ+0.45θ) p +0.07, R 2The data fusion module employs a fusion strategy combining weighted voting and MobileNetV2 deep learning (C = 0.98) to deeply integrate three-dimensional information on specific heat capacity, phase delay, and peak polarization angle, generating a pseudo-color functional map containing integrated information on structure, activity, and composition. Specifically, the data fusion module normalizes and denoises the phase delay, specific heat capacity, and peak polarization angle. Trained with 500 sets of mycorrhizal-root samples, the module establishes a quantitative correlation model between specific heat capacity, phase delay, and peak polarization angle, C = 0.28δ + 0.45θ. p +0.07, where C is the specific heat capacity, δ is the phase delay, and θ is the phase delay. p The peak polarization angle is given by R. 2 =0.98. The data fusion module adopts a "weighted voting + deep learning" fusion strategy to weight the three-modal features, with phase delay having a weight of 0.35, specific heat capacity having a weight of 0.3, and peak polarization angle having a weight of 0.35. The input is a lightweight CNN (MobileNetV2), and the output is a fused feature vector. The fused feature vector is compared with the mycorrhizal-root feature database using cosine similarity matching to obtain a pseudo-color functional map of the mycorrhizal-root system. Mycorrhizal regions are marked in red, active root regions in blue, dormant root regions in yellow, and the soil background in black, to distinguish mycorrhizal-root systems based on structure, activity, and composition.

[0034] Specifically, the database pre-stores 256-dimensional fused feature vectors for 1000 sets of samples with known categories (covering crops such as wheat, corn, and soybeans), with each vector clearly labeled with a category label: Active mycorrhizal vector (label A): corresponds to "δ=0.6-1.0π rad, C=2.8-3.5 J / (g·℃), θ p The characteristic combination of =40°-45°”; Active root vector (label B): corresponds to "δ=0.3-0.5π rad, C=2.2-2.7 J / (g·℃), θ p The characteristic combination of =35°-39°”; Dormant root vector (label C): corresponds to "δ=0.2-0.4π rad, C=1.8-2.1 J / (g·℃), θ p The characteristic combination of =32°-34°”; Soil background vector (label D): corresponds to "δ<0.1π rad, C<1.5 J / (g·℃)", θ p The characteristic combination of "no fixed peak value".

[0035] During detection, the 256-dimensional fused feature vector (denoted as V) output by the MobileNetV2 network is compared with the cosine similarity of each labeled vector (denoted as Vi, i=1-1000) in the database. The formula is as follows: cosθ = V·Vi / ||V|| × ||Vi|| The numerator is the dot product of the two vectors, and the denominator is the product of the magnitudes of the two vectors. The calculation result is in the range of [0,1]. The closer the value is to 1, the more similar the features of the two vectors are.

[0036] Set a similarity threshold of ≥0.92 (the optimal threshold validated with 500 samples): If the similarity between the vector V to be classified and multiple vectors of a certain type of label (such as label A) in the database is all ≥0.92, and the average similarity is the highest, then the detection area corresponding to V is determined to be "active mycorrhizae" and marked in red. If the similarity with the vector of label B is the highest and ≥0.92, it is determined to be "active root system" and marked in blue. Similarly, matching label C is "dormant root system" (yellow), and matching label D is "soil background" (black), thus obtaining a pseudo-color image of mycorrhizae-root system. If the similarity of all categories is <0.92, the system automatically marks it as "area to be verified" and prompts manual review.

[0037] For example, in wheat sample testing, if the average similarity between the vector to be classified, V, and the "active mycorrhizae" class vector in the database is 0.95, and the similarity with other classes is <0.85, then the region is determined to be an active mycorrhizae and marked in red.

[0038] Optionally, the linearly polarized laser 10 is a 532nm linearly polarized laser, the wavelength of which is determined according to the characteristic response wavelength of mycelial chitin and root cellulose; the power of the linearly polarized laser 10 is 10-20mW, which can ensure that the mycelial survival rate is ≥97% after continuous irradiation for 30 minutes; the spectral linewidth is ≤1nm to ensure laser monochromaticity and reduce dispersion interference.

[0039] In this embodiment, the sample stage includes a platform, an x-axis linear module, a y-axis linear module, and a z-axis linear module to drive the sample stage to move along the x-axis, y-axis, and z-axis directions. This allows the first detector to collect light signals from different regions of the mycorrhizal-root system, and the second detector to collect light intensity from different regions of the mycorrhizal-root system. In this embodiment, the platform is a biocompatible transparent platform made of quartz glass. The structures of the x-axis linear module, y-axis linear module, and z-axis linear module can be referenced from existing technologies and will not be described in detail here.

[0040] The mycorrhizal-root three-modal synergistic imaging system provided in this invention, by setting up a linearly polarized laser, a first semi-transparent and semi-reflective mirror, a detection unit, an activity acquisition mechanism, and a processor, achieves three-dimensional synchronous analysis of the structure, activity, and composition of mycorrhizal-root systems, providing multi-dimensional quantitative data for the study of mycorrhizal-root symbiotic mechanisms.

[0041] Optionally, in one embodiment of the present invention, the detection unit 40 includes a first polarizer and a first detector. The first polarizer is movably disposed above the first semi-transparent mirror 30 and is used to filter out S-polarized light. The first detector is disposed above the first polarizer and is used to collect P-polarized light when the first polarizer is stationary, to collect S-polarized light when the first polarizer is moved away, and to collect θ-polarized light when the first polarizer rotates.

[0042] Specifically, in this embodiment, when it is necessary to collect P-polarized light, the S-polarized light emitted by the linearly polarized laser 10 is reflected by the first semi-transparent mirror 30 onto the mycorrhizal-root system. The mycorrhizal-root system reflects the S-polarized light, and the reflected light also includes both P-polarized and theta-polarized light. At this time, the first detector only collects S-polarized and P-polarized light, and does not collect theta-polarized light. When the S-polarized and P-polarized light pass through the first polarizer, the first polarizer filters out the S-polarized light, and the light signal collected by the first detector is pure P-polarized light. In this embodiment, the first polarizer is movable; when the first polarizer is removed, the light signal received by the first detector is both S-polarized and P-polarized light.

[0043] In this embodiment, the P-polarized light signal represents the mycorrhizal signal, and the S-polarized light signal represents the root signal. The processor is used to separate the P-polarized light signal and the S-polarized light signal. By testing the polarization reflectance of mycorrhizal chitin and root fiber cellulose of four crops, including wheat and corn, the following results were obtained: the reflectance of mycorrhizal chitin to P-polarized light is 85%-90%, while the reflectance of root fiber cellulose to P-polarized light is 35%-40%, with the mycorrhizal reflectance being 2.1-2.6 times that of the root fiber; the reflectance of mycorrhizae to S-polarized light is 3%-5%, while the reflectance of root fiber to S-polarized light is 15%-18%, with the root fiber reflectance being 3-6 times that of the hyphae.

[0044] In this embodiment, the mixed light signal = P-polarized light signal from mycorrhizae + S-polarized light signal from mycorrhizae + P-polarized light signal from root fibers + S-polarized light signal from root fibers. Since: the reflectivity of mycorrhizae for S-polarized light is extremely low (3%-5%), its contribution to the total S-polarized light signal in the mixed light is ≤15% (negligible); the reflectivity of root fibers for P-polarized light is extremely low (35%-40%), its contribution to the total P-polarized light signal in the mixed light is ≤30%, which can be offset through correlation model calibration. Therefore: The mixed optical signal - P-polarized optical signal ≈ S-polarized optical signal of root fibers (accounting for ≥85%), meets the requirements for precise separation.

[0045] When the processor receives the P-polarized light signal and the mixed light signal, the mixed light at this time refers only to the S-polarized light and the P-polarized light. The processor will perform signal preprocessing on the P-polarized light and the mixed light. The Otsu adaptive threshold segmentation (threshold T=0.3±0.05) will automatically filter out the S-polarized light noise that contributes weakly to the hyphae, ensuring that the purity of the final root fiber signal is ≥95%.

[0046] The following details the specific separation process of P-polarized light signals and S-polarized light signals.

[0047] Assuming the total light intensity incident on the mycorrhiza-root system is 100 μW, and the reflectance of both mycorrhizae and roots to the two polarized light types is taken as the median, i.e., mycorrhizae (chitin): 88% reflectance to P-polarized light (strong reflectance), and 4% reflectance to S-polarized light (extremely weak reflectance); root fibers: 38% reflectance to P-polarized light (weak reflectance), and 16% reflectance to S-polarized light (moderate reflectance). Assuming the detection area occupies half the area (50%:50%) for both mycorrhizae and roots, and there is no soil background.

[0048] The P-polarized light signal reflected by hyphae = total incident light intensity × mycorrhizal P-polarized light reflectance × mycorrhizal area ratio = 100μW × 88% × 50% = 44μW; The S-polarized light signal reflected by mycorrhizae = Total incident light intensity × Mycorrhizae S-polarized reflectivity × Mycorrhizae area ratio = 100μW × 4% × 50% = 2μW; Total mycorrhizal signal = 44 μW (P light) + 2 μW (S light) = 46 μW; The mycorrhizal signal mainly comes from P-polarized light (44 μW, accounting for 95.7%), while the S-polarized light signal is almost negligible (2 μW, accounting for 4.3%).

[0049] The P-polarized light signal reflected by the root system = total incident light intensity × root P-light reflectivity × root area ratio = 100μW × 38% × 50% = 19μW; The S-polarized light signal reflected by the root system = total incident light intensity × root S-polarized light reflectivity × root area ratio = 100μW × 16% × 50% = 8μW; Total root signal = 19 μW (P light) + 8 μW (S light) = 27 μW; The root system signal mainly comes from S-polarized light (8μW, accounting for 29.6%), while P-polarized light has a high proportion (19μW, accounting for 70.4%), but this will be distinguished later.

[0050] Mixed light signal = total mycorrhizal signal + total root signal = 46μW + 27μW = 73μW. The mixed light contains 44μW of mycorrhizal P light + 2μW of mycorrhizal S light + 19μW of root P light + 8μW of root S light.

[0051] Pure P light signal = Mycorrhizal P light signal + Root fiber P light signal = 44μW + 19μW = 63μW.

[0052] The mixed light signal - pure P-polarized light signal = 73μW - 63μW = 10μW. This 10μW is actually the "mycorrhizal S-polarized light signal + root S-polarized light signal", which is the total signal of all S-polarized light. In this embodiment, the mycorrhizal S-polarized light signal is 2μW, which is lower than the set threshold. The Otsu threshold segmentation algorithm in the processor will automatically filter out the mycorrhizal S-polarized light signal. The remaining 8μW light signal after filtering is the root S-polarized light signal.

[0053] Within the pure P-polarized light signal (63 μW), there is also 19 μW of polarized light reflected from the root system. This portion of polarized light can be subsequently calibrated using a correlation model to isolate the root system's contribution to the P-polarized light; that is, the P-polarized light signal is the pure mycorrhizal light signal. The mycorrhizal signal and the root signal can be separated using the above method.

[0054] Phase retardation reflects the difference in refractive index between mycorrhizal chitin and root cellulose. The formula for calculating phase retardation is as follows: δ=arctan[(I pmax -I pmin ) / (I s+pmax -I s+pmin )]×π / 2, where δ is the phase delay, I pmaxFor p-polarized light, I is the maximum value. pmin For the minimum value of p-polarized light, I s+pmax For the maximum value of the mixed light, I s+pmin This represents the minimum value of the mixed light. Correspondingly, the phase delay of the S-polarized light can also be calculated based on the maximum and minimum values ​​of the S-polarized light.

[0055] Mycorrhizal and root samples were collected, mycorrhizal diameters were calibrated, and mycorrhizal activity levels were determined using staining methods to establish a mycorrhizal sample database. A linear correlation model between phase lag and mycorrhizal diameter was obtained by fitting phase lag and mycorrhizal diameter using the least squares method. Based on this linear correlation model, different mycorrhizal diameters can be obtained for different phase lags. Correspondingly, a root sample database was established, and a linear correlation model between phase lag and root diameter was obtained by fitting phase lag and root diameter using the least squares method. Based on this linear correlation model, different root diameters can be obtained for different phase lags, thus providing clear outlines of mycorrhizae and roots on the pseudocolor functional map.

[0056] Next, the first polarizer is rotated, and the mixed light is filtered out after transmission through the first polarizer, leaving only theta-polarized light. The first detector collects the intensity of theta-polarized light and sends it to the processor. The processor generates a polarization angle spectrum data map based on the intensity of theta-polarized light and the polarization angle of the first polarizer. For chitin (mycorrhizal hyphae) and cellulose (plant roots), chitin molecules are arranged in a helical pattern, with a peak polarization angle θ. p ≈45°, spectral width Δθ≈15°, polarization extinction ratio R≥5.0; cellulose molecules are arranged in parallel, peak polarization angle θ p ≈90°, spectral width Δθ≈25°, polarization extinction ratio R≤3.0; through θ p The combined threshold with R can distinguish mycorrhizal and root systems by their composition.

[0057] In this embodiment, the first polarizer is disposed on the optical support 60, the optical support 60 is provided with a rotating shaft, the first polarizer is connected to the rotating shaft, the rotating shaft can rotate relative to the optical support 60, the rotating shaft is connected to a rotary motor, and when the rotary motor rotates, it can drive the rotating shaft to rotate, thereby driving the first polarizer to rotate; at the same time, the first polarizer can also move along the length direction of the rotating shaft so that the first polarizer is staggered from the first semi-transparent and semi-reflective mirror 30.

[0058] Optionally, such as Figure 2As shown, in another embodiment of the present invention, the detection unit 40 includes: a second semi-transparent mirror 41, a first detection component, and a second detection component. The second semi-transparent mirror 41 is disposed above the first semi-transparent mirror 30, the first detection component is disposed above the second semi-transparent mirror 41, and the first detection component is used to collect p-polarized light. The second detection component is disposed on one side of the second semi-transparent mirror, and the second detection component is used to collect θ-polarized light.

[0059] Specifically, the mixed light transmitted through the first semi-transparent and semi-reflective mirror 30 enters the second semi-transparent and semi-reflective mirror 41, and the second semi-transparent and semi-reflective mirror 41 transmits part of the light to the first detection component and reflects part of the light to the second detection component.

[0060] In this embodiment, the first detection component includes a second polarizer 42 and a second detector. The second polarizer 42 is movably disposed above the second semi-transparent mirror 41 and is used to filter out S-polarized light. The second detector is disposed above the second polarizer 42 and is used to collect P-polarized light.

[0061] Specifically, in this embodiment, the second polarizer 42 is movable. The mixed light reflected by the mycorrhiza-root system is transmitted through the first semi-transparent mirror 30 and then through the second semi-transparent mirror 41. After passing through the second polarizer 42, the S-polarized light in the mixed light is filtered out, and the second detector collects pure P-polarized light. When the second polarizer 42 is removed, the light collected by the second detector includes both S-polarized and P-polarized light. Subtracting the P-polarized light yields pure S-polarized light. In this embodiment, the second polarizer 42 can be mounted on a guide rail and connected to a driver. The driver can move the second polarizer 42 along the guide rail. Optionally, the driver can be a hydraulic cylinder, a pneumatic cylinder, or a linear motor.

[0062] like Figure 2 As shown, in an embodiment of the present invention, the second detection component includes a third polarizer 43 and a third detector. The third polarizer 43 is rotatably disposed on one side of the second semi-transparent mirror 41, and is used to filter out S-polarized light. The third detector is disposed on one side of the third polarizer 43, and is used to collect θ-polarized light.

[0063] Specifically, the mixed light reflected by the mycorrhiza-root system is transmitted through the first semi-transparent mirror 30 and then reflected by the second semi-transparent mirror 41. The reflected light enters the third polarizer 43, which filters out the S-polarized light. In this embodiment, the third polarizer 43 gradually rotates from 0° to 180°. During this rotation, the third detector collects the intensity of the θ-polarized light and sends the intensity with the highest clarity, along with the polarization angle of the third polarizer 43 at that moment, to the processor. The processor generates a polarization angle spectrum based on this intensity and polarization angle. In this embodiment, the rotation angle of the third polarizer 43 is the polarization angle, and the third polarizer 43 can be rotated by a motor.

[0064] In an embodiment of the present invention, the first semi-transparent mirror 30 and the second semi-transparent mirror 41 can be a single mirror body, with a reflective film coated on the side facing the linearly polarized laser. When light passes through the semi-transparent mirror, 50% of the light is reflected and 50% is transmitted.

[0065] Optionally, in embodiments of the present invention, the detector can be a CMOS or CCD imaging camera with ≥2 million pixels, a frame rate ≥30fps, a response wavelength of 400~1100nm, supporting light intensity signal imaging acquisition, and having a built-in image magnification and noise reduction module for capturing the intensity of polarized light modulated by biological tissue.

[0066] In an embodiment of the present invention, the mycorrhizal-root trimodal collaborative imaging system further includes a temperature sensor and a humidity sensor. The temperature sensor is used to detect the ambient temperature, the humidity sensor is used to detect the ambient humidity, and the processor is used to compensate for the spectral capacity when the difference between the current ambient temperature and the initial ambient temperature is greater than a first preset value, and to compensate for the spectral width when the difference between the current ambient humidity and the initial ambient humidity is greater than a second preset value.

[0067] Specifically, in this embodiment, the temperature sensor has a detection range of -40℃ to 125℃, and the humidity sensor has a detection range of 0% to 100%. When collecting the light signal reflected by mycorrhizae and roots, if the difference between the current ambient temperature and the initial temperature is > ±2℃, it is calculated according to C0 = C meas +0.02(T-25) is used to compensate for the specific heat capacity, where C0 is the compensated specific heat capacity, and C meas The specific heat capacity is calculated based on the heat flux density and the heating area, where T is the current temperature. When the difference between the current humidity and the initial humidity of the environment is > ±5% RH, it is calculated according to Δθ0 = Δθ meas -0.01×(RH-30) is used to compensate for the spectral width, where θ0 is the compensated spectral width, and θ measThe spectral width is obtained based on the polarization angle spectrum, and RH represents the current humidity. In this embodiment, the processor automatically calibrates every 5 seconds to ensure that the system's recognition accuracy fluctuates by ≤2% when the ambient temperature fluctuates by ±5℃ and the ambient humidity fluctuates by ±10% RH.

[0068] like Figure 3 As shown, in an embodiment of the present invention, the activity acquisition mechanism 50 includes: multiple thermostatic heaters 51, multiple heat sinks 52, and a thermal infrared camera 53. The multiple thermostatic heaters 51 are arranged around the sample stage 20 and are used to heat the mycorrhizal-root system. Each heat sink 52 is used to dissipate heat from one thermostatic heater 51, and the heat sink 52 is turned on after the thermostatic heater 51 is turned off to rapidly reduce the temperature of the thermostatic heater 51. The thermal infrared camera 53 is positioned above and opposite the sample stage 20 and is used to acquire the heat flux density of biological tissue. A processor is electrically connected to the thermal infrared camera 53 and is used to obtain the specific heat capacity of the mycorrhizal-root system based on the heat flux density acquired by the thermal infrared camera 53 and the heating area.

[0069] Specifically, in this embodiment, the constant temperature heater 51 has an airflow channel, and a heating element is installed inside the constant temperature heater 51. When the heating element is running, the airflow channel is closed. When the temperature inside the constant temperature heater 51 reaches the target temperature, the airflow channel opens to rapidly heat the biological tissue, shorten the heating time of the biological tissue, and maintain the biological activity of the biological tissue. After the constant temperature heater 51 is turned on, the thermal infrared camera 53 begins to collect the heat flux density of the mycorrhizal-root system. After the thermal infrared camera 53 has finished collecting data, the constant temperature heater 51 is turned off, and the radiator 52 is turned on to rapidly cool the constant temperature heater 51. The processor calculates the specific heat capacity C of the mycorrhizal-root system based on the heat flux density and the heating area, wherein the formula for calculating the specific heat capacity C is: C = Q / (m·ΔT), where Q is calculated by heat flux integration: Q = ∫q·S·dt, q is the heat flux density, S is the heating area, t is the heating time after the airflow channel is opened, the integral is the average value of the heat flux density variation range, m is the mass of the mycorrhizal-root system, and ΔT is the temperature change of the mycorrhizal-root system. In this embodiment, the heating area is the effective heat radiation coverage area of ​​the constant temperature heater 20, which is the factory data of the constant temperature heater 20.

[0070] In this embodiment, the thermal radiation coverage of multiple constant temperature heaters 51 is ≥ 120% of the sample stage area, and the heating uniformity error is ≤ ±5%, in order to avoid local overheating that could lead to tissue damage.

[0071] Optionally, in this embodiment, the thermal infrared camera has a resolution of ≥300,000 pixels, a field of view of ≥60°, and a conventional detection accuracy of ≤0.01W / cm². 2For tiny samples (microbial biopsy films, minimally invasive biopsy tissues), a "micro-acquisition mode" is adapted, improving detection accuracy to ≤0.008W / cm². 2 The response time is ≤10ms. The thermal infrared camera is mounted directly above the sample stage, with a field of view that completely covers the sample stage. For large crop leaves (≤15cm×10cm), the thermal infrared camera can extend the field of view to ≥80°.

[0072] The radiator 52 uses 1-2 sets of silent brushless high-power fans, which are installed on the back or side of the constant temperature heater 51. The air duct is directly facing the constant temperature heater 51. The normal air speed is ≥3m / s. The radiator 52 is turned off when the temperature of the constant temperature heater 51 drops to within room temperature +3℃.

[0073] In an embodiment of the present invention, the constant temperature heater 51 includes a housing 511, louvers 512, and a heating element. The heating element is disposed inside the housing 511, and the louvers 512 are disposed within the housing 511. When the louvers 512 are open, they form multiple airflow channels that communicate with the interior of the housing 511 and are directed towards the biological tissue. A radiator 52 is disposed on the back or side of the housing 511, i.e., the radiator 52 is not opposite to the louvers 512. Before heating the biological tissue, the louvers 512 are closed, and the heating element is turned on. When the temperature inside the housing 511 reaches the target temperature of the biological tissue, the louvers 512 are opened to rapidly heat the biological tissue. After the thermal infrared camera 53 has finished collecting data, the louvers 512 are closed, and the radiator 52 is turned on to cool the housing 511, achieving rapid cooling.

[0074] Optionally, in embodiments of the present invention, the heating element can be an infrared heating lamp with a heating power of 5-20W, and the lamp body can be made of tungsten wire alloy. The 512 louver blades are made of mirror stainless steel or high-reflectivity aluminum foil with a reflectivity ≥90%.

[0075] In this embodiment, the acquisition of the pseudo-color image is related to the positions of the first detection component, the second detection component, and the active acquisition mechanism 50. In order to avoid the misalignment of multimodal data caused by differences in the installation position and detector distortion of the first detection component, the second detection component, and the active acquisition mechanism 50, such as the mycorrhizal-root system in the same physical location having coordinate deviations in different modal images, a spatial consistency basis is provided for subsequent cross-modal data fusion.

[0076] Specifically, a 12×9 checkerboard calibration plate (square size 1mm, accuracy ±0.01mm) was selected and placed on the sample stage 20. By adjusting the three-axis linear module of the sample stage 20, the calibration plate was ensured to be simultaneously at the center of the field of view of the first detection component, the second detection component, and the active acquisition mechanism 50. The tilt angle (0°-15°) and height (20-25cm) of the calibration plate were finely adjusted by the sample stage 20, and 15 sets of calibration plate images in different postures were acquired (15 sets for each module), with a uniform image resolution of 2048×2048 pixels. The corner points of the calibration plate were extracted using the Zhang Zhengyou algorithm, and the intrinsic parameters of each detector (focal length f, principal point coordinates (u0, v0), distortion coefficients k1 / k2 / p1 / p2) were calculated, where the focal length error of the first detector was ≤0.1mm and the distortion coefficient was ≤0.001. Using the coordinate system of the first detector as the world coordinate system, the extrinsic parameters (rotation matrix R, translation vector T) of the active acquisition mechanism 50 and the second detector relative to the first detector are calculated. The rotation error is ≤0.05°, and the translation error is ≤0.2mm. Using the calculated intrinsic parameters, radial and tangential distortion corrections are performed on the original images of the active acquisition mechanism 50 and the second detector assembly. The distortion error of the corrected image is ≤0.05%. Based on the extrinsic parameters, a coordinate mapping matrix M (3×3 matrix) is generated. The pixel coordinates (x,y) of the corrected specific heat capacity image and polarization angle spectrum image are converted into unified coordinates (x',y') of the first detector assembly through the coordinate mapping matrix M, with a mapping accuracy ≤1μm.

[0077] Spatial matching error refers to the Euclidean distance between the coordinates (x1, y1) of the same physical marker point (such as the corner of the calibration plate or the mycorrhizal branching point) in the polarization state image and its coordinates (x2, y2) in the specific heat capacity image / polarization angle spectrum image after mapping, i.e., error d = √[(x1-x2)]. 2 +(y1-y2) 2 ].

[0078] A standard calibration plate (containing 108 corner points) was selected, and the same corner point was marked in different module images. The coordinate deviation was calculated. Wheat mycorrhizal samples (containing 20 clear bifurcation points) were selected, and the detection was repeated 10 times. The coordinate mapping deviation of the bifurcation points was statistically analyzed.

[0079] The verification results are as follows: the average matching error of the calibration plate corner points is ≤1.2μm, and the maximum error is ≤2.5μm; the average matching error of the mycorrhizal branching points is ≤1.8μm, and the maximum error is ≤2.8μm; considering both verification scenarios, the maximum error upper limit is taken as ≤3μm to ensure spatial consistency in actual detection.

[0080] Furthermore, in an embodiment of the present invention, the mycorrhizal-root trimodal synergistic imaging system further includes a light shield, which is disposed outside the linearly polarized laser 10, the detection unit 40, the activity acquisition mechanism 50, and the sample stage 20. In this embodiment, the light shield is used to suppress ambient light and electromagnetic interference. The light shield is made of black light-absorbing material and also has a circuit anti-interference shielding layer.

[0081] Furthermore, in an embodiment of the present invention, the mycorrhizal-root trimodal collaborative imaging system also includes a display screen. Both the display screen and the processor are disposed outside the light shield. The display screen is disposed on the processor and is used to display a pseudo-color image.

[0082] like Figure 4 As shown, the horizontal axis represents the typical ambient temperature in farmland (15℃-35℃), and the vertical axis represents the accuracy rate of mycorrhizal-root identification, covering all scenarios including low temperature (15℃), suitable temperature (25℃), and high temperature (35℃). The red solid line represents the imaging system provided in this embodiment of the invention, with an accuracy rate of ≥98% across the entire temperature range and a fluctuation range of only ≤1%. The accuracy rate in the conventional field temperature range of 20-30℃ is ≤97.5%, with no significant performance degradation. The blue dashed line represents the accuracy curve of the dual-modal (structure + activity) system, and the black dashed line represents the accuracy curve of the single-modal (composition) system. For every 5℃ temperature fluctuation, the accuracy rate of both the dual-modal and single-modal systems decreases by 3%-5%, with a fluctuation range of up to 8% across the entire range, making them unsuitable for complex field temperature environments.

[0083] This invention also provides a mycorrhizal-root three-modal synergistic imaging method, which specifically includes the following steps: Step 100: Obtain the phase delay of P-polarized and S-polarized light, the heat flux density of mycorrhizal-root system, the light intensity of θ-polarized light, and the polarization angle of the detection unit. Calculate the specific heat capacity of the mycorrhizal-root system based on the heat flux density and heating area. Generate a polarization angle spectrum based on the light intensity of θ-polarized light and the polarization angle of the detection unit. Obtain the peak polarization angle and spectral width based on the polarization angle spectrum. Step 102: Establish a quantitative correlation model between specific heat capacity, phase delay, and peak polarization angle. Employ a weighted voting + deep learning fusion strategy, weighting the heat capacity, phase delay, and peak polarization angle. Input the weighted data into the MobileNetV2 network to obtain a fused feature vector. Step 103: Compare the fused feature vector with the mycorrhizal-root system feature database to obtain a pseudo-color image of the mycorrhizal-root system.

[0084] Specifically, the phase delay of P-polarized light is calculated based on the optical signals of P-polarized light and mixed light, and the phase delay of S-polarized light is calculated based on the optical signals of S-polarized light and mixed light. The specific heat capacity of mycorrhizal-root system is calculated based on the heat flux density and heating area of ​​mycorrhizal-root system. A polarization angle spectrum is generated based on the light intensity of θ-polarized light and the polarization angle of the detection unit, and the peak polarization angle and spectral width are obtained based on the polarization angle spectrum.

[0085] Phase delay, specific heat capacity, and peak polarization angle were normalized and denoised. The data fusion module was trained using 500 sets of mycorrhizal-root samples to establish a quantitative correlation model between specific heat capacity, phase delay, and peak polarization angle, C = 0.28δ + 0.45θ. p +0.07, where C is the specific heat capacity, δ is the phase delay, and θ is the phase delay. p The peak polarization angle is given by R. 2 =0.98. The data fusion module adopts a "weighted voting + deep learning" fusion strategy to weight the three modal features, with phase delay having a weight of 0.35, specific heat capacity having a weight of 0.3, and peak polarization angle having a weight of 0.35. The input is a lightweight MobileNetV2, and the output is a fused feature vector. The fused feature vector is compared with the mycorrhizal-root feature database using cosine similarity matching to obtain a pseudo-color image of the mycorrhizal-root system. Mycorrhizal regions are marked in red, active root regions in blue, dormant root regions in yellow, and the soil background in black, to distinguish mycorrhizal-root systems based on structure, activity, and composition.

[0086] The mycorrhizal-root three-modal synergistic imaging method provided in this invention realizes three-dimensional synchronous analysis of mycorrhizal-root structure, activity and composition, providing multi-dimensional quantitative data for the study of mycorrhizal-root symbiosis mechanism.

[0087] Furthermore, in an embodiment of the present invention, before the step of establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle, the imaging method further includes: acquiring the current temperature of the environment, comparing the current temperature with the initial temperature, and compensating for specific heat capacity when the difference between the current temperature and the initial temperature is greater than a first preset value.

[0088] Furthermore, before establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle, the imaging method also includes: acquiring the current humidity of the environment; comparing the current humidity with the initial humidity; and compensating for the spectral width when the difference between the current humidity and the initial humidity is greater than a second preset value.

[0089] Specifically, the temperature sensor has a detection range of -40℃ to 125℃, and the humidity sensor has a detection range of 0% to 100%. When collecting the light signal reflected by mycorrhizae and roots, if the difference between the current ambient temperature and the initial temperature is > ±2℃, it is calculated according to C0 = C meas +0.02(T-25) is used to compensate for the specific heat capacity, where C0 is the compensated specific heat capacity, and C meas The specific heat capacity is calculated based on the heat flux density and the heating area, where T is the current temperature. When the difference between the current humidity and the initial humidity of the environment is > ±5% RH, it is calculated according to Δθ0 = Δθ meas -0.01×(RH-30) is used to compensate for the spectral width, where θ0 is the compensated spectral width, and θ meas The spectral width is obtained from the polarization angle spectrum, and RH is the current humidity.

[0090] Table 1 compares the mycorrhizal-root trimodal imaging system provided in the embodiments of the present invention with the single-modal system.

[0091] Table 1: Comparison Dimensions Trimodal system Polarization state single mode Specific heat capacity single mode Polarization angle spectrum single mode Information Dimensions <![CDATA[Structure (δ) + Activity (C) + Component (θ p )]]> Structure only (δ) Activity only (C) <![CDATA[Only ingredients (θ p > Mycorrhizal / root differentiation accuracy ≥98% 82% 75% 85% Active mycelium recognition rate 96% Unable to identify (no active dimension) 80% (Boundary blurred) Unable to identify (no active dimension) Environmental adaptability (fluctuation within ±5℃) Accuracy fluctuation ≤2% Accuracy fluctuation ≥6% Accuracy fluctuation ≥10% Accuracy fluctuation ≤3% Single sample detection time ≤8 seconds 2 seconds 5 seconds 3.6 seconds As shown in Table 1, the mycorrhizal-root trimodal imaging system provided in this embodiment of the invention significantly improves the accuracy of mycorrhizal-root differentiation and the identification rate of active hyphae compared with the single-modal system. Through "trimodal cross-calibration", the accuracy fluctuation is ≤2% when the environmental fluctuation is ±5℃ and ±10% RH, which meets the long-term monitoring needs of the field.

[0092] Table 2 compares the mycorrhizal-root trimodal imaging system with the dual-modal system provided in the embodiments of the present invention.

[0093] Table 2: Comparison Dimensions Trimodal system Polarization state + specific heat capacity dual mode Polarization state + angular spectrum dual mode Specific heat capacity + angular spectrum dual mode Spatiotemporal synchronization error Time ≤ 10ms, space ≤ 3μm Time ≥ 50ms, Space ≥ 8μm Time ≥ 30ms, space ≤ 5μm Time ≥ 40ms, Space ≥ 7μm Data fusion depth <![CDATA[Quantitative association model (R 2 = 0.98)]]> Pixel overlay (unrelated) <![CDATA[Feature splicing (R 2 = 0.82)]]> <![CDATA[Feature splicing (R 2 = 0.78)]]> Environmental calibration mechanism Three-modal cross-calibration No calibration (dependent on hardware stability) Dual-modal unidirectional calibration Dual-modal unidirectional calibration Symbiotic interface parsing capabilities It can distinguish between active / dormant mycelia and root areas. Unable to distinguish between active / dormant hyphae Unable to distinguish between active / dormant hyphae Unable to locate hyphal morphological boundaries Long-term monitoring stability (24 hours) Accuracy rate maintained at ≥96% Accuracy dropped to 85% Accuracy dropped to 88%. Accuracy dropped to 82%. The following describes the process of conducting field measurements on wheat mycorrhizal-root systems using the mycorrhizal-root trimodal imaging system provided in this embodiment of the invention. (Time ≤ 10 minutes) (1) Sample preparation (takes 2 minutes) Select wheat seedlings (30 days old, inoculated with rhizobium, mycorrhizal infection rate about 60%), carefully peel off the soil from the roots (retaining native soil particles, moisture content 25%), gently rinse the root surface with clean water, absorb the surface moisture with filter paper, and fix it on the glass slide of sample stage 20.

[0094] (2) Parameter settings (1 minute) The start-up time difference between the second and third detection components and the active acquisition mechanism is ≤10ms; polarization state acquisition takes 2 seconds; angle spectrum acquisition takes 3.6 seconds; and thermal capacity acquisition takes 5 seconds. Automatic calibration threshold (calibration is initiated when temperature fluctuation >2℃ or humidity fluctuation >5% RH).

[0095] (3) System operation (time taken 5 minutes) All components are preheated to a stable state. The temperature sensor collects the initial ambient temperature, and the humidity sensor collects the initial ambient humidity. (T0=25℃, RH0=50% RH); data is collected synchronously to generate δ, C, and θ. p Raw data; ambient temperature fluctuation ≤1℃, no additional calibration required; data preprocessed and validated using correlation models (R). 2 =0.98), algorithm fusion, output recognition result.

[0096] (4) Results analysis (2 minutes) The display screen clearly shows the red active mycorrhizal area, the blue active root area, and the yellow dormant root area, with no interference from the soil background.

[0097] Quantitative report: Mycorrhizal infection rate 59.8% (actual value 60%, deviation 0.2%); active root system percentage 78.5% (manual count, actual value 79%, deviation 0.5%); average C value of symbiotic area 2.9 J / (g·℃).

[0098] Process log: Records acquisition time, environmental parameters, calibration record (no calibration, data qualified), and identification confidence level (98.5%).

[0099] Ten wheat samples (mycorrhizal infection rate 30%-80%) were selected and tested using this system, single-modal, dual-modal, and manual counting methods (staining method, gold standard). Table 3 shows the comparison of test results.

[0100] Table 3: Detection methods Deviation between measured and true infection rate Detection rate of active mycelia Detection time Trimodal system ≤1% 96.5% 5 minutes Polarization state single mode ≤5% Unable to detect 2 minutes Specific heat capacity single mode ≤8% 80.2% 5 minutes Polarization state + angular spectrum dual mode ≤3% Unable to detect 5.6 minutes Manual counting method (gold standard) ≤0.5% 98% 45 minutes Conclusion: The mycorrhizal-root trimodal synergistic imaging system provided in this embodiment of the invention has a detection accuracy close to the gold standard, takes only 1 / 9 of the time of manual counting, and can simultaneously detect the activity state.

[0101] The recognition accuracy of the same wheat sample was tested at temperatures ranging from 15℃ to 35℃ (5℃ intervals) and humidity ranging from 30% to 60% (10% RH intervals). Table 4 shows the recognition accuracy of the same wheat sample.

[0102] Table 4: Environmental conditions (temperature / humidity) Accuracy of this system Polarization state + specific heat capacity dual-mode accuracy Deviation (this system - bimodal) 15℃ / 30%RH 98.5% 90.2% +8.3% 25℃ / 50% RH (Baseline) 98.2% 89.8% +8.4% 35℃ / 60%RH 96.8% 82.5% +14.3% As shown in the table above, the greater the environmental fluctuation, the more significant the advantages of the mycorrhizal-root three-modal collaborative imaging system provided in this embodiment of the invention, and the three-modal cross-calibration can effectively cancel out interference.

[0103] Table 5 compares the core indicators of the mycorrhizal-root trimodal synergistic imaging system provided in this embodiment of the invention with the commercial "hyperspectral + thermal imaging" system.

[0104] Table 5: index This system Commercial dual-modal system Advantage multiple Single sample detection time 5 minutes 20 minutes 4 times Mycorrhizal-root differentiation accuracy 98.2% 85.3% 1.15 times Field portability Weight 5kg Weight 30kg 6 times It should be noted that the single-sample testing time includes the entire process of sample preparation, data acquisition, calibration and fusion, and result analysis. The total single-sample acquisition time of ≤8 seconds mentioned in this embodiment specifically refers to the simultaneous acquisition of the three modules; there is no contradiction between the two.

[0105] As can be seen from the table above, the mycorrhizal-root three-modal synergistic imaging system provided in this embodiment of the invention has significant advantages in efficiency, accuracy and portability, and fully meets the needs of field scientific research and production.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A mycorrhizal-root three-modal synergistic imaging system, characterized in that, include: Linearly polarized lasers are used to emit S-polarized light; A sample stage, which is movable, is used to place mycorrhizal-root systems. A first semi-transparent and semi-reflective mirror is disposed opposite to the linearly polarized laser. The first semi-transparent and semi-reflective mirror is used to reflect S-polarized light to mycorrhizal-root system and transmit mixed light reflected by mycorrhizal-root system, wherein the mixed light includes S-polarized light, P-polarized light and θ-polarized light. The detection unit is used to collect the S-polarized light, the P-polarized light, and the θ-polarized light; An active collection mechanism for collecting the heat flux density of mycorrhizae-root systems; The processor is used to calculate the phase delay of the P-polarized light and the S-polarized light, and to obtain the specific heat capacity of the mycorrhizal-root system based on the heat flux density and the heating area. The processor is also used to generate a polarization angle spectrum based on the light intensity of the θ-polarized light and the polarization angle of the detection unit, and to obtain the peak polarization angle and spectral width based on the polarization angle spectrum. The processor is also used to establish a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle. It adopts a weighted voting + deep learning fusion strategy to weight the specific heat capacity, phase delay, and peak polarization angle, and inputs the weighted data into the MobileNetV2 network to obtain a fused feature vector. The processor is also used to compare the fused feature vector with the mycorrhizal-root feature database to obtain a pseudo-color image of the mycorrhizal-root system.

2. The mycorrhizal-root three-modal synergistic imaging system according to claim 1, characterized in that, The detection unit includes: A first polarizer is movably disposed above the first semi-transparent and semi-reflective mirror. The first polarizer is used to filter out S-polarized light in the mixed light. A first detector is disposed above the first polarizer. The first detector is used to collect the P-polarized light when the first polarizer is stationary. The first detector is also used to collect the S-polarized light when the first polarizer is moved away. The first detector is also used to collect the θ-polarized light when the first polarizer is rotated.

3. The mycorrhizal-root three-modal synergistic imaging system according to claim 1, characterized in that, The detection unit includes: The second semi-transparent and semi-reflective mirror is positioned above the first semi-transparent and semi-reflective mirror; A first detection component is disposed above the second semi-transparent mirror, and the first detection component is used to collect the P-polarized light and the S-polarized light; The second detection component is disposed on one side of the second semi-transparent mirror, and the second detection component is used to collect the θ-polarized light.

4. The mycorrhizal-root three-modal synergistic imaging system according to claim 3, characterized in that, The first detection component includes: A second polarizer is movably disposed above the second semi-transparent mirror, and the second polarizer is used to filter out S-polarized light; The second detector is positioned above the second polarizer and is used to collect the P-polarized light and the S-polarized light.

5. The mycorrhizal-root three-modal synergistic imaging system according to claim 3, characterized in that, The second detection component includes: A third polarizer is rotatably disposed on one side of the second semi-transparent mirror, and the third polarizer is used to filter out S-polarized light; A third detector is disposed on one side of the third polarizer, and the third detector is used to collect the θ-polarized light.

6. The mycorrhizal-root three-modal synergistic imaging system according to claim 1, characterized in that, Also includes: A temperature sensor is used to detect the ambient temperature, and the temperature sensor is electrically connected to the processor; A humidity sensor is used to detect ambient humidity, and the humidity sensor is electrically connected to the processor; The processor is further configured to compensate for the specific heat capacity when the difference between the current temperature and the initial temperature of the environment is greater than a first preset value, and to compensate for the spectral width when the difference between the current humidity and the initial humidity of the environment is greater than a second preset value.

7. The mycorrhizal-root three-modal synergistic imaging system according to claim 1, characterized in that, The active acquisition mechanism includes: Multiple thermostatic heaters are arranged around the sample stage. The thermostatic heaters are used to heat the mycorrhizal-root system after their own temperature reaches the target temperature of the mycorrhizal-root system. Multiple radiators, each of which is used to dissipate heat from one of the constant temperature heaters, and the radiator is turned on after the constant temperature heater is turned off; A thermal infrared camera is positioned above and opposite the sample stage. The thermal infrared camera is used to collect the heat flux density of the mycorrhizal-root system. The thermal infrared camera is electrically connected to the processor.

8. The mycorrhizal-root three-modal synergistic imaging system according to claim 7, characterized in that, The constant temperature heater includes: The housing, with the air outlet of the radiator facing the housing; A louver is provided in the housing. When the louver is opened, it forms multiple airflow channels. The airflow channels are connected to the interior of the housing and are directed toward the mycorrhizal-root system. A heating element is disposed within the housing.

9. An imaging method based on the mycorrhizal-root three-modal synergistic imaging system according to any one of claims 1-8, characterized in that, include: The phase delay of P-polarized light and S-polarized light, the heat flux density of mycorrhizal-root system, the light intensity of θ-polarized light and the polarization angle of the detection unit are obtained. The specific heat capacity of mycorrhizal-root system is obtained based on the heat flux density and the heating area. A polarization angle spectrum is generated based on the light intensity of θ-polarized light and the polarization angle of the detection unit. The peak polarization angle and spectral width are obtained based on the polarization angle spectrum. A quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle is established. A weighted voting + deep learning fusion strategy is adopted to weight specific heat capacity, phase delay, and peak polarization angle. The weighted data is then input into the MobileNetV2 network to obtain the fused feature vector. The fused feature vector is compared with the mycorrhizal-root feature database to obtain a pseudo-color image of the mycorrhizal-root system.

10. The imaging method according to claim 9, characterized in that, Before establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle, the imaging method further includes: Get the current ambient temperature; Compare the current temperature with the initial temperature; When the difference between the current temperature and the initial temperature is greater than a first preset value, the specific heat capacity is compensated.

11. The imaging method according to claim 9, characterized in that, Before establishing a quantitative correlation model of specific heat capacity, phase delay, and peak polarization angle, the imaging method further includes: Obtain the current humidity of the environment; Compare the current humidity with the initial humidity; When the difference between the current humidity and the initial humidity is greater than a second preset value, the spectral width is compensated.

Images