Underwater sediment particle size measuring system based on laser light curtain imaging

Abstract

The invention discloses an underwater sediment particle size measurement system based on laser light curtain imaging. The underwater sediment particle size measurement system comprises an upper computer, a waterproof metal isolation bin, and a laser light curtain irradiation module, a microscopic optical imaging module and an image sensing module which are mounted in the waterproof metal isolation bin, an optical window is arranged on the waterproof metal isolation bin; the laser light curtain irradiation module is used for generating a stable and uniform laser light curtain, outputting the laser light curtain from an optical window, and exposing and displaying to-be-detected sediment particles; the microscopic optical imaging module is used for realizing high-definition dark field imaging of sediment particles in a laser light curtain plane; the image sensing module is used for acquiring an image of the sediment particles to be detected in real time and uploading the image to the upper computer; and the upper computer is used for processing the to-be-measured sediment particle image so as to measure the sediment particle size. The invention provides a new solution for real-time measurement of the particle size of underwater sediment particles, and has a great application prospect.

Description

Technical Field

[0001] The invention belongs to the technical field of underwater sediment particle size measurement, in particular to an underwater sediment particle size measurement system based on laser light curtain imaging. Background Art

[0002] Particle size measurement methods are usually benchmarked based on the use environment, the target to be measured, and the technical level. Typical methods include screening method, particle size meter method, laser particle size method, Coulter method, and image method. Different methods have different measurement characteristics, and the specific measurement method needs to be determined based on multiple factors.

[0003] Sieving is a classic particle size measurement method. With the advancement of the Industrial Revolution from the late 18th century to the early 19th century, it was widely used in the mining and construction industries. In the 19th century, unified screening standards were established by the International Organization for Standardization and other organizations. The principle is to screen particles of varying size using sieves of varying mesh sizes or a dedicated screening machine to intuitively determine the particle size range. This method is easy to use, intuitive, and low-cost, but it relies on manual operation and can significantly affect the accuracy of the results when multiple particles cling or the sieve is damaged.

[0004] The particle size counter method, also known as the sedimentation method, began to be used to measure particle size following the introduction of Stokes' law in 1851. This method relies on the physical properties of sediment particles. Because different particles experience different forces in water, when the forces of gravity and buoyancy are balanced, the particles are located in different layers within the liquid. Physical calculations reveal the particle size distribution of these layers. While the sedimentation method is simple to use and provides relatively accurate results, it is time-consuming, performs poorly with clinging particles, and can only measure particles within a certain size range.

[0005] The Coulter method, also known as the resistance method, was developed in 1949 by Wallace Coulter, who discovered that blood cells cause a change in resistance when passing through micropores. Based on this discovery, he proposed the Coulter method for particle size measurement. When a non-conductive particle suspended in a conductive electrolyte passes through a tiny pore, it displaces an amount of electrolyte equal to its volume, causing an instantaneous increase in resistance between the two electrodes and generating a voltage pulse. This impedance change is then used to directly measure the particle volume. While this method offers the advantages of high precision and a wide range, it cannot be used for real-time measurement in underwater environments.

[0006] Laser scattering. In the late 1970s, BJ Berne and GD Harp demonstrated the feasibility of particle size measurement using dynamic light scattering. Laser particle size measurement has become a mainstream method for fine particle measurement. The laser particle size method is based on the scattering of laser light. A sample is suspended in a sample cell. When the laser passes through the sample and strikes the particle surface, light scattering occurs. Different particles have different diameters and scattering angles, resulting in varying amounts of scattered light. Through light amplification and analog-to-digital conversion, specific particle size parameters can be determined. While the laser particle size method provides relatively accurate results for micron-sized particles, it exhibits significant errors for particles larger than millimeters. It also has high environmental requirements and requires strict control of measurement parameters.

[0007] Imaging, also known as microscopy, is a measurement method based on a microscope. In 1931, Ernst Ruska and Max Knoll developed the first prototype of an electron microscope at the Technical University of Berlin in Germany, and by 1934, they had clearly documented tasks related to particle detection. From the 1980s to the 1990s, optical microscopes were combined with image processing technology to form the first digital microscope systems. Computers captured microscopic images and used image processing methods to analyze particle size and morphology, greatly improving the automation and accuracy of particle measurement. However, imaging methods are limited by the microscope equipment required and cannot measure in water. Summary of the Invention

[0008] The purpose of the present invention is to address the problems existing in the above-mentioned prior art and provide an underwater sediment particle size measurement system that meets the requirements of underwater measurement environment and measurement real-time and accuracy. It can realize particle chip selection and measurement of fixed depth of field measurement layer in an underwater measurement environment with weak background light, many noise points and different depth of field of sediment particles, avoid the influence of interference items such as microorganisms and bubbles in water on particle size analysis, and realize the segmentation, counting and analysis of the particles to be measured.

[0009] The technical solution to achieve the purpose of the present invention is: an underwater sediment particle size measurement system based on laser light curtain imaging, the system comprising a host computer, a waterproof metal isolation chamber, and a laser light curtain irradiation module, a microscopic optical imaging module, and an image sensing module installed in the waterproof metal isolation chamber; the waterproof metal isolation chamber is provided with an optical window; The laser light curtain irradiation module is used to generate a stable and uniform laser light curtain and output it from the optical window to expose and display the sediment particles to be measured; The microscopic optical imaging module is used to achieve high-resolution dark-field imaging of sediment particles within the laser light curtain plane; The image sensor module is used to collect images of sediment particles to be measured in real time and upload them to the host computer; The host computer is used to process the image of the sediment particles to be measured to measure the sediment particle size.

[0010] Furthermore, the laser light curtain irradiation module includes two symmetrically and coaxially arranged laser light curtain irradiation units, which together generate a fan-shaped laser light curtain within the laser irradiation range. The mud and sand particles in the plane where the laser light curtain is located block the propagation of light, and the edge feature information is displayed with high exposure.

[0011] Furthermore, each laser light curtain irradiation unit includes a coaxially arranged visible light semiconductor laser and a prism; The visible light semiconductor laser generates high-power 532 nm wavelength laser light through stimulated emission, and then passes through a prism to achieve controllable steering of the incident laser light, thereby generating a fan-shaped laser light curtain.

[0012] Furthermore, the microscopic optical imaging module is a symmetrical Gaussian relay imaging system, comprising a relay front group structure and a relay rear group structure coaxially arranged in sequence, which jointly image the mud and sand particles in the laser light curtain illumination area onto the image sensing module; wherein, the mud and sand particles to be measured are located on the front focal plane of the relay front group structure, and the target surface of the image sensing module is located on the rear focal plane of the relay rear group structure.

[0013] Furthermore, the image sensing module includes a back-illuminated CMOS image sensor.

[0014] Furthermore, the system also includes two line interfaces arranged on the waterproof metal isolation chamber: a power interface and a waterproof transmission cable RJ-45 interface. The power supply of the laser light curtain irradiation module and the image sensor module, and the communication between the image sensor module and the host computer are all completed through the waterproof transmission cable.

[0015] Furthermore, the host computer processes the image of the sediment particles to be measured to measure the sediment particle size, specifically by using an improved random forest algorithm to process the image of the sediment particles to be measured to measure the sediment particle size, specifically including: Step 1: Collect multiple images of sediment particles with known characteristic information and add labels to form an original sample dataset; the labels include large sediment particles, small sediment particles, and background; large particles and small particles are divided according to a preset particle area threshold. Particles with an area larger than the preset particle area threshold are considered large particles, otherwise they are considered small particles; Step 2: Divide the original sample data set into a training set and a test set; Step 3: Create and optimize a random forest classifier, including: Step 3-1, build an initial random forest classifier based on all the features of the sample; Step 3-2, randomly extract some samples from the training set and calculate the importance of each feature of the sample; Step 3-3, sort the features in descending order according to their importance, and select the top m features, and eliminate other features to obtain the optimized random forest classifier; Step 4: Using the training set to train the random forest classifier to obtain a classification model; Step 5: Preprocess the sediment particle image to be tested, enhance the image features, and obtain the region of interest; Step 6: inputting the sediment particle image to be tested after the processing in step 5 into the classification model to obtain sediment particle classification results, including large sediment particles, small sediment particles, and background; Step 7: Calculate the diameters of large and small sediment particles.

[0016] Furthermore, the importance of each feature of the sample is calculated in step 3-2, specifically including: (1) For each sample, calculate the importance of each feature: in, For the n The importance of features, Respectively n The feature is passed through the random forest classifier node m Before and after the split Index of T Represents a collection of nodes; The calculation formulas are: Where, Representation node m The Gini index is used to represent Calculation, For nodes m Middle k The sample ratio of each target category reflects the distribution of the target category; K is the total number of target categories, including large sediment particles, small sediment particles and background; (2) For each feature, the importance of the feature corresponding to all samples is averaged to obtain the importance of the feature.

[0017] Furthermore, during the training process of step 4, the random forest classifier performs dynamic weight adjustment. The specific formula is: Where, To initialize the weights, is the predicted probability output by the random forest classifier, Is a positive number, indicating an adjustment factor; if Around 0.5, that is The difference from 0.5 is within the first preset range, then The value is greater than 1; if Around 0 or 1, that is The difference from 0 or 1 is within the second preset range or the third preset range, The value is less than 1.

[0018] Furthermore, step 5 pre-processes the sediment particle image to be measured, enhances image features, and obtains the region of interest, specifically including: Step 5-1, convert the format of the sediment particle image to be measured into PNG format; Step 5-2, performing grayscale conversion processing on the image of the sediment particles to be measured after format conversion; Step 5-3, performing bilateral filtering on the image after grayscale conversion; In step 5-4, image enhancement processing is performed, and edge detection is performed to obtain the region of interest.

[0019] Compared with the prior art, the present invention has the following significant advantages: (1) A modular design was adopted to meet measurement requirements, completing the implementation of a laser light curtain and the acquisition of image information. A 532 nm wavelength green semiconductor laser was used to generate a laser light curtain that was confocal with the imaging plane, and the presentation of particle feature information was far superior to conventional imaging. Two imaging optical systems were used to perform microscopic imaging of sediment particles. A 20-megapixel CMOS sensor captured images based on the coplanarity of the object plane and transmitted them to a computer via a waterproof cable for image processing.

[0020] (2) The laser light curtain is realized by green lasers and prisms at both ends. The laser light curtain is generated by the laser. The sand particles on the plane that coincides with the light curtain block the propagation of the laser, and the edge details are presented with high brightness, which can effectively improve the accuracy of subsequent measurements.

[0021] (3) The random forest machine learning algorithm was optimized and improved according to the measurement environment and measurement target. A data set was obtained by interactive learning through manual selection of ROI (region of interest). The Gini importance parameter and dynamic weight adjustment were introduced to improve the random forest classification algorithm. The processing time of the single image algorithm was reduced from 13086 ms to 9441 ms, the small particle recall rate was increased from 69.4% to 87.3%, and the large particle detection error was reduced from 10.2% to 6.7%. Efficient classification and segmentation of sediment particles in the image was achieved, and the final particle size information was obtained.

[0022] (4) Through image preprocessing and image enhancement, the algorithm workload and overall redundancy are effectively reduced, and the edge information of the sediment particles to be measured is enhanced: Based on the amount of storage space occupied and the algorithm load, the PNG format is selected as the image storage and transmission format in the system. After grayscale conversion, the storage space occupied is reduced to 1 / 3 of the original. Bilateral filtering is used to smooth noise while preserving edge features. Histogram equalization is used to enhance the contrast of sediment particles, and the edge information of sediment particles is enhanced based on the Hessian matrix. In combination with the image information characteristics of sediment particle edges, the Sobel operator is selected for edge detection to avoid large-scale diffusion of particle edges.

[0023] (5) In terms of engineering applications, it can obtain relatively accurate sediment distribution in the fields of water ecology and water environment, judge the environmental quality of water areas, and provide water conservancy and waterway departments with on-site solutions that do not require laboratory analysis. It is particularly suitable for reservoir sedimentation monitoring or related research on estuary delta evolution. In terms of scientific research, this measurement system can effectively promote further research in related water conservancy fields. In scenarios such as debris flows and reservoir dam failures, it can quickly obtain sediment particle size data to improve the accuracy of disaster prediction models. In terms of economic value, this measurement system reduces the cost of manual sampling and laboratory analysis. Based on the real-time performance of the system, it can optimize the operation and maintenance strategies of water conservancy projects and reduce the difficulty and time consumption of related operations.

[0024] The present invention is further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 The present invention is a framework diagram of an underwater sediment particle size measurement system based on laser light curtain imaging in one embodiment.

[0026] Figure 2 FIG. 1 is an optical principle diagram of an imaging optical system in an embodiment.

[0027] Figure 3 FIG. 1 is an optical structure diagram of an imaging optical system in an embodiment.

[0028] Figure 4 It is a schematic diagram of the laser light curtain illumination module in one embodiment.

[0029] Figure 5 FIG. 1 is a light path diagram of a 1:1 microscopic imaging system in an embodiment.

[0030] Figure 6 FIG. 1 is a light path diagram of a 1:2 microscopic imaging system in an embodiment.

[0031] Figure 7 FIG. 1 is a diagram of a back-illuminated CMOS structure in an embodiment.

[0032] Figure 8 The present invention is a flow chart of an algorithm for measuring sediment particle size based on images of sediment particles to be measured in one embodiment. DETAILED DESCRIPTION

[0033] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0034] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), such directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0035] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features specified as "first" and "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that ordinary technicians in this field can implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0036] In one embodiment, combined Figure 1 , provides an underwater sediment particle size measurement system based on laser light curtain imaging, the system includes a host computer, a waterproof metal isolation chamber, and a laser light curtain irradiation module, a microscopic optical imaging module and an image sensing module installed in the waterproof metal isolation chamber; the waterproof metal isolation chamber is provided with an optical window; The laser light curtain irradiation module is used to generate a stable and uniform laser light curtain and output it from the optical window to expose and display the sediment particles to be measured; The microscopic optical imaging module is used to achieve high-resolution dark-field imaging of sediment particles within the laser light curtain plane; The image sensor module is used to collect images of sediment particles to be measured in real time and upload them to the host computer; The host computer is used to process the image of the sediment particles to be measured to measure the sediment particle size.

[0037] Here, the microscopic optical imaging module collects and focuses the optical information of the sediment particles onto the imaging plane. By arranging the optical elements, the light is converged at the appropriate angle and path to the imaging target surface of the image sensor module. The high-resolution image sensor module realizes the photoelectric conversion and transmits the image data to the host computer through a waterproof transmission cable for image algorithm processing and analysis. The optical principle diagram is shown in the figure below. Figure 2 shown.

[0038] Furthermore, in one embodiment, the laser light curtain irradiation module includes two symmetrically and coaxially arranged laser light curtain irradiation units, which together generate a fan-shaped laser light curtain within the laser irradiation range. The mud and sand particles in the plane where the laser light curtain is located block the propagation of light, and the edge feature information is displayed with high exposure.

[0039] Here, a laser light curtain is manufactured during actual measurements in an underwater environment, maintaining a stable waveform for a long time. Excessive divergence of the light beam is effectively avoided during the manufacture of the light curtain, ensuring that the shape of the light curtain is stable, and almost no scattering or expansion occurs during the propagation process, maintaining a high beam quality.

[0040] Preferably, in some embodiments, each laser light curtain irradiation unit includes a coaxially arranged visible light semiconductor laser and a prism; The visible light semiconductor laser generates high-power 532 nm wavelength laser light through stimulated emission, and then passes through a prism to achieve controllable steering of the incident laser light, thereby generating a fan-shaped laser light curtain. Figure 4 shown.

[0041] Here, gas lasers have poor beam quality, solid-state lasers require additional cooling systems for temperature control, and dye lasers are expensive. Semiconductor lasers are small and lightweight, and their built-in design in underwater detection equipment will not affect the overall performance of the equipment. Their long lifespan means that the measurement device does not need to be frequently disassembled, which prevents the life of the waterproof rubber rings at the device connections from being reduced. Therefore, semiconductor lasers are selected. In order to display sediment particles with high brightness, infrared and ultraviolet light are not perceptible to the human eye, so visible light semiconductor lasers are used. Within the visible light range, water strongly absorbs short-wavelength light and weakly absorbs medium-wavelength light such as green light. Therefore, green light travels a relatively long distance in water and can effectively penetrate the water layer. Green light also has a smaller scattering effect than other wavelengths of light, and its transmission and focusing capabilities in water are also stronger. Therefore, a green light semiconductor laser with a wavelength of 532 nm is selected.

[0042] Furthermore, in one embodiment, in combination Figure 3The microscopic optical imaging module is a symmetrical Gaussian relay imaging system, which includes a relay front group structure and a relay rear group structure coaxially arranged in sequence, which jointly image the mud and sand particles in the area illuminated by the laser light curtain onto the image sensing module; wherein, the mud and sand particles to be measured are located on the front focal plane of the relay front group structure, and the image sensing module is located on the rear focal plane of the relay rear group structure.

[0043] Here, the microscopic optical imaging system module is designed based on the symmetrical double-Gaussian structure imaging principle and the relay lens group as the structural basis to achieve high-definition dark-field imaging of sediment particles in the laser light curtain plane.

[0044] Preferably, in some embodiments, in order to facilitate the study of measurement accuracy and measurement efficiency, two sets of microscopic optical imaging systems are designed: one is a 1:1 microscopic imaging system, and the other is a 1:2 microscopic imaging system. The 1:1 microscopic imaging system is designed to be completely symmetrical, with the focal lengths of the front relay group and the rear relay group both being 50 mm. The sediment sample is located at the front focal plane of the front relay group, with an object field size of Φ20 mm. The sediment sample is imaged by the microscopic imaging system onto the target surface of the image sensor, i.e., the rear focal plane of the rear relay group. The optical path diagrams of the two microscopic imaging systems are as follows: Figure 5 and Figure 6 shown.

[0045] Furthermore, in one embodiment, the image sensing module includes a back-illuminated CMOS image sensor.

[0046] Here, combined Figure 7 The back-illuminated CMOS image sensor places the circuit layer below the photoelectric sensing layer to avoid blocking incident light. It is combined with a lens array to optimize laser efficiency, and filters are used to achieve spectral selection and noise suppression. During system construction, the sensor is located behind the lens and connected to the image processing device via a waterproof cable to transmit the collected image information. MVS software is used for driver control, and parameters such as gain and exposure time can be manually adjusted to obtain high-quality images of sediment particles.

[0047] Here, the image sensing module uses a high-resolution CMOS sensor to collect pixel array data of sediment particle image signals based on the photoelectric effect and photoelectric conversion principle, completes analog-to-digital conversion, and transmits it to the host computer for sediment particle size measurement and analysis. Furthermore, in one embodiment, the system also includes two line interfaces arranged on the waterproof metal isolation chamber: a power interface and a waterproof transmission cable RJ-45 interface. The power supply of the laser light curtain irradiation module and the image sensor module, and the communication between the image sensor module and the host computer are all completed through the waterproof transmission cable.

[0048] Here, in the overall design of the measurement system hardware, the particularity of underwater environment measurement is taken into consideration. The device is sealed with a waterproof metal isolation chamber. The power supply of the laser and CMOS is completed through a waterproof cable. Two line interfaces are designed at the end of the device, namely the power interface and the waterproof transmission cable RJ-45 interface. After connection, the entire system meets the waterproof requirements. In addition, the laser, lens group and CMOS device in the device are fixed by a metal structure to prevent the image from being blurred due to changes in the laser irradiation direction or imaging depth of field due to movement. The lens group and CMOS are combined and connected through a C interface. The system device is designed as follows: Figure 3 shown.

[0049] Furthermore, in one embodiment, the host computer processes the image of the sediment particles to be measured to measure the sediment particle size, and combines Figure 8 Specifically, the improved random forest algorithm is used to process the sediment particle images to measure the sediment particle size, including: Step 1: Collect multiple images of sediment particles with known characteristic information and add labels to form an original sample dataset; the labels include large sediment particles, small sediment particles, and background; large particles and small particles are divided according to a preset particle area threshold. Particles with an area larger than the preset particle area threshold are considered large particles, otherwise they are considered small particles; Step 2: Divide the original sample data set into a training set and a test set; Step 3: Create and optimize a random forest classifier, including: Step 3-1, build an initial random forest classifier based on all the features of the sample; Step 3-2, randomly extract some samples from the training set and calculate the importance of each feature of the sample; Step 3-3, sort the features in descending order according to their importance, and select the top m features, and eliminate other features to obtain the optimized random forest classifier; Step 4: Using the training set to train the random forest classifier to obtain a classification model; Step 5: Preprocess the sediment particle image to be tested, enhance the image features, and obtain the region of interest; Step 6: inputting the sediment particle image to be tested after the processing in step 5 into the classification model to obtain sediment particle classification results, including large sediment particles, small sediment particles, and background; Step 7: Calculate the diameters of large and small sediment particles.

[0050] Preferably, in some embodiments, calculating the importance of each feature of the sample in step 3-2 specifically includes: (1) For each sample, calculate the importance of each feature: in, For the n The importance of features, Respectively n The feature is passed through the random forest classifier node m Before and after the split Index of T Represents a collection of nodes; The calculation formulas are: Where, Representation node m The Gini index is used to represent Calculation, For nodes m Middle k The sample ratio of each target category reflects the distribution of the target category; K is the total number of target categories, including large sediment particles, small sediment particles and background; (2) For each feature, the importance of the feature corresponding to all samples is averaged to obtain the importance of the feature.

[0051] For example, features include but are not limited to particle area, laser scattering intensity, texture entropy, roundness, edge gradient mean, high-frequency texture, background water variance, and ambient light interference intensity. The feature importance calculated for some samples is shown in Table 1 below.

[0052] Preferably, in some embodiments, during the training process of step 4, the random forest classifier performs dynamic weight adjustment, and the specific formula is: Where, To initialize the weights, is the predicted probability output by the random forest classifier, Is a positive number, indicating an adjustment factor; if Around 0.5, that is The difference from 0.5 is within the first preset range, then The value is greater than 1; if Around 0 or 1, that is The difference from 0 or 1 is within the second preset range or the third preset range, The value is less than 1. In summary, we propose an improved random forest classification algorithm. Using the Gini importance parameter as a guideline, we screen key splitting features and eliminate redundant features, reducing the average algorithm processing time for a single image. To address issues such as missed detection and misclassification of small particles, we introduce a dynamic weight adjustment method based on the Focal Loss concept. This increases the weight of small particles and difficult-to-identify samples, improving both the recall rate for small particles and the detection error for large particles.

[0053] Preferably, in some embodiments, the step 5 of preprocessing the image of the sediment particles to be measured, enhancing the image features, and obtaining the region of interest specifically includes: Step 5-1, convert the format of the sediment particle image to be measured into PNG format; Here, the storage space can be reduced and the processing speed can be increased; Step 5-2, performing grayscale conversion processing on the image of the sediment particles to be measured after format conversion; Here, grayscale conversion is used to reduce the workload of each image in batch processing; Step 5-3, performing bilateral filtering on the image after grayscale conversion; Here, bilateral filtering effectively smooths the noise and preserves the edges completely; Step 5-4: perform image enhancement and edge detection to obtain the region of interest; Here, preferably, image enhancement processing is performed through histogram equalization and Hessian matrix to enhance the contrast and edge feature information of sediment particles.

[0054] Here, preferably, the Sobel operator is selected for edge detection to eliminate the influence of edge diffusion and provide a feature information set for subsequent screening and segmentation algorithms.

[0055] In one embodiment, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the functions of the underwater sediment particle size measurement system based on laser light curtain imaging are realized.

[0056] For the specific limitations of each module, please refer to the above limitations of the underwater sediment particle size measurement system based on laser light curtain imaging, which will not be repeated here.

[0057] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the functions of the underwater sediment particle size measurement system based on laser light curtain imaging are realized.

[0058] For the specific limitations of each module, please refer to the above limitations of the underwater sediment particle size measurement system based on laser light curtain imaging, which will not be repeated here.

[0059] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only illustrative of the principles of the present invention. Without departing from the spirit and scope of the present invention, any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention shall be included in the scope of protection of the present invention.

Claims

1. An underwater sediment particle size measurement system based on laser light curtain imaging, characterized in that: The system includes a host computer, a waterproof metal isolation chamber, and a laser light curtain irradiation module, a microscopic optical imaging module and an image sensing module installed in the waterproof metal isolation chamber; the waterproof metal isolation chamber is provided with an optical window; The laser light curtain irradiation module is used to generate a stable and uniform laser light curtain and output it from the optical window to expose and display the sediment particles to be measured; The microscopic optical imaging module is used to achieve high-resolution dark-field imaging of sediment particles within the laser light curtain plane; The image sensor module is used to collect images of sediment particles to be measured in real time and upload them to the host computer; The host computer is used to process the image of the sediment particles to be measured to measure the sediment particle size.

2. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 1 is characterized in that: The laser light curtain irradiation module includes two symmetrical and coaxially arranged laser light curtain irradiation units, which together generate a fan-shaped laser light curtain within the laser irradiation range. The mud and sand particles in the plane where the laser light curtain is located block the propagation of light, and the edge feature information is displayed with high exposure.

3. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 2 is characterized in that: Each laser light curtain irradiation unit includes a coaxially arranged visible light semiconductor laser and a prism; The visible light semiconductor laser generates high-power 532 nm wavelength laser light through stimulated emission, and then passes through a prism to achieve controllable steering of the incident laser light, thereby generating a fan-shaped laser light curtain.

4. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 1 is characterized in that: The microscopic optical imaging module is a symmetrical Gaussian relay imaging system, which includes a relay front group structure and a relay rear group structure arranged coaxially in sequence, which jointly image the mud and sand particles in the area illuminated by the laser light curtain onto the image sensing module; wherein, the mud and sand particles to be measured are located on the front focal plane of the relay front group structure, and the target surface of the image sensing module is located on the rear focal plane of the relay rear group structure.

5. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 1 is characterized in that: The image sensing module includes a back-illuminated CMOS image sensor.

6. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 1 is characterized in that: The system also includes two line interfaces arranged on the waterproof metal isolation chamber: a power interface and a waterproof transmission cable RJ-45 interface. The power supply of the laser light curtain irradiation module and the image sensor module, and the communication between the image sensor module and the host computer are all completed through the waterproof transmission cable.

7. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 1 is characterized in that: The host computer processes the image of the sediment particles to be measured to measure the sediment particle size, specifically: using an improved random forest algorithm to process the image of the sediment particles to be measured to measure the sediment particle size, specifically including: Step 1: Collect multiple images of sediment particles with known characteristic information and add labels to form an original sample dataset; the labels include large sediment particles, small sediment particles, and background; large particles and small particles are divided according to a preset particle area threshold. Particles with an area larger than the preset particle area threshold are considered large particles, otherwise they are considered small particles; Step 2: Divide the original sample data set into a training set and a test set; Step 3: Create and optimize a random forest classifier, including: Step 3-1, build an initial random forest classifier based on all the features of the sample; Step 3-2, randomly extract some samples from the training set and calculate the importance of each feature of the sample; Step 3-3, sort the features in descending order according to their importance, and select the top m features, and eliminate other features to obtain the optimized random forest classifier; Step 4: Using the training set to train the random forest classifier to obtain a classification model; Step 5: Preprocess the sediment particle image to be tested, enhance the image features, and obtain the region of interest; Step 6: inputting the sediment particle image to be tested after the processing in step 5 into the classification model to obtain sediment particle classification results, including large sediment particles, small sediment particles, and background; Step 7: Calculate the diameters of large and small sediment particles.

8. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 7 is characterized in that: In step 3-2, the importance of each feature of the sample is calculated, including: (1) For each sample, calculate the importance of each feature: in, For the n The importance of features, Respectively n The feature is passed through the random forest classifier node m Before and after the split index; T Represents a collection of nodes; The calculation formulas are: Where, Representation node m The Gini index is used to represent Calculation, For nodes m Middle k The sample ratio of each target category reflects the distribution of the target category; K is the total number of target categories, including large sediment particles, small sediment particles and background; (2) For each feature, the importance of the feature corresponding to all samples is averaged to obtain the importance of the feature.

9. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 7 is characterized in that: During the training process of step 4, the random forest classifier performs dynamic weight adjustment. The specific formula is: Where, To initialize the weights, is the predicted probability output by the random forest classifier, Is a positive number, indicating an adjustment factor; if Around 0.5, i.e. | The difference between | and 0.5 is within the first preset range, then The value is greater than 1; if Around 0 or 1, i.e. | | and 0 or 1 is within the second preset range or the third preset range, then The value is less than 1.

10. The underwater sediment particle size measurement system based on laser light curtain imaging according to claim 7, characterized in that: Step 5 pre-processes the sediment particle image to be measured, enhances image features, and obtains the region of interest, specifically including: Step 5-1, convert the format of the sediment particle image to be measured into PNG format; Step 5-2, performing grayscale conversion processing on the image of the sediment particles to be measured after format conversion; Step 5-3, performing bilateral filtering on the image after grayscale conversion; In step 5-4, image enhancement processing is performed, and edge detection is performed to obtain the region of interest.

Images