Portable long-distance water level observation device and method based on multi-sensor fusion

Abstract

The portable long-distance water level observation device and method based on multi-sensor fusion belong to the technical field of hydrogeological observation and hydraulic measurement, and aims to solve the problems of traditional wall water gauge, such as easy pollution and fading, large error in long-distance observation, water surface fluctuation affecting reading stability, and the problem that the existing measuring equipment is difficult to balance portability and precision. The device is composed of a fixed reference scale and a portable handheld main machine. The handheld main machine integrates image acquisition, laser ranging, attitude sensing and other modules, and combines image recognition, tilt compensation, time series filtering and triangular height conversion to complete measurement. The present application can realize long-distance non-contact automatic water level observation, the scale is resistant to erosion and has long service life, effectively eliminates the error caused by handheld posture and water surface fluctuation, and has excellent measurement precision and stability. The device is portable and flexible, suitable for multi-point outdoor patrol measurement, supports wireless data upload, safe and convenient operation, and has strong environmental adaptability, and is suitable for various outdoor water level observation scenes such as karst channels and water measuring weirs.

Description

Technical Field

[0001] This invention relates to the field of hydrogeological observation and hydraulic engineering measurement technology, and in particular to a portable long-distance water level observation device and method based on multi-sensor fusion. Background Technology

[0002] In karst hydrogeological surveys, groundwater flow monitoring, and seepage observation in hydraulic engineering projects, weirs are a widely used infrastructure. By constructing weirs with specific cross-sectional shapes in karst channels or irrigation canals, and in conjunction with water level observations, real-time flow rates can be calculated using hydraulic formulas. Therefore, the accuracy and convenience of water level observations directly determine the accuracy and reliability of flow rate data.

[0003] Currently, the most common method for water level observation in karst channel weirs in the field is the wall gauge method. This involves vertically installing a graduated enamel, stainless steel, or plastic water gauge on the sidewall or side wall of the weir. Observers approach the gauge and visually read the value at the intersection of the water level and the graduations to obtain the water level elevation. This method has advantages such as low cost, simple installation, and intuitive results, and has therefore been widely used for a long time.

[0004] However, the wall water gauge method has revealed the following prominent shortcomings in long-term field applications: 1. Grading Fading and Damage: Karst passages are characterized by extremely high humidity year-round, with some areas even containing corrosive moisture containing sulfur or carbonates. The printed graduations or paint coating on the wall-mounted water level gauges are highly susceptible to fading, peeling, mold growth, or being covered by travertine deposits after prolonged exposure. Once the graduations become unclear, observers struggle to accurately interpret water level values, often requiring frequent replacement of the gauges or manual redrawing. Replacing the gauges not only increases equipment costs but, more importantly, necessitates recalibrating the zero-value elevation datum after each replacement. This cumbersome calibration process is prone to introducing systematic errors, further increasing maintenance costs and observational errors.

[0005] 2. Due to terrain limitations, some measuring weirs have high and steep sidewalls, or the water flow in front of the weir is rapid and the banks are slippery, making it impossible for observers to safely approach the water gauge for close-range readings. Furthermore, some karst caves have narrow spaces, or the measuring weirs are covered by enclosed panels, leaving only small observation windows. In these situations, observers are forced to make long-distance visual estimations from a safe location away from the water gauge. Because the human eye's resolving power decreases significantly with distance, coupled with insufficient lighting inside the cave, fading water gauge markings, and water surface reflections, the error in long-distance estimations can often reach several centimeters, severely affecting the accuracy of flow rate calculations.

[0006] 3. Water surface fluctuations affect reading stability: Water flow in karst channels often exhibits pulsating characteristics, and the water surface is not a static plane but contains tiny waves. When observers estimate the reading from a distance, it is difficult to accurately determine the average position of the wave surface, and they can only make a rough judgment based on the instantaneous wave crests or troughs, introducing a large random error.

[0007] To address the above problems, several improvements have emerged in the existing technology, but each has its own limitations: For example, CN216283811U discloses a portable laser head measuring device for measuring water head. By setting a laser measuring instrument on a bracket and using a connecting rod and a hollow tube to measure the head, it avoids the problem of laser reflection on the water surface affecting the measurement value to a certain extent and simplifies the steps of traditional manual measurement. However, its bracket needs to be installed on the wall through the connecting rod, and the hollow tube needs to be inserted vertically into the water. In essence, it is still a semi-fixed installation. In the field karst channel scenario where multiple points need to be inspected, the rapid setup and transfer of the device is not convenient, and its portability and flexibility are still insufficient.

[0008] For example, CN220039585U discloses a low-power telemetry weir gauge based on laser ranging technology. By setting a vertical pipe below the laser monitoring host and placing a hemispherical foam float inside the pipe that rises and falls with the water level, it can effectively avoid interference from external light on the laser beam and ensure measurement accuracy. However, the pipe needs to be fixed by clamps, and the whole structure is a fixed telemetry structure. The probe and data acquisition terminal must be deployed at each measurement point, resulting in high equipment cost. In addition, it is necessary to solve the problems of power supply and communication in the field, which makes it difficult to meet the low-cost requirements of multi-point inspection.

[0009] For example, CN220120203U discloses a portable non-contact water level measuring device, which uses an L-shaped main rod nested with an axially movable long rod. A laser sensor is set at the outer end of the long rod, and a level bubble is set on the main rod for observing the levelness. Although it solves the problem that traditional water level probes need to contact the water surface and have low accuracy, and realizes portable non-contact measurement, its measurement accuracy is highly dependent on the operator's manual interpretation of the level bubble and the stable control of the handheld rod. When observing from a distance, even a slight tilt or sway of the rod will introduce measurement errors that are difficult to compensate for effectively, which has a significant impact on the ease of operation and measurement consistency in complex terrain.

[0010] In summary, existing water level observation methods have consistently failed to achieve an ideal balance between portability, long-distance measurement accuracy, and adaptability to harsh field environments. The industry urgently needs a water level observation solution with superior overall performance. Summary of the Invention

[0011] The technical problem to be solved by this invention is to provide a portable long-distance water level observation device and method based on multi-sensor fusion, which solves the problems of faded and soiled scales on existing wall water level gauges, difficulty in long-distance observation, unstable readings due to water surface fluctuations, and the difficulty in balancing portability and accuracy in existing improved solutions. It realizes the coordinated operation of a fixed reflective ruler and a portable handheld host, and automatically obtains accurate water level elevation at any observation location through multi-sensor fusion calculation of image acquisition, laser ranging, and attitude sensing, meeting the needs of long-distance, high-precision, and portable water level observation in complex field scenarios such as karst channel weirs.

[0012] To achieve the above technical objectives, the present invention adopts the following technical solution: The portable long-distance water level observation device and method based on multi-sensor fusion provided by this invention adopts a split structure design of a fixed reference scale and a portable handheld host, and integrates multiple types of sensors such as image acquisition, laser ranging, and attitude sensing. Combined with image recognition, time-series filtering, tilt compensation, and trigonometric leveling algorithms, it realizes non-contact, long-distance, automated, and high-precision water level observation. It takes into account the durability, portability, measurement stability, and adaptability to complex field environments, and meets the application requirements of multi-point inspection in scenarios such as karst channels, water measuring weirs, and flow measurement weirs.

[0013] The portable long-distance water level observation device based on multi-sensor fusion mainly consists of two parts: a fixedly deployed reference scale and a portable handheld host. The portable handheld host and the reference scale are used separately and in conjunction. The portable handheld host completes image acquisition, distance measurement and attitude detection of the reference scale from a safe position far away from the reference scale, and obtains the water level elevation by relying on multi-sensor fusion calculation.

[0014] The reference scale is fixedly installed on the sidewall of the weir and is made of corrosion-resistant metal substrate, which can withstand the corrosion of high humidity and corrosive water vapor in the karst environment for a long time. The reference scale is formed by laser etching, and the surface is integrally formed with basic division lines, auxiliary division lines, scale values, barcode, and spatial positioning codes evenly distributed along the longitudinal direction. The surface is coated with a high-reflective coating to ensure clear imaging even in low light environments and long-distance conditions. The barcode is located in the top area of ​​the reference scale and stores unique coding information such as the scale number, installation position, and calibration date. The spatial positioning code is a continuously arranged black and white coded pattern, with each group of patterns corresponding to independent absolute position information. The geometric dimensions of adjacent patterns are fixed known values, providing a precise spatial position reference for the equipment. The reference scale is installed vertically, and its zero scale position is precisely aligned with the elevation benchmark at the bottom of the weir.

[0015] The portable handheld host is an integrated device that can be held by a single person. The entire device is encased in a waterproof and moisture-proof shell, making it suitable for use in complex outdoor environments. The mainboard inside the host integrates an image acquisition module, a laser rangefinder module, an attitude sensing module, a computing and control unit, and a wireless communication module. It also features a built-in rechargeable battery to power the entire device. Externally, the host has an image acquisition lens at the front, a display screen on the top, and a power switch, measurement button, upload button, and reset button arranged sequentially on the sides. An integrated grip is located at the bottom, and a charging port is located on the bottom of the device for charging the built-in rechargeable battery. Specifically, the rechargeable battery is electrically connected to the image acquisition module, laser rangefinder module, attitude sensing module, computing and control unit, display screen, wireless communication module, power switch, measurement button, upload button, and reset button. The charging port is electrically connected to the rechargeable battery to charge it.

[0016] The image acquisition module is equipped with a high-definition industrial camera module, and the matching image acquisition lens has optical zoom capability. The lens field of view has been specially calibrated, and it can completely capture the boundary area between the reference scale and the water surface within the maximum observation distance of the equipment, ensuring the integrity of image acquisition.

[0017] The laser ranging module employs a phase-type laser ranging sensor, which features a long measurement range and high measurement accuracy. In long-distance operation, the laser spot can stably act on the scale area, ensuring precise ranging point positioning. The emission optical axis of the laser ranging module and the optical axis of the image acquisition lens are precisely calibrated by a mechanical structure, and their lateral offset is pre-stored as a fixed system parameter within the calculation and control unit.

[0018] The attitude sensing module adopts a six-axis inertial measurement unit, which is composed of a three-axis accelerometer and a three-axis gyroscope. It can detect the roll angle, pitch angle and yaw angle and other full-dimensional spatial attitude data of the portable handheld host in real time and accurately, providing the original basis for subsequent error compensation.

[0019] The computing and control unit, based on an embedded high-performance processor, serves as the data processing and logic control center of the entire device. Internally, it incorporates a water surface line extraction algorithm module and a timing filtering module. During operation, the computing and control unit first analyzes the images transmitted from the image acquisition module, identifying and reading the identity and parameter information within the barcode. Simultaneously, it extracts spatial positioning coding features and, combining pixel coordinates with known physical dimensions, calculates the spatial position and orientation of the image acquisition device relative to the reference scale coordinate system. Utilizing edge detection operators combined with grayscale gradient analysis technology, it automatically identifies the boundary area between the water body and the reference scale, and, based on the mapping relationship between pixels and physical scale, calculates the corresponding water level scale value. For the slant distance data transmitted from the laser ranging module, the computing and control unit, in conjunction with the attitude angle acquired by the attitude sensing module, performs tilt compensation correction. Then, relying on a trigonometric leveling model, it calculates the current absolute water level elevation.

[0020] Furthermore, the temporal filtering module can batch process the scale data obtained from recognizing multiple consecutive frames of images, automatically removing abnormal values ​​and calculating the average value to offset reading fluctuations caused by water surface waves and water flow pulsations. Simultaneously, the calculation and control unit can overlay electronic aiming crosshairs on the real-time display screen based on the image recognition results, guiding the operator to accurately align the laser point with the target area of ​​the reference scale. When the reference scale is partially obstructed or severely damaged, causing the barcode or spatial positioning code to be unreadable, the device can automatically switch to a backup measurement mode: prioritizing the extraction of unobstructed basic and auxiliary scale lines to establish a temporary reference scale; if the scale texture is also unreadable, a feature point detection algorithm is invoked to extract texture features such as joints, rock layer boundaries, and moss boundaries of the natural rock on the weir wall, constructing a virtual reference coordinate system, and cooperating with a single auxiliary distance measurement to complete scale calibration, ultimately achieving water level elevation conversion and ensuring continuous measurement work.

[0021] The wireless communication module supports multiple communication standards and can transmit measurement data remotely to the hydrological monitoring platform in real time; the built-in rechargeable battery has the power to meet the needs of continuous field surveys for several days; each physical button has a clear function: the power switch controls the start and stop of the whole machine, the measurement button triggers a single data acquisition and calculation, the upload button supports manual data upload, and the reset button can restart the whole machine in case of system abnormality.

[0022] The portable long-range water level observation method based on multi-sensor fusion, applied to the aforementioned portable long-range water level observation device based on multi-sensor fusion, specifically includes the following steps: Step 1, Pre-measurement preparation: Vertically fix the reference ruler to the side wall of the measuring weir, use a leveling tool to ensure the ruler is installed vertically, and then use professional measuring equipment to accurately calibrate the zero mark of the reference ruler with the elevation benchmark at the bottom of the weir, record and bind the zero elevation data of the ruler, and complete the layout of the measuring point equipment.

[0023] Step 2, Powering on and Aiming Positioning: The operator stands in a safe area away from the water, holds the portable handheld host, and presses the power switch to start the equipment; the image acquisition module acquires the scene in real time and transmits it to the display screen for display. The equipment automatically recognizes the barcode and spatial positioning code in the image, retrieves the corresponding scale number, elevation benchmark and other preset parameters, and calculates the relative position according to the spatial code, and superimposes the electronic aiming crosshairs on the display screen.

[0024] Triggering multi-source data acquisition: The operator fine-tunes the placement of the portable handheld host, ensuring that the electronic aiming crosshairs on the screen are precisely aligned with the preset target area of ​​the reference scale, and presses the measurement button; the device simultaneously starts working on multiple modules, the laser ranging module collects the slant distance value from the host to the target point on the scale, the attitude sensing module collects various attitude angle data of the host in real time, and the image acquisition module continuously collects multiple frames of on-site images containing the scale and the water surface according to the set frame rate and duration.

[0025] Multi-sensor data fusion and calculation: The calculation and control unit retrieves attitude angle data and performs tilt compensation correction on the laser slant range value; combining the image pixel spacing of spatial positioning encoding and the actual physical spacing, the three-dimensional spatial coordinates of the portable handheld host in the scale coordinate system are calculated. The water surface line extraction algorithm is run, first performing grayscale conversion and filtering preprocessing on the image, then using edge detection operators and grayscale gradient analysis to locate the water surface boundary line position, and calculating the corresponding scale value frame by frame. Subsequently, temporal filtering processing is initiated, sorting all frame data within a preset time period, removing extreme values, and calculating the arithmetic mean to obtain a stable average water level scale; the system can adaptively adjust the number of sampling frames and the outlier removal ratio according to the intensity of water surface fluctuations. If scale encoding or scale recognition fails, the device automatically switches to a backup measurement mode, constructing a virtual benchmark based on on-site texture features to complete the calculation.

[0026] Results Display and Data Upload: Combining the pre-calibrated zero-point elevation of the scale, the calculated average water level scale, and spatial coordinate corrections, the current absolute water level elevation is calculated and displayed in real time on the screen. The wireless communication module automatically uploads the integrated data packet to the remote monitoring platform. The data packet contains information such as the unique scale code, measurement timestamp, water level elevation value, host attitude angle, and spatial coordinates. Operators can also manually upload data by pressing the upload button. In case of system lag or malfunction, pressing the reset button will restart the device.

[0027] Multi-point survey operation: After the measurement at a single measuring point is completed, the operator moves to the next measuring point with a portable handheld host and repeats all the steps of powering on, aiming, data acquisition, calculation, and result uploading to complete the multi-point water level observation work in the entire measuring area.

[0028] The portable long-distance water level observation device and method based on multi-sensor fusion provided by this invention have the following beneficial effects: 1. The reference scale of this invention adopts a corrosion-resistant metal substrate, laser-etched graduations, and a highly reflective coating, which is suitable for the high humidity and corrosive water vapor environment in karst areas. It completely solves the problems of fading, peeling, and contamination of traditional water gauge graduations, eliminates the need for frequent scale replacement and repeated elevation calibration, reduces maintenance labor and equipment costs, avoids system observation errors caused by repeated calibration, and addresses the technical defects of water gauges in the background technology, such as easy damage and cumbersome calibration.

[0029] 2. This invention adopts a separate matching structure of reference scale and portable handheld host. A single host can be adapted to multiple measuring point scales, resulting in a high equipment reuse rate. Measuring points only need to install the scale, without the need for separate power supply and communication equipment. This simplifies the field measuring point construction process, shortens the deployment cycle, and significantly reduces the hardware investment in multi-point survey scenarios, overcoming the shortcoming of high single-point deployment cost of existing fixed telemetry equipment.

[0030] 3. This invention supports long-distance non-contact water level measurement, allowing operators to work safely outside dangerous and restricted areas such as steep weirs, slippery banks, and narrow karst caves without having to get close to the water surface. This eliminates the safety hazards of falling into the water or slipping in the wild from the source and solves the problem that traditional water gauges require close-range readings and have limited observation conditions.

[0031] 4. This invention integrates three modules: image acquisition, phase-type laser ranging, and six-axis inertial attitude sensing. It pre-calibrates the parallelism between the laser optical axis and the lens optical axis, and with attitude tilt compensation and trigonometric leveling fusion calculation, it eliminates systematic errors caused by observation distance and hand-held shaking. The measurement accuracy is far superior to manual visual estimation, solving the defect of existing portable laser devices that cannot compensate for hand-held tilt errors.

[0032] 5. This invention incorporates a water surface line extraction algorithm and a time-series filtering module, relying on Gaussian filtering and Canny edge detection to identify water surface boundaries; by sampling multiple frames of images and removing extreme values ​​to take the mean, it offsets the interference of water flow pulsation and water surface waves, and can also adaptively adjust the number of sampling frames and the proportion of outlier removal according to the intensity of water surface fluctuations, taking into account both measurement accuracy and operation speed, and solving the technical problem of unstable readings caused by water flow fluctuations.

[0033] 6. This invention sets up a two-level backup measurement mechanism. When the scale is obstructed or damaged, causing the barcode and spatial positioning code to be unrecognizable, the residual scale texture is extracted first to establish a temporary benchmark. If the scale fails, the natural rock texture of the weir wall is extracted and combined with the feature point algorithm to construct a virtual coordinate system to complete the conversion. Monitoring is uninterrupted under extreme working conditions, improving the robustness of the equipment in field operation.

[0034] 7. This invention automatically binds the unique code and elevation benchmark information of the measuring point through image recognition barcode, eliminating the confusion and misrecording problems caused by manual recording of measuring points. It has strong data traceability and is suitable for large-scale multi-point inspection and measurement operations.

[0035] 8. This invention is equipped with a wireless communication module, which supports two data transmission modes: automatic upload and manual upload. The data packet contains multi-dimensional information such as scale code, timestamp, water level elevation, host attitude, and spatial coordinates, which facilitates unified storage in the background, hydrological trend analysis, and flow inversion calculation, and has a higher degree of intelligence.

[0036] 9. The main unit of this invention uses a six-axis inertial measurement unit to synchronously acquire three-dimensional attitude data of roll, pitch, and yaw, and is equipped with a dedicated tilt compensation algorithm to correct the slant distance of the ranging, which greatly reduces the requirements for the operator's hand stability, lowers the operation threshold, and improves measurement consistency.

[0037] 10. This invention is manufactured according to the standardization of the scale, with high precision in scale processing and excellent imaging effect of the highly reflective coating on the surface. It can still be stably identified in low light and long-distance cave environments, reducing the interference of insufficient light on image acquisition and recognition, and is suitable for observation scenarios in karst caves with no light or low light.

[0038] 11. The portable handheld host of this invention has a built-in high-capacity rechargeable battery and a universal charging interface. The battery life can meet the needs of continuous multi-day field surveys. The measurement point does not require external power supply. The equipment is flexible in carrying and resupplying, and is free from the limitations of field power supply facilities.

[0039] 12. The main unit of this invention can be operated by hand or fixedly mounted on a tripod for long-term fixed-point observation, taking into account both mobile surveying and fixed-point continuous monitoring modes, thus having stronger functional adaptability.

[0040] 13. The present invention has a compact hardware structure, standardized operation process, and operators can quickly get started without professional training. The single measurement process is short and the overall field operation efficiency is significantly improved.

[0041] 14. The laser ranging module of this invention uses a dedicated phase-type ranging sensor, which limits the dedicated measurement wavelength, range and accuracy range. Combined with pre-stored optical axis offset parameters, it suppresses optical path deviation from the hardware level and further stabilizes the measurement data throughout the process.

[0042] 15. The algorithm system of this invention fully covers the entire process of image preprocessing, spatial pose calculation, adaptive filtering, and virtual reference conversion. All technical logic can be run automatically through an embedded processor without manual intervention, resulting in a high degree of automation.

[0043] 16. This invention has a wide range of applicable scenarios and can be adapted to all scenarios that require long-distance, high-precision, non-contact water level measurement, such as karst channels, various water measuring weirs, flow measuring weirs and flumes, and seepage monitoring in water conservancy projects. Its industrialization and promotion application value is outstanding. Attached Figure Description

[0044] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a schematic diagram of the overall structure of the device of the present invention; Figure 2 This is a schematic diagram of the overall structure of the portable handheld host of the present invention; Figure 3 This is a partial cross-sectional view of the portable handheld host of the present invention; Figure 4 for Figure 1 A schematic diagram of the front structure of the reference scale; In the diagram: Reference scale 1, portable handheld host 2, weir 3, weir plate 4, basic division line 11, auxiliary division line 12, scale value 13, barcode 14, spatial positioning code 15, image acquisition module 21, laser ranging module 22, attitude sensing module 23, calculation and control unit 24, display screen 25, wireless communication module 26, rechargeable battery 27, charging port 28, motherboard 29, image acquisition lens 210, power switch 211, measurement button 212, upload button 213, reset button 214, grip 215. Detailed Implementation

[0045] The technical solution of the present invention will be further described below with reference to the accompanying drawings and embodiments: Example 1 like Figures 1 to 4 As shown, this embodiment discloses a portable long-distance water level observation device based on multi-sensor fusion. The device mainly consists of two parts: a fixedly deployed reference scale 1 and a portable handheld host 2. The two work together to achieve automatic acquisition of water level elevation over long distances without contact, relying on a multi-sensor fusion calculation method. A weir plate 4 is installed inside the weir channel 3, and the reference scale 1 is fixedly installed on the inner wall of the weir channel 3.

[0046] like Figure 4As shown, the scale 1 is made of corrosion-resistant metal substrates such as 304 stainless steel plate or indium steel strip, with a plate thickness of 2mm, which can withstand the high humidity environment and corrosive water vapor containing sulfur and carbonates in karst channels for a long time. The scale surface is processed by laser etching to form basic division lines 11, auxiliary division lines 12, scale values ​​13, barcode 14, and spatial positioning code 15. The basic division lines 11 have a line width of 2mm, and scale values ​​13 are set every 5cm. The average length of the meter interval is no more than 0.1mm different from the nominal value. The entire surface of the scale is covered with a glass microsphere reflective film, and the scale lines and numerical markings are coated with a high-reflective coating, which can still provide clear imaging in low light and long-distance environments. Barcode 14 is located on top of reference scale 1 and stores unique coded information such as scale number, installation location, and calibration date. Spatial positioning code 15 consists of alternating black and white graphics arranged longitudinally along reference scale 1. Each group of graphics corresponds to independent absolute position information, and the geometric dimensions of adjacent graphics are precisely known, providing a reliable spatial position reference for portable handheld host 2. Reference scale 1 is vertically fixed to the side wall of the measuring weir and installed vertically, with its zero mark strictly calibrated and aligned with the elevation reference at the bottom of the weir channel 3.

[0047] like Figure 2 and Figure 3 As shown, the portable handheld host 2 is encased in a waterproof and moisture-proof shell, with a compact overall structure, making it easy for a single person to carry and operate. It can be used by hand using the handle 215 or mounted and fixed on a tripod, making it suitable for long-term fixed-point observation operations. The mainboard 29 inside the host integrates an image acquisition module 21, a laser ranging module 22, an attitude sensing module 23, a calculation and control unit 24, a wireless communication module 26, and a rechargeable battery 27. An image acquisition lens 210 is installed at the front of the host, and a display screen 25 is set on the top. The power switch 211, measurement button 212, and upload button 213 are arranged sequentially on the side. The handle 215 is set at the bottom, and a USB-C (Universal Serial Bus Type-C) charging port 28 is provided at the bottom. The rechargeable battery 27 is charged through the charging port 28. Among them, the power switch 211 controls the opening and closing of the whole machine, the measurement button 212 is used to trigger a single water level measurement, the upload button 213 can manually upload measurement data, the reset button 214 is used to reset the system fault, and the handle 215 makes it convenient for the operator to hold the equipment with one hand.

[0048] Furthermore, the rechargeable battery 27 is electrically connected to the image acquisition module 21, the laser ranging module 22, the attitude sensing module 23, the calculation and control unit 24, the display screen 25, the wireless communication module 26, the power switch 211, the measurement button 212, the upload button 213, and the reset button 214, respectively. The charging port 28 is electrically connected to the rechargeable battery 27 to charge and power the rechargeable battery 27.

[0049] Furthermore, the image acquisition module 21 is equipped with a high-definition industrial camera module with a resolution of no less than 1920×1080 pixels, paired with an image acquisition lens 210 featuring 5-10x optical zoom. The lens's field of view has been optimized, enabling complete imaging of the boundary area between the reference scale 1 and the water surface within the maximum observation distance. The laser ranging module 22 employs a phase-type laser ranging sensor with a wavelength of 640nm-660nm, preferably 650nm. Its measurement accuracy reaches ±1.5mm, and its measurement range is 0.05m-200m. When operating at long distances, the laser spot diameter can be controlled within the width range of the scale lines. The optical axis of this module and the optical axis of the image acquisition lens 210 are precisely parallelized, with a parallelism deviation of less than 0.05°. The lateral offset of the optical axis is pre-stored as a fixed parameter in the calculation and control unit 24.

[0050] Furthermore, the attitude sensing module 23, with a three-axis accelerometer as its core, achieves a measurement resolution of 0.01°. It is preferably paired with a three-axis gyroscope to form a six-axis inertial measurement unit. Combined with a fusion filtering algorithm, it outputs real-time spatial attitude data such as roll angle, pitch angle, and yaw angle of the portable handheld host 2. The computing and control unit 24, with an embedded high-performance processor as its core, carries functions such as image recognition, data processing, water surface line extraction, and temporal filtering. During operation, this unit reads the barcode 14 to obtain scale information, extracts the spatial positioning code 15, and combines pixel coordinates and physical dimensions. Through three-dimensional pose calculation algorithms such as PnP (Perspective-n-Point) and homography matrix decomposition, it calculates the spatial position and attitude of the image acquisition module 21 relative to the reference scale 1 coordinate system, while simultaneously identifying the scale value 13 corresponding to the water surface boundary line.

[0051] Furthermore, the computing and control unit 24 can also overlay electronic aiming crosshairs on the display screen 25 based on the image recognition results. The center of the crosshairs corresponds to the theoretical landing point of the laser rangefinder spot, assisting the operator in accurately aiming at the target area. The water surface line extraction algorithm first converts the image into a grayscale image and smooths it with a Gaussian filter. After delineating the region of interest, it uses a Canny edge detection operator with a high threshold of 150 and a low threshold of 50, combined with grayscale gradient analysis, to extract the water surface boundary line. Then, based on the spatial positioning code 15, it establishes a mapping relationship between pixels and physical scales to calculate the water level scale. The temporal filtering module can process multi-frame image data. Under normal conditions, it acquires 30 frames of images at a rate of 10 frames / second within 3 seconds, removes the first 10% minimum values ​​and the last 10% maximum values, and then takes the average. The system can also adaptively adjust the number of sampling frames and the outlier removal ratio according to the intensity of water surface ripples, effectively eliminating reading fluctuations caused by water surface waves.

[0052] Furthermore, the calculation and control unit 24, combined with the angle data collected by the attitude sensing module 23, performs tilt compensation on the slant distance value obtained by laser ranging, and then calculates the absolute water level elevation through trigonometric leveling, which is then refreshed and displayed on the display screen 25 in real time. When the reference scale 1 is obstructed or severely damaged, causing the barcode 14 or spatial positioning code 15 to be unrecognizable, the device automatically switches to the backup measurement mode: first, the texture of the basic division line 11 and the auxiliary division line 12 is extracted to establish a temporary scale. If the scale cannot be recognized, the natural texture of the weir wall is extracted through SIFT (Scale-Invariant Feature Transform) and ORB (Oriented Fast and Rotated BRIEF) feature point detection algorithms to construct a virtual coordinate system. This is combined with auxiliary ranging to complete the scale calibration and water level conversion. In this mode, the measurement work can still be carried out normally.

[0053] Furthermore, the portable handheld host 2 has a built-in rechargeable lithium battery, which can meet the needs of multi-day field survey operations when fully charged. The wireless communication module 26 supports 4G / 5G (4th / 5th Generation Mobile Communication Technology), Bluetooth Low Energy and other communication standards, and can remotely upload data packets containing scale codes, timestamps, water level elevations, attitude angles and positioning coordinates to the data platform, which is convenient for backend traceability and data analysis.

[0054] This embodiment successfully integrates visual spatial positioning, precise laser ranging, and attitude error compensation through a collaborative design combining a fixed reflective ruler and a portable multi-sensor handheld host. This enables long-distance, non-contact, high-precision water level measurement in the field. The entire device is compact and requires no pre-installation of power supply and communication infrastructure at the measurement points. A single unit can serve multiple measurement points, significantly reducing the cost of multi-point surveys. It is particularly suitable for hydrogeological survey scenarios such as karst channels, weirs, and flow measurement flumes, demonstrating outstanding practical value and broad prospects for promotion.

[0055] Example 2 In another preferred embodiment, based on Embodiment 1 above, this embodiment provides a portable long-distance water level observation method based on multi-sensor fusion, implemented using the portable long-distance water level observation device based on multi-sensor fusion described in Embodiment 1. The specific operation steps are as follows: Step 1, Preparation before measurement: such as Figure 1As shown, the reference scale 1 is vertically fixed on the side wall of the measuring weir. The scale is kept vertical by using a level bubble or digital level. Then, the zero mark of the reference scale 1 is calibrated with the bottom elevation benchmark of the weir channel 3 using a total station to determine the zero elevation of the scale and complete the layout of the measuring point equipment.

[0056] Step 2, Start-up and Aiming: The operator stands in a safe area away from the water gauge, holding the portable handheld host 2 with one hand on the handle 215. Alternatively, the portable handheld host 2 can be fixedly mounted on a tripod for fixed-point observation. Press the power switch 211 to start the device. The image acquisition module 21 starts working and transmits the real-time image to the display screen 25. The operator adjusts the orientation of the host so that the reference scale 1 is fully displayed on the screen. The calculation and control unit 24 automatically identifies the barcode 14 and the spatial positioning code 15, reads the unique code of the scale and loads the preset zero elevation. Based on the pixel spacing and physical spacing of the spatial positioning code 15, a spatial reference frame is established between the host and the scale. At the same time, the electronic aiming crosshairs are superimposed on the screen 25.

[0057] Step 3, Trigger Measurement: The operator fine-tunes the attitude of the portable handheld host 2, aligning the electronic aiming crosshairs with the designated target point on the reference scale 1, and presses the measurement button 212 to trigger the measurement. At this time, the device simultaneously performs multi-sensor data acquisition. The laser ranging module 22 acquires the slant distance value from the host to the target point on the scale, the attitude sensing module 23 acquires attitude data such as the host's roll angle and pitch angle, and the image acquisition module 21 continuously acquires multiple frames of high-resolution on-site images within a preset time period.

[0058] Step 4, Data Calculation: The calculation and control unit 24, combining the acquired image frames, recalculates the spatial transformation relationship between the portable handheld host 2 and the reference scale 1 to determine the precise pose of the camera in the scale coordinate system. Tilt compensation is applied to the laser slant range value using pitch and roll angles, and optimized calculations are performed using the spatial reference frame to obtain the three-dimensional spatial coordinates of the host's optical center. Simultaneously, a water surface line extraction algorithm is run to identify the boundary line between the water body and the reference scale 1 frame by frame and convert it into the corresponding scale value. Then, a time-series filtering module is used to remove outliers and calculate the mean value for all scale data, outputting a stable average boundary line scale. If the scale encoding or scale recognition is abnormal, the device will automatically switch to a backup measurement mode, relying on the weir wall texture features to complete the water level calculation.

[0059] Step 5, Water Level Output and Upload: Combining the zero elevation of reference scale 1, the average boundary line scale, and the three-dimensional spatial coordinate correction, the current absolute water level elevation is calculated and displayed in real time on display screen 25. Wireless communication module 26 automatically uploads the measurement data packet to the remote monitoring platform. Operators can also manually upload data by pressing upload button 213. In case of system lag or malfunction, pressing reset button 214 will perform a system reset operation.

[0060] Step 6, Multi-point Survey: After completing the measurement at a single measuring point, the operator carries the portable handheld host 2 and walks to the next measuring point on the weir body, repeating steps 2 to 5 to complete the multi-point water level observation of the entire measurement area. Since each measuring point's reference scale 1 has a unique code, the measurement data can be automatically associated with the corresponding measuring point, completely avoiding the confusion caused by manual recording.

[0061] This embodiment relies on the collaborative operation of a fixed reflective ruler and a portable multi-sensor host, integrating multiple technologies such as visual positioning, laser ranging, and attitude error compensation to achieve long-distance, non-contact, high-precision water level measurement in the field. The entire operation process does not require additional power supply or communication facilities at the measurement point, and a single host can be used for multiple measurement points, effectively reducing the cost of multi-point surveys. It is particularly suitable for various hydrogeological observation scenarios such as karst channels, water measuring weirs, and flow measurement flumes.

[0062] Example 3 In another preferred embodiment, based on the above embodiment 1, this embodiment provides a portable long-distance water level observation device based on multi-sensor fusion. It is a portable long-distance water level observation device adapted to karst caves and narrow observation window scenarios. The overall structure is consistent with the core architecture of the above embodiment 1. It still consists of two main components: a reference scale 1 fixedly installed on the side wall of the weir 3 and a portable handheld host 2. The two work together to realize long-distance non-contact automatic water level measurement in low light and confined space.

[0063] Reference scale 1 is still made of corrosion-resistant metal substrate. The surface is laser-etched with basic division lines 11, auxiliary division lines 12, scale values ​​13, barcode 14, and spatial positioning code 15. The surface high-reflective coating uses high-brightness night vision reflective material to further improve the imaging clarity in the low-light environment of the cave. Barcode 14 stores the scale's unique identification code, installation point, elevation benchmark, and cave environment adaptation parameters; the vertically arranged spatial positioning code 15 maintains a fixed geometric size, continuously providing a spatial positioning benchmark for the portable handheld host 2. Reference scale 1 is installed strictly vertically, with the zero mark precisely aligned with the elevation benchmark at the bottom of the weir 3, adapting to the long-term high humidity and corrosive water vapor environment inside karst caves.

[0064] The portable handheld host 2 features a thickened waterproof and moisture-proof shell, enhancing its resistance to condensation and dust, making it suitable for the enclosed and humid environment of caves. The internal motherboard 29 integrates an image acquisition module 21, a laser ranging module 22, an attitude sensing module 23, a calculation and control unit 24, a wireless communication module 26, and a rechargeable battery 27. The rechargeable battery 27 is electrically connected to the image acquisition module 21, laser ranging module 22, attitude sensing module 23, calculation and control unit 24, display screen 25, wireless communication module 26, power switch 211, measurement button 212, upload button 213, and reset button 214. The charging port 28 is electrically connected to the rechargeable battery 27, providing power for charging.

[0065] An image acquisition lens 210 is located at the front of the device, and a high-brightness display screen 25 is embedded on the top surface. A power switch 211, a measurement button 212, an upload button 213, and a reset button 214 are arranged sequentially on the sides. The lower integrated grip 215 features an anti-slip texture, and a USB-C charging port 28 is located at the bottom. The power switch 211 controls the device's start and stop, the measurement button 212 triggers a single measurement, the upload button 213 enables manual data upload, and the reset button 214 is used for system malfunction reset.

[0066] The image acquisition module 21 is equipped with a high-definition industrial camera module. The image acquisition lens 210 optimizes optical zoom and low-light imaging performance, enabling complete image acquisition of the boundary area between the reference scale 1 and the water surface under low-light conditions in the cave. The laser ranging module 22 is a phase-type laser ranging sensor. The emitted optical axis is strictly parallel and calibrated with the optical axis of the image acquisition lens 210. The optical axis offset parameters are pre-stored in the calculation and control unit 24, allowing the long-distance laser spot to accurately fall on the scale marking area. This module can stably output ranging data. The attitude sensing module 23 consists of a three-axis accelerometer and a three-axis gyroscope, forming a six-axis inertial measurement unit. It has high angle measurement resolution and acquires the roll angle, pitch angle, and yaw angle of the host in real time, providing data support for tilt error compensation.

[0067] The computing and control unit 24 has a built-in embedded processor that integrates functions such as image recognition, water surface line extraction, temporal filtering, and backup measurement modes. During operation, the device uses PnP and homography matrix decomposition, combined with barcode 14 and spatial positioning code 15, to parse the relative spatial pose of the host and reference scale 1. It employs the Canny edge detection operator in conjunction with grayscale gradient analysis to identify the water surface boundary line and calculate the water level value based on the pixel-to-physical scale mapping relationship. The temporal filtering module performs extreme value removal and mean calculation on multiple frames of image data according to preset rules. Simultaneously, it can adaptively adjust the number of sampling frames and the outlier removal ratio based on the intensity of water flow fluctuations within the cave, suppressing reading deviations caused by water surface waves.

[0068] When the reference scale 1 is obscured by cave debris, silt, or severely damaged, making the barcode 14 and spatial positioning code 15 unrecognizable, the calculation and control unit 24 automatically activates the backup measurement mode: it prioritizes recognizing the texture features of the basic division line 11 and auxiliary division line 12 to establish a temporary reference scale; if the scale is completely unrecognizable, it calls the SIFT and ORB feature point detection algorithms to extract the natural rock texture of the weir channel 3 sidewall to construct a virtual reference coordinate system, and completes the scale calibration and water level elevation conversion with the help of auxiliary distance measurement to ensure continuous measurement under extreme working conditions.

[0069] The rechargeable battery 27 uses a high-capacity lithium battery to meet the needs of long-term operation in caves without external power supply. The wireless communication module 26 is compatible with 4G, 5G and Bluetooth (Bluetooth, short-range wireless communication technology), and can remotely upload measurement data packets containing scale codes, timestamps, water level elevation and attitude data to the monitoring platform to achieve unified data management and traceability.

[0070] Example 4 In another preferred embodiment, based on embodiments 1 to 3 above, this embodiment provides a portable long-distance water level observation method based on multi-sensor fusion, implemented using the portable long-distance water level observation device based on multi-sensor fusion described in embodiment 3. The specific operation steps are as follows: Step 1, Preliminary setup and benchmark calibration: In advance, the operators vertically fix the reference scale 1 on the side wall of the weir channel 3 of all the water measuring weirs in the measurement area. The level instrument is used to ensure that the scale is vertical. Then, the zero mark of the reference scale 1 is calibrated with the bottom elevation benchmark of the weir channel 3 one by one using a total station. The barcode 14 code and zero elevation parameter of each reference scale 1 are recorded to complete the hardware setup and parameter filing of all measuring points.

[0071] Step 2, Equipment Power-On and Point Aiming: The operator, carrying the portable handheld host 2, arrives at the first measuring point, stands in a safe area such as the bank or steep slope, holds the device with the handle 215, or the entire machine can be mounted and fixed on a tripod for long-term fixed-point observation, and presses the power switch 211 to start the entire machine. The image acquisition module 21 starts working, transmitting real-time images to the display screen 25. The operator adjusts the host angle so that the reference scale 1 is fully displayed in the lens field of view. The calculation and control unit 24 automatically identifies the barcode 14 and spatial positioning code 15 in the image, retrieves the preset elevation parameters of the measuring point, and establishes a spatial reference frame by combining the pixel spacing and physical spacing of the spatial positioning code 15. The electronic aiming crosshairs are then superimposed on the display screen 25 to guide precise aiming.

[0072] Step 3, Multi-source Data Synchronous Acquisition: The operator fine-tunes the attitude of the portable handheld host 2, aligning the center of the electronic aiming crosshair with the preset measurement area of ​​the reference scale 1, and presses the measurement button 212 to trigger the measurement task. At this time, the equipment synchronously starts working on all sensors: the laser ranging module 22 collects the slant distance value from the host to the target point of the reference scale 1, the attitude sensing module 23 collects attitude data such as the host's roll angle and pitch angle in real time, and the image acquisition module 21 continuously acquires multiple frames of high-definition images containing the scale and the water surface according to the set frame rate and duration.

[0073] Step 4: Multi-sensor data fusion and anomaly handling: The calculation and control unit 24 combines the acquired images to calculate the spatial transformation relationship between the portable handheld host 2 and the reference scale 1, determining the camera's three-dimensional pose; using attitude angle data, tilt compensation correction is performed on the laser slant range value to complete spatial coordinate optimization. The water surface line extraction algorithm is run, first performing grayscale conversion and Gaussian filtering preprocessing on the image, then using the Canny edge detection algorithm and grayscale gradient analysis to locate the water surface boundary line, converting it frame by frame into scale values. The temporal filtering module sorts all scale data, removes extreme values, and calculates the arithmetic mean. Simultaneously, it adaptively adjusts the sampling parameters according to water surface fluctuations, outputting a stable average water level scale.

[0074] If, during image recognition, the reference scale 1 is detected to be obstructed or damaged, causing the barcode 14 or spatial positioning code 15 to become invalid, the system will automatically switch to the backup measurement mode, sequentially calling the scale texture and weir wall rock texture features, and combining the SIFT and ORB algorithms to construct a virtual coordinate system to complete the water level calculation.

[0075] Step 5: Result Display and Data Upload: The calculation unit, combining the zero elevation of reference scale 1, the average water level scale, and the spatial coordinate correction, calculates the absolute water level elevation of the current measuring point and displays it in real time on the display screen 25. The wireless communication module 26 automatically uploads the data packet carrying the scale code, measurement timestamp, water level value, attitude angle, and positioning coordinates to the remote hydrological monitoring platform; when the network signal is poor, the operator can press the upload button 213 to manually cache or upload offline. When screen freezes, program crashes, or other malfunctions occur, pressing the reset button 214 will perform a system reset and restore normal equipment operation.

[0076] Step 6, Multi-point Cyclic Measurement: After the first measurement point is completed, the operator moves the portable handheld host 2 to the next water level measuring point and repeats the process of powering on, aiming, data acquisition, calculation, and data uploading from Steps 2 to 5. Since each reference scale 1 has a unique barcode 14, the measurement data can be automatically bound to the corresponding measuring point, completely avoiding the problem of confusion caused by manual recording. Following this process, water level observations are completed at all points within the measurement area. After a single full-area measurement is completed, the rechargeable battery 27 can be recharged via the USB-C interface.

[0077] This embodiment gradually realizes long-distance water level observation in karst caves and narrow observation window scenarios by following the above measurement steps, data processing methods, equipment operation logic and multi-point survey mode. It can be stably applied to various hydrological observation scenarios such as karst channels, field water measuring weirs, and flow measuring weirs.

[0078] In the preferred embodiment, the reference scale 1 is made of a corrosion-resistant metal substrate, with the surface marked using laser etching and covered with a highly reflective coating. The reference scale 1 is vertically positioned, with its zero mark aligned with the elevation benchmark at the bottom of the weir / channel 3. This configuration ensures that the reference scale 1 can operate stably for extended periods in high-humidity, sulfur- or carbonate-containing corrosive environments such as karst channels, preventing corrosion, deformation, or wear, thus guaranteeing the durability of the scale accuracy. The laser etching process ensures that the edges of the basic graduation line 11, auxiliary graduation line 12, scale values ​​13, barcode 14, and spatial positioning code 15 are clear and their dimensions are precise. The highly reflective coating, combined with the glass microsphere reflective film, allows the scale to be clearly imaged even in low light and at long distances. The strict alignment of the zero mark with the elevation benchmark at the bottom of the weir / channel 3 eliminates systematic errors caused by installation deviations, ensuring that the absolute accuracy of the water level elevation calculation is reliably guaranteed from the benchmark.

[0079] In a preferred embodiment, the image acquisition module 21 is equipped with an image acquisition lens 210, which has optical zoom capabilities. The emission optical axis of the laser ranging module 22 is parallel-calibrated with the optical axis of the image acquisition lens 210, with a parallelism deviation of less than 0.05°. This configuration ensures that the image acquisition lens 210 can completely capture the reference scale 1 and the water surface boundary area within a 5 to 10x optical zoom range, meeting imaging requirements at different observation distances. The strict parallel calibration of the two optical axes, with a deviation controlled within 0.05°, ensures that the projection position of the measurement spot of the laser ranging module 22 in the image is highly consistent with the actual illumination position. The image recognition result can be directly used to correct the ranging data, avoiding spatial position deviations introduced by parallax. This design tightly couples visual observation and laser ranging geometrically, providing precise geometric constraints for subsequent image-based spatial calculations and slant range correction, significantly improving the reliability of long-distance non-contact measurement.

[0080] In a preferred embodiment, the attitude sensing module 23 is a six-axis inertial measurement unit, integrating a three-axis accelerometer and a three-axis gyroscope, used to detect the pitch, roll, and yaw angles of the portable handheld host 2 in real time, with an angle measurement resolution of 0.01°. This configuration allows for real-time and accurate perception of attitude changes in the portable handheld host 2 caused by hand shake or body tilt during single-person operation. The high resolution of 0.01° ensures that even minute attitude deviations are not ignored, and the calculation and control unit 24 uses this information to precisely correct the tilt angle of the slant distance value acquired by the laser ranging module 22, eliminating measurement deviations caused by the host's non-horizontal state. Simultaneously, the complete output of pitch, roll, and yaw angles provides accurate spatial attitude parameters for the on-site images acquired by the image acquisition module 21, supporting algorithms such as PnP calculation or homography matrix decomposition to accurately reconstruct the spatial positional relationship between the handheld host 2 and the reference scale 1, ensuring overall calculation accuracy.

[0081] In a preferred embodiment, the calculation and control unit 24 incorporates a water surface line extraction algorithm module and a temporal filtering module. The water surface line extraction algorithm module identifies the boundary line between the water surface and the reference scale 1, while the temporal filtering module performs outlier removal and mean calculation on the recognition data from multiple frames of images. This configuration ensures that boundary line identification errors caused by water surface waves, light spot jitter, or partial occlusion in a single frame image are not directly transmitted to the final water level result. The water surface line extraction algorithm module locates the boundary line between the water body and the scale using Canny edge detection and grayscale gradient analysis, converting it into a scale value 13. The temporal filtering module removes outliers from multiple consecutive frames of data and takes the mean, effectively mitigating the instantaneous fluctuation interference caused by water surface pulsation. Working together, without increasing hardware costs, the two modules significantly improve the stability and repeatability of water level readings in a purely software-based manner, making the output result closer to the actual water level and meeting the requirements of hydrological monitoring for data continuity and reliability.

[0082] In the preferred embodiment, the calculation and control unit 24 overlays an electronic aiming crosshair onto the display screen 25 based on the image recognition result, guiding the laser ranging module 22 to align with the target point on the reference scale 1. When the reference scale 1 is partially obstructed or damaged, causing the code recognition to fail, the calculation and control unit 24 automatically switches to the backup measurement mode. This setup eliminates the need for operators to rely on experience for estimation; they only need to observe the electronic aiming crosshair on the display screen 25 to quickly align the laser spot with the target point on the reference scale 1, significantly reducing the difficulty of long-distance operation and improving the efficiency of single-person operation. When the barcode 14 or spatial positioning code 15 cannot be fully recognized due to obstruction or damage, the system automatically switches to the backup measurement mode, maintaining the measurement function without manual intervention. This avoids the loss of data for the entire measurement point due to localized damage, enhancing the device's fault tolerance and operational continuity in complex field environments.

[0083] In the preferred embodiment, under the backup measurement mode, the calculation and control unit 24 sequentially extracts the residual scale texture of the scale and the natural rock texture features of the weir sidewall to construct a virtual reference coordinate system, and combines this with the laser ranging module 22 to complete the water level elevation conversion. This setup enables the device to complete water level measurement even under extreme conditions where the reference scale 1 is severely damaged or heavily obscured. The system first attempts to establish a temporary reference scale using the local texture of the basic reticle 11 or auxiliary reticle 12. If the scale lines are also indistinguishable, it further utilizes the natural textures of the weir sidewall, such as joints, fissures, rock layer boundaries, and moss growth boundaries, to extract robust feature points using SIFT or ORB algorithms to construct a virtual reference coordinate system. The operator only needs to perform one auxiliary ranging measurement at a known feature point to complete the absolute scale calibration. This mechanism extends the device's applicability from normal operating conditions to extreme field scenarios, significantly enhancing environmental adaptability and robustness.

[0084] In a preferred embodiment, the portable handheld host 2 is further equipped with a power switch 211, an upload button 213, a reset button 214, and a grip 215. The host integrates a rechargeable battery 27, and a USB-C charging port 28 is located at the bottom. These features allow operators to hold the device with one hand via the grip 215, completing all operations from a safe area away from the water level gauge, thus ensuring personal safety during fieldwork. The power switch 211 controls the entire device's on / off state, the measurement button 212 triggers a single measurement, the upload button 213 supports manual data upload, and the reset button 214 is used for system reset. The functions of each button are clearly defined, and the operation logic is simple, reducing the risk of misoperation. The rechargeable battery 27, combined with the USB-C charging port 28, provides a convenient charging method. A full charge can support continuous survey operations for several days without frequent battery replacements or reliance on external power, meeting the endurance requirements for long-term independent fieldwork.

[0085] In the preferred embodiment, the laser ranging module 22 is a phase-type laser ranging sensor with a measurement wavelength of 640nm to 660nm, a measurement accuracy of ±1.5mm, and a maximum measurement range of 0.05m to 200m. These settings ensure that the laser ranging module 22 maintains a high measurement accuracy of ±1.5mm across a wide range of 0.05m to 200m, covering all application scenarios from close-range calibration to long-range observation. The 640nm to 660nm red light band has strong atmospheric penetration capabilities, maintaining stable ranging performance even in karst tunnel environments containing fog and dust. Compared to pulse-type measurement, the phase-type measurement principle offers higher accuracy and faster response speed. Combined with a narrow beam design, the spot diameter can be controlled within the width of the scale lines at long distances, ensuring that the laser spot accurately falls on the target area of ​​the reference scale 1, providing a hardware foundation for high-precision acquisition of slant range data.

[0086] In the preferred embodiment, in step 4, the water surface line extraction algorithm preprocesses the image using Gaussian filtering, and combines the Canny edge detection operator and gray-level gradient analysis to determine the water surface boundary line. The Canny algorithm sets the high threshold to 150 and the low threshold to 50. These settings enable the water surface line extraction algorithm to sensitively capture the abrupt changes in brightness at the boundary between the water body and the surface of the reference scale 1 while suppressing noise. Gaussian filtering effectively eliminates high-frequency noise and particle interference in the image. In the Canny algorithm's dual-threshold strategy, the high threshold of 150 is used to confirm strong edges, and the low threshold of 50 is used to track weak edges. The combination of these two can completely extract the boundary line contour, while suppressing the weak edge interference of the basic reticle 11 and the auxiliary reticle 12 themselves, avoiding the scale markings being misjudged as water surface boundaries. The gray-level gradient is calculated row by row along the vertical axis of the scale, and the maximum value is taken to locate the row position of the boundary line. This method has low computational load and strong real-time performance, making it suitable for efficient operation on the embedded computing and control unit 24.

[0087] In the preferred embodiment, the temporal filtering process in step 4 specifically involves: acquiring 30 frames of images within 3 seconds at a frame rate of 10 frames / second; removing the first 10% minimum values ​​and the last 10% maximum values ​​of the data sequence; and calculating the arithmetic mean of the remaining valid data as the water level scale value. This setup allows for the acquisition of 30 independent water level scale samples within 3 seconds for a single measurement. By removing the three extreme values ​​at each end and averaging the remaining 24 valid data, the impact of random interference such as water surface waves and light spot vibrations on the single measurement result is significantly reduced. Compared to single-frame recognition, this method reduces the random error of water level readings by approximately 40% to 60% without significantly increasing the measurement time, utilizing redundant sampling and statistical filtering, resulting in more stable and reliable output results. The 3-second acquisition duration and 10 frames / second frame rate achieve a good balance between measurement efficiency and data quality, making it suitable for most hydrological survey scenarios.

[0088] In the preferred embodiment, the time-series filtering adaptively adjusts the sampling parameters according to the degree of water surface fluctuation. When the data standard deviation is less than 2 mm, the number of sampling frames is reduced to 15 frames; when the data standard deviation is greater than 5 mm, the number of sampling frames is increased to 50 frames, and the outlier rejection ratio is increased to 20%. These settings enable the measurement strategy to dynamically match the actual water surface fluctuation state. Under calm water conditions with a standard deviation of less than 2 mm, 15 sampling frames are sufficient to obtain sufficient accuracy, shortening the measurement time and improving survey efficiency. Under severe fluctuations with a standard deviation greater than 5 mm, the system automatically increases the sampling to 50 frames and increases the outlier rejection ratio to 20%, using more data redundancy to offset fluctuation interference and ensure the reliability of the output water level reading. This adaptive mechanism avoids the applicability contradiction of fixed parameters under different water surface conditions, enabling the device to automatically adjust to the optimal working state in various hydrological scenarios, from still water tanks to rapid flow weirs.

[0089] In the preferred embodiment, in step 4, a pixel-scale mapping relationship is established using the pixel spacing of the spatial positioning code 15 and the actual physical spacing. Combined with trigonometric leveling and attitude angle compensation correction, the absolute water level elevation is finally calculated. This setup ensures that the water level elevation calculation does not rely on a single measurement method but integrates visual positioning, laser ranging, and attitude compensation constraints. The known physical spacing provided by the spatial positioning code 15 establishes a precise pixel-scale mapping scale with the image pixel spacing, directly converting the boundary line positions in the image into physical scale values ​​13. Based on this, the slant distance value from the laser ranging module 22, corrected by the pitch and roll angles of the attitude sensing module 23, and the absolute elevation at the zero point of the reference scale 1 are used to finally output the absolute water level elevation through trigonometric leveling relationships. Errors in each step are independent and traceable.

[0090] In the preferred embodiment, step 5 involves data uploading. The uploaded data includes a unique code for the reference scale 1, a measurement timestamp, water level elevation, attitude angle of the portable handheld host 2, and spatial coordinate information. Both automatic and manual upload modes are supported. This configuration ensures that each measurement data point is automatically associated with the unique code of the reference scale 1, allowing the backend to directly trace and manage data by measurement point, eliminating potential errors caused by manual recording. The complete recording of fields such as water level elevation, attitude angle, and spatial coordinates provides rich metadata support for subsequent long-term water level change analysis. The automatic upload mode pushes data to the remote platform immediately after measurement, meeting real-time monitoring needs. The manual upload mode, triggered by the upload button 213, is suitable for offline caching and batch uploading in areas with unstable network signals. The two modes complement each other, ensuring the flexibility and integrity of data transmission.

[0091] In summary, the portable long-distance water level observation device and method based on multi-sensor fusion provided by this invention effectively solves the problems of faded and soiled scales on existing wall-mounted water level gauges, difficulty in long-distance observation, unstable readings due to water surface fluctuations, and the difficulty in balancing portability and accuracy in existing improved solutions. It achieves the coordinated operation of a fixed reflective ruler and a portable handheld host 2, and automatically obtains accurate water level elevation at any observation location through multi-sensor fusion calculation of the image acquisition module 21, laser ranging module 22, and attitude sensing module 23, meeting the needs for long-distance, high-precision, and portable water level observation in complex field scenarios such as karst channel weirs 3.

[0092] This invention integrates a fixed reference scale 1 with a portable handheld host 2 into a single design. Through multi-sensor fusion calculation, it achieves long-distance, non-contact automatic acquisition of water level elevation, overcoming the limitations of traditional water level gauges that require close contact or pre-buried cables. The reference scale 1 is made of corrosion-resistant metal substrate, with its surface laser-etched to create basic graduations 11, auxiliary graduations 12, scale values ​​13, barcodes 14, and spatial positioning codes 15, and covered with a highly reflective coating. The zero-scale position is aligned with the bottom elevation benchmark of the weir 3, ensuring the durability of accuracy and the reliability of the benchmark for long-term field use from a hardware perspective.

[0093] In terms of geometric coupling, the emission optical axis of the laser ranging module 22 and the optical axis of the image acquisition lens 210 are strictly parallelized and the deviation is controlled within 0.05°, tightly coupling visual observation and laser ranging in geometric relationship. This allows the recognition results acquired by the image acquisition module 21 to be directly used to correct the ranging data, eliminating spatial position deviations introduced by parallax. The spatial positioning codes 15 distributed longitudinally on the reference scale 1 are black and white alternating coded patterns. Each coded pattern corresponds to a unique absolute position information, providing a spatial position reference benchmark for the portable handheld host 2, replacing the on-site calibration process of traditional total stations or GPS (Global Positioning System). The calculation and control unit 24 establishes a pixel-scale mapping scale based on this, directly converting the pixel row position of the boundary line into physical water level scale values, realizing on-site measurement and use without calibration.

[0094] To ensure measurement accuracy, the water surface line extraction algorithm employs Gaussian filtering to preprocess the image, combined with the Canny edge detection operator and grayscale gradient analysis to determine the water surface boundary line. The Canny algorithm uses a high threshold of 150 and a low threshold of 50 to specifically suppress weak edge interference from the basic reticle 11 and auxiliary reticle 12, preventing scale markings from being misidentified as water surface boundaries. The temporal filtering module acquires 30 frames of images at a frame rate of 10 frames / second within 3 seconds. After removing the top 10% minimum values ​​and the bottom 10% maximum values, the arithmetic mean of the remaining valid data is taken, effectively smoothing out water surface pulsation interference. Furthermore, the temporal filtering adaptively adjusts the sampling parameters based on the degree of water surface fluctuation. When the standard deviation of the data is less than 2mm, the number of sampling frames is reduced to 15 to improve efficiency; when the standard deviation is greater than 5mm, the number of sampling frames is increased to 50, and the outlier removal ratio is increased to 20%, achieving a dynamic balance between measurement efficiency and data quality.

[0095] In terms of ease of operation and fault tolerance, the computing and control unit 24 overlays the electronic aiming crosshairs on the display screen 25 in real time based on the image recognition results. The center of the crosshairs corresponds to the theoretical landing point of the laser ranging module 22's measured spot on the scale plane. The operator only needs to observe the display screen 25 to complete the long-distance accurate alignment. Combined with the grip 215, it can be operated with one hand, greatly reducing the difficulty of single-person operation. The barcode 14 and the spatial positioning code 15 constitute a dual identification mechanism. When the reference scale 1 is partially obscured or damaged, causing the code recognition to fail, the computing and control unit 24 automatically switches to the backup measurement mode. It sequentially extracts the residual scale texture of the basic division line 11 or auxiliary division line 12 and the natural rock texture features of the sidewall of the measuring weir 3. It extracts robust feature points through SIFT or ORB algorithms to construct a virtual reference coordinate system. After completing an auxiliary distance measurement with the laser ranging module 22, the absolute scale calibration can be established. Measurement can still be completed when the scale is severely damaged, and the measurement function can be maintained without manual intervention.

[0096] In terms of data calculation and output, the computing and control unit 24 uses an embedded high-performance processor as its core, integrating a complete calculation chain of PnP calculation, homography matrix decomposition, tilt compensation correction, and trigonometric leveling conversion, deeply fusing visual spatial positioning, precise laser ranging, and attitude error compensation. The attitude sensing module 23, as a six-axis inertial measurement unit, outputs the pitch, roll, and yaw angles of the portable handheld host 2 in real time, with an angle measurement resolution of 0.01°. Based on this, the computing and control unit 24 performs precise tilt compensation correction on the slant distance value obtained by the laser ranging module 22, and finally outputs the absolute water level elevation rather than a relative scale value. In step 5, the data upload includes the unique code of the reference scale 1, the measurement timestamp, the water level elevation value, the attitude angle of the portable handheld host 2, and spatial coordinate information. It supports both automatic and manual upload modes. The wireless communication module 26 pushes the data to a remote platform, facilitating data traceability management and long-sequence analysis in the background. The portable handheld host 2 has a built-in rechargeable battery 27, which can be charged through the charging port 28. When fully charged, it can meet the needs of continuous field survey operations for several days. The power switch 211, measurement button 212, upload button 213 and reset button 214 have clear functional divisions and a compact overall structure. A single unit can serve multiple measurement points, which greatly reduces the cost of multi-point surveys.

Claims

1. A portable long-range water level observation device based on multi-sensor fusion, characterized in that: The device includes a reference scale (1) fixedly installed on the side wall of the water measuring weir and a portable handheld host (2). The portable handheld host (2) and the reference scale (1) are used separately and in conjunction. The portable handheld host (2) completes image acquisition, distance measurement and attitude detection of the reference scale (1) from a safe position away from the reference scale (1). The water level elevation is obtained by multi-sensor fusion calculation. The surface of the reference scale (1) is provided with basic division lines (11), auxiliary division lines (12), scale values ​​(13), barcode (14) and spatial positioning code (15). The barcode (14) stores the unique coding information of the reference scale (1). The spatial positioning code (15) is laid out along the longitudinal direction of the reference scale (1) to provide spatial position reference. The portable handheld host (2) integrates an image acquisition module (21), a laser ranging module (22), an attitude sensing module (23), a calculation and control unit (24), a display screen (25) and a wireless communication module (26). It is also provided with a measurement button (212).

2. The portable long-distance water level observation device based on multi-sensor fusion according to claim 1, characterized in that: The reference scale (1) is made of corrosion-resistant metal substrate, and the surface is processed with laser etching to form the marking structure and covered with a high reflective coating. The reference scale (1) is vertically arranged, and its zero mark position is aligned with the bottom elevation benchmark of the weir (3).

3. The portable long-distance water level observation device based on multi-sensor fusion according to claim 1, characterized in that: The image acquisition module (21) is equipped with an image acquisition lens (210), which has an optical zoom function; the emission optical axis of the laser ranging module (22) and the optical axis of the image acquisition lens (210) are mutually parallel and calibrated, and the parallelism deviation between the two optical axes is less than 0.05°.

4. The portable long-distance water level observation device based on multi-sensor fusion according to claim 1, characterized in that: The attitude sensing module (23) is a six-axis inertial measurement unit that integrates a three-axis accelerometer and a three-axis gyroscope to detect the pitch angle, roll angle and yaw angle of the portable handheld host (2) in real time.

5. The portable long-distance water level observation device based on multi-sensor fusion according to claim 1, characterized in that: The calculation and control unit (24) has a built-in water surface line extraction algorithm module and a time-series filtering module. The water surface line extraction algorithm module is used to identify the boundary line between the water surface and the reference scale (1), and the time-series filtering module is used to remove outliers and calculate the mean value of the recognition data of multiple frames of images.

6. The portable long-distance water level observation device based on multi-sensor fusion according to claim 5, characterized in that: The calculation and control unit (24) overlays an electronic aiming crosshair on the display screen (25) according to the image recognition result, guiding the laser ranging module (22) to align with the target point of the reference scale (1); when the reference scale (1) is partially obscured or damaged, causing the code recognition to fail, the calculation and control unit (24) automatically switches to the backup measurement mode.

7. The portable long-distance water level observation device based on multi-sensor fusion according to claim 6, characterized in that: In the backup measurement mode, the calculation and control unit (24) sequentially extracts the residual scale texture of the scale and the natural rock texture features of the weir sidewall to construct a virtual reference coordinate system, and combines the laser ranging module (22) to complete the water level elevation conversion.

8. The portable long-distance water level observation device based on multi-sensor fusion according to claim 1, characterized in that: The portable handheld host (2) is also equipped with a power switch (211), an upload button (213), a reset button (214) and a grip (215). The host integrates a rechargeable battery (27) and has a charging port (28) at the bottom. The rechargeable battery (27) is electrically connected to the image acquisition module (21), the laser ranging module (22), the attitude sensing module (23), the calculation and control unit (24), the display screen (25), the wireless communication module (26), the power switch (211), the measurement button (212), the upload button (213), and the reset button (214). The charging port (28) is electrically connected to the rechargeable battery (27) to charge and power the rechargeable battery (27).

9. The portable long-distance water level observation device based on multi-sensor fusion according to claim 1, characterized in that: The laser ranging module (22) is a phase-type laser ranging sensor with a measurement wavelength of 640nm to 660nm, a measurement accuracy of ±1.5mm, and a maximum measurement range of 0.05m to 200m.

10. A portable long-distance water level observation method based on multi-sensor fusion, applied to the portable long-distance water level observation device based on multi-sensor fusion as described in any one of claims 1 to 9, characterized in that, Includes the following steps: Step 1, Measurement preparation: Fix the reference scale (1) vertically to the side wall of the measuring weir, install it vertically, and calibrate the zero mark of the reference scale (1) with the bottom elevation benchmark of the weir channel (3); Step 2, Power on and aim: The operator turns on the portable handheld host (2) in a safe area. The image acquisition module (21) acquires the scene. The host automatically recognizes the barcode (14) and spatial positioning code (15) and overlays the electronic aiming crosshairs on the display screen (25). Step 3, Trigger Measurement: Adjust the attitude of the portable handheld host (2) so that the electronic aiming crosshair is aligned with the target point of the reference scale (1), press the measurement button (212), and simultaneously collect the slope distance value, attitude angle and multiple consecutive frames of on-site images; Step 4, Data calculation: The calculation and control unit (24) completes tilt compensation and spatial position calculation, identifies the water surface boundary line scale through the water surface line extraction algorithm, and then obtains a stable average scale value through time-series filtering. Step 5, Result Output and Data Upload: Calculate the absolute water level elevation by combining the zero elevation of the reference scale (1) and display it on the display screen (25). Upload the measurement data through the wireless communication module (26). Step 6, Multi-point survey: Carry the portable handheld host (2) to the next measuring point and repeat steps 2 to 5 to complete the multi-point water level observation.

11. The portable long-distance water level observation method based on multi-sensor fusion according to claim 10, characterized in that: In step 4, the water surface line extraction algorithm uses Gaussian filtering to preprocess the image and combines the Canny edge detection operator with gray-level gradient analysis to determine the water surface boundary line.

12. The portable long-distance water level observation method based on multi-sensor fusion according to claim 10, characterized in that: The temporal filtering process described in step 4 is as follows: 30 frames of images are acquired within 3 seconds at a frame rate of 10 frames / second, the first 10% minimum values ​​and the last 10% maximum values ​​of the data sequence are removed, and the arithmetic mean of the remaining valid data is calculated as the water level scale value.

13. The portable long-distance water level observation method based on multi-sensor fusion according to claim 12, characterized in that: The time-series filtering adaptively adjusts the sampling parameters according to the degree of water surface fluctuation; when the data standard deviation is less than 2mm, the number of sampling frames is reduced to 15 frames; when the data standard deviation is greater than 5mm, the number of sampling frames is increased to 50 frames, and the outlier removal ratio is increased to 20%.

14. The portable long-distance water level observation method based on multi-sensor fusion according to claim 10, characterized in that: In step 4, the pixel spacing of the spatial positioning code (15) and the actual physical spacing are used to establish a pixel-scale mapping relationship. Combined with trigonometric elevation conversion and attitude angle compensation correction, the absolute water level elevation is finally calculated.

15. The portable long-distance water level observation method based on multi-sensor fusion according to claim 10, characterized in that: In step 5, data is uploaded. The uploaded data includes the unique code of the reference scale (1), the measurement timestamp, the water level elevation value, the attitude angle and spatial coordinate information of the portable handheld host (2), and supports both automatic and manual upload modes.

Images