A method and apparatus for measuring the flow velocity of water on a laboratory slope.

Abstract

This invention discloses a method and apparatus for measuring the flow velocity of water on a laboratory slope, belonging to the field of fluid linear velocity and motion state measurement technology. It includes acquiring the inlet and outlet cross-sectional width values ​​of a guide groove and extracting the pixel coordinates of the thin-layer water flow crest and bottom surface, calculating and generating a head loss compensation coefficient; receiving video frames of projected water jets and performing inter-frame difference operations to extract the coordinates of the detachment point and the contact point, and calculating and generating pixel values ​​of the horizontal displacement and vertical drop of the water jet; adding the component placement angle and the slope inclination angle to generate the actual projection angle value; and performing fluid projection algebraic solutions on the displacement, drop, angle values, and compensation coefficient to generate the initial flow velocity value. This invention employs a data-driven mechanism that integrates image coordinate analysis and kinematic equations, which can eliminate the interference of physical probes on the motion state of thin-layer water flow and improve the accuracy of local instantaneous linear velocity measurement.

Description

Technical Field

[0001] This invention relates to the field of fluid linear velocity and motion state measurement technology, and in particular to a method and apparatus for measuring the flow velocity of water on a laboratory slope. Background Technology

[0002] In physical model tests in fields such as water conservancy engineering, hydrogeology, and soil and water conservation, it is often necessary to accurately measure the linear velocity of surface runoff on laboratory slopes. Fluid linear velocity data is the core foundation for evaluating the surface water flow state and hydrodynamic characteristics.

[0003] In related technologies, Chinese utility model patent CN216900608U discloses a slope runoff velocity measuring instrument suitable for fixed-bed experiments, comprising: a starting point measuring device and an ending point measuring device; the starting point measuring device includes a retrievable rigid measuring tape, a vertical guide tube, and a stopwatch; the bottom of the vertical guide tube is equipped with a solution outlet valve, and the top of the vertical guide tube is connected to a liquid storage box; the bottom of the ending point measuring device is equipped with an ion sensor group. This device calculates the flow velocity by timing the dripping of a high-concentration electrolyte solution containing a dye, sensing ions through the ending point measuring device to stop the timing, and combining this with the slope length measured by the measuring tape.

[0004] The aforementioned technologies, which calculate flow velocity by adding electrolytes and recording the total time it takes to flow over a fixed distance, only obtain the average flow velocity over a relatively long slope. This cannot accurately measure the instantaneous linear velocity at a specific local location on the slope, and the overall measurement accuracy is low, making it difficult to meet the needs of high-precision laboratory velocity measurement and unfavorable for long-term fluid kinematics monitoring. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a method and apparatus for measuring the flow velocity of water on a laboratory slope. It employs a data-driven mechanism that integrates image coordinate analysis and kinematic equations, which can eliminate the interference of physical probes on the motion state of thin-layer water flow and improve the accuracy of local instantaneous linear velocity measurement.

[0006] The above objectives can be achieved through the following approach:

[0007] A method for measuring the flow velocity of water on a laboratory slope includes: acquiring the inlet cross-sectional width and outlet cross-sectional width of a guide groove; receiving a thin layer of water flowing through the guide groove; extracting the peak pixel coordinates and bottom pixel coordinates of the thin layer of water; performing a division operation on the inlet cross-sectional width and the outlet cross-sectional width to generate a cross-sectional contraction ratio; calculating the difference between the peak pixel coordinates and the bottom pixel coordinates to generate a standing wave height difference; performing a multiplication operation on the standing wave height difference and the cross-sectional contraction ratio to generate a head loss compensation coefficient; receiving video frames of the projected water jet as the thin layer of water exits the outlet of the guide groove; and analyzing the projected water jet... Inter-frame difference operations are performed on the video frames of the water tongue to extract the pixel coordinates of the detachment point and the water contact point. Horizontal and vertical difference calculations are then performed on the detachment point and water contact point pixel coordinates to generate the horizontal displacement pixel value and the vertical drop pixel value of the water tongue. The component placement angle value and the slope inclination angle value of the experimental slope are obtained from the bottom surface of the guide groove. An addition operation is performed on the component placement angle value and the slope inclination angle value to generate the actual water flow projection angle value. The horizontal displacement pixel value, the vertical drop pixel value, the actual water flow projection angle value, and the head loss compensation coefficient are then solved using fluid projection algebra to generate the initial flow velocity value.

[0008] Optionally, the step of extracting the peak pixel coordinates and bottom pixel coordinates of the thin-layer water flow includes: receiving a top-view projection image of the guide groove, extracting the contour boundary lines of the inner walls on both sides, calculating the lateral pixel distance of the inner wall contour boundary lines at the water flow inlet and outlet ends, and generating the inlet cross-sectional width value and the outlet cross-sectional width value; receiving a side-view image of the thin-layer water flow in the guide groove, performing grayscale threshold segmentation and contour connectivity analysis, extracting the extreme points of the vertical coordinates of the top boundary of the connectivity region to generate peak pixel coordinates, and extracting the set of horizontal straight line pixels at the solid-liquid interface to generate bottom pixel coordinates.

[0009] Optionally, generating the head loss compensation coefficient includes: extracting the inlet cross-sectional width value and the outlet cross-sectional width value, performing a division operation, and extracting the dimensionless quotient result as the cross-sectional contraction ratio value; parsing the two-dimensional matrix index of the wave crest pixel coordinates and the bottom pixel coordinates, stripping the horizontal coordinate data and extracting the vertical coordinate data, performing a subtraction operation to generate the standing wave bulge height difference value; and using the standing wave bulge height difference value as the base and performing a multiplication operation with the cross-sectional contraction ratio value to generate the head loss compensation coefficient.

[0010] Optionally, the extraction of the detachment point pixel coordinates and the water contact point pixel coordinates includes: extracting a first video frame and a second video frame from adjacent time series using the water jet video frame; performing pixel-level grayscale subtraction on the first video frame and the second video frame to generate a binarized motion difference matrix; traversing the binarized motion difference matrix to obtain continuous bright pixel clusters; extracting the initial divergence endpoint coordinates of the continuous bright pixel clusters to generate the detachment point pixel coordinates; and extracting the final truncated endpoint coordinates of the continuous bright pixel clusters to generate the water contact point pixel coordinates.

[0011] Optionally, the method further includes: multiplying the reciprocal of the head loss compensation coefficient with the pixel value of the horizontal displacement of the water tongue to generate a credibility score for the water tongue trajectory camera observation.

[0012] Optionally, generating the horizontal displacement pixel value and the vertical drop pixel value of the water tongue includes: performing a two-dimensional matrix dimension decomposition on the pixel coordinates of the detachment point and the pixel coordinates of the water contact point to generate the horizontal coordinate component of the detachment point, the vertical coordinate component of the detachment point, and the horizontal coordinate component and the vertical coordinate component of the water contact point; performing an absolute value operation on the difference between the horizontal coordinate component of the water contact point and the horizontal coordinate component of the detachment point to generate the horizontal displacement pixel value of the water tongue, and performing an absolute value operation on the difference between the vertical coordinate component of the water contact point and the vertical coordinate component of the detachment point to generate the vertical drop pixel value of the water tongue.

[0013] Optionally, generating the actual water flow projection angle value includes: receiving a first digital tilt sensor data packet from the bottom surface of the guide groove and a second digital tilt sensor data packet from the experimental slope, performing data load decoding, and extracting the component placement angle value and the slope tilt angle value; performing a binary floating-point addition instruction on the component placement angle value and the slope tilt angle value to generate the actual water flow projection angle value.

[0014] Optionally, generating the initial flow velocity value includes: substituting the horizontal displacement pixel value of the water tongue, the vertical drop pixel value of the water tongue, and the actual projection angle value of the water flow into a preset fluid projection mapping model to perform algebraic calculations to generate an uncompensated velocity variable; multiplying the uncompensated velocity variable with the head loss compensation coefficient to generate a compensated transition velocity value; and performing a weighted fusion calculation with the compensated transition velocity value and the credibility score value of the water tongue trajectory camera observation to generate the initial flow velocity value.

[0015] Based on the same inventive concept, this invention also provides a device for measuring the flow velocity of water on a laboratory slope. The device includes: an image parameter acquisition module, used to acquire the inlet cross-sectional width and outlet cross-sectional width values ​​of a flow guide groove, receive a thin layer of water flowing through the flow guide groove, and extract the peak pixel coordinates and bottom pixel coordinates of the thin layer of water; a loss compensation coefficient calculation module, used to perform a division operation between the inlet cross-sectional width and the outlet cross-sectional width values ​​to generate a cross-sectional contraction ratio value, perform a vertical coordinate difference calculation between the peak pixel coordinates and the bottom pixel coordinates to generate a standing wave height difference value, and perform a multiplication operation between the standing wave height difference value and the cross-sectional contraction ratio value to generate a head loss compensation coefficient; and a video frame parsing module, used to receive the projectiles of the thin layer of water leaving the outlet of the flow guide groove. The system employs a water jet video frame analysis technique, performing inter-frame difference operations to extract the pixel coordinates of the detachment point and the water contact point. A displacement pixel calculation module calculates the horizontal and vertical differences between the detachment and water contact point pixel coordinates to generate horizontal displacement and vertical drop pixel values ​​for the water jet. A projection angle calculation module obtains the component placement angle at the bottom of the guide groove and the slope inclination angle of the experimental slope, performing addition to generate the actual projection angle value. An initial velocity calculation module performs fluid projection algebra calculations on the horizontal displacement pixel value, the vertical drop pixel value, the actual projection angle value, and the head loss compensation coefficient to generate the initial velocity value.

[0016] Compared with the prior art, the present invention has the following advantages:

[0017] 1. By capturing video frames of the water jet and performing inter-frame difference operations, the coordinates of the water flow separation point and the water contact point are extracted using computer vision. This completely transforms the traditional manual distance and time measurement into objective image matrix analysis, eliminating the lag of subjective operation.

[0018] 2. By utilizing the cross-sectional contraction ratio characteristics of the flow guide groove, the head loss compensation coefficient is dynamically generated by extracting the height difference of the standing wave bulge generated by the water flow under pressure. This deeply integrates the internal physical loss with the external motion trajectory, realizing real-time automatic verification of energy loss.

[0019] 3. It minimizes physical interference in the velocity measurement process, avoiding the fatal flaw of traditional physical probes that must be submerged in water, thus disrupting the flow field in extremely shallow water. Simultaneously, all calculation steps are reduced to matrix extraction and algebraic operations at the data processing terminal. Only basic tilt angle values ​​and image streams are needed to automatically and continuously output flow velocity values, making it extremely suitable for long-term, unattended motion state monitoring in complex hydraulic model experiments.

[0020] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures pointed out in the description, claims and drawings. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the axial structure of the flow-lifting component of the measuring device in an embodiment of the present invention.

[0023] Figure 2 These are three views of the flow-lifting component of the measuring device in an embodiment of the present invention.

[0024] Figure 3 This is a side view of the general working principle of the measuring device on a slope in an embodiment of the present invention.

[0025] Figure 4 This is a flowchart illustrating a method for measuring the flow velocity of water on a laboratory slope according to an embodiment of the present invention.

[0026] Figure 5 This is a scatter plot showing the correlation between the reliability score of camera observation and the relative error of speed measurement in an embodiment of the present invention.

[0027] Figure 6 This is a diagram showing the evolution of flow velocity numerical distribution at different calculation stages in an embodiment of the present invention.

[0028] Figure 7 This is a schematic diagram of a device for measuring the flow velocity of water on a laboratory slope, according to an embodiment of the present invention.

[0029] Explanation of reference numerals in the attached diagram: 1-pointed wedge-shaped leading edge, 2-guide groove, 3-flow-lifting component, 4-experimental slope, 5-thin layer of water flow, 6-jetting water tongue, L-horizontal lifting distance, H-lifting height. - Slope angle - Slope angle of the flow-carrying component. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0031] Reference Figure 1 and Figure 2 Environmental data acquisition is based on a solid jetting component 3 that can be tightly attached to and fixed to the slope. This jetting component 3 is a rigid right-angled triangular block with a flat bottom, allowing it to be tightly fitted or fixed to the laboratory slope. The core structure of this component includes: a pointed wedge-shaped leading edge 1: located at the front end of the component in the direction of water flow, its cross-section is a sharp triangle. This design smoothly guides the thin layer of water flow to a proper upward turn, forming a clear parabolic trajectory while minimizing head loss. A guiding groove 2: located at the top of the component, it is a U-shaped or V-shaped groove that runs through the component along the water flow direction. Its function is to constrain and collect the thin layer of water flowing through the device, preventing the water flow from spreading to both sides, and ensuring that after flowing out, it forms a concentrated, stable, and easily captured jet of water by machine vision.

[0032] Reference Figure 3 This demonstrates the general operating state of the device on slope 4. In hydrogeological or hydraulic engineering model tests, when a very shallow, thin layer of water flows along the slope at an angle of... As the water flows down the experimental slope 4, it reaches the leading edge of the flow-lifting component at an unknown velocity. The thin layer of water 5, after being collected by the guide groove 2, flows along the slope angle... The inclined plane is forcibly guided and detaches from its restraint at the end of the component, projecting into the air to form a distinct jet of water 6. The image feature extraction module performs real-time high-speed video recording of this physical process, converting the physical trajectory in space into a two-dimensional image pixel matrix. By extracting the contact point and detachment point of the jet of water 6 under gravity, the corresponding horizontal jet distance L and jet height H are calculated. Finally, the initial velocity of the thin-layer water flow 5 is calculated using the underlying fluid projection mapping model of the computer. In particular, when the component slope angle... equal to slope angle At this point, the launch angle is equivalent to 0, and the thin layer of water is thrown horizontally.

[0033] Reference Figure 4 One embodiment of the present invention proposes a method for measuring the flow velocity of water on a laboratory slope. It adopts a data-driven mechanism that integrates image coordinate analysis and kinematic equations, which can eliminate the interference of physical probes on the motion state of thin-layer water flow and improve the accuracy of local instantaneous linear velocity measurement.

[0034] The method described in this embodiment specifically includes:

[0035] Obtain the inlet cross-sectional width and outlet cross-sectional width values ​​of the flow guide groove, receive the thin layer of water flowing through the flow guide groove, and extract the peak pixel coordinates and bottom pixel coordinates of the thin layer of water.

[0036] The inlet cross-section width value and the outlet cross-section width value are divided to generate the cross-section contraction ratio value. The peak pixel coordinate and the bottom pixel coordinate are calculated by the vertical coordinate difference to generate the standing wave bulge height difference value. The standing wave bulge height difference value and the cross-section contraction ratio value are multiplied to generate the head loss compensation coefficient.

[0037] Receive video frames of the thin layer of water flowing out of the outlet of the guide groove and ejecting water jets; perform inter-frame difference operation on the video frames of the ejecting water jets and extract the pixel coordinates of the detachment point and the pixel coordinates of the water contact point.

[0038] Perform horizontal and vertical difference calculations on the pixel coordinates of the detachment point and the pixel coordinates of the water contact point to generate the pixel values ​​of the horizontal displacement of the water tongue and the vertical drop of the water tongue.

[0039] Obtain the component placement angle value of the bottom surface of the flow guide groove and the slope inclination angle value of the experimental slope. Perform an addition operation on the component placement angle value and the slope inclination angle value to generate the actual water flow projection angle value.

[0040] The initial flow velocity value of the water flow is generated by performing fluid projection algebra solution on the pixel values ​​of the horizontal displacement of the water tongue, the pixel values ​​of the vertical drop of the water tongue, the actual projection angle of the water flow, and the head loss compensation coefficient.

[0041] Optionally, extracting the peak pixel coordinates and bottom pixel coordinates of the thin-layer water flow includes:

[0042] Receive the top-view projection image of the flow guide groove, extract the contour boundary lines of the inner walls on both sides, calculate the lateral pixel distance of the inner wall contour boundary lines at the water inlet and outlet ends, and generate the inlet cross-sectional width value and the outlet cross-sectional width value.

[0043] In the data acquisition and analysis phase targeting the cross-sectional features of the flow guide groove, a high-definition area array camera orthogonally hovering above the experimental water tank received a top-view projection image containing the global geometric features of the flow guide groove. When processing the pixel matrix of this image, a Gaussian smoothing filter was first applied to remove speckle noise caused by local reflections from the water surface. Subsequently, the connected set of pixels in the image matrix whose gray-level gradient changes exceeded adaptive high and low thresholds was extracted, thereby identifying the contour boundary lines of the inner walls on both sides. Since the flow guide groove has a strip-like structure extending along the water flow direction in image space, the image processing process locates the first region of interest (ROI) at the top of the image where the water flows in and the second region of interest (ROI) at the bottom of the image where the water flows out. Within the first ROI, the average horizontal pixel coordinates of the left inner wall and the average horizontal pixel coordinates of the right inner wall are obtained; the horizontal pixel distance between them is the inlet cross-sectional width. Similarly, the outlet cross-sectional width is calculated within the second ROI. The horizontal pixel distance is calculated based on a one-dimensional coordinate difference equation. The one-dimensional coordinate difference equation is:

[0044] ,

[0045] in, The pixel value representing the cross-sectional width to be determined has the physical meaning of the geometric net width of the flow guide groove under a specific observation section. The arithmetic mean of the x-coordinates of the set of pixels to the right of the inner wall contour boundary line in the image plane coordinate system; The arithmetic mean of the x-coordinates of the pixel set to the left of the inner wall contour boundary line in the same observation area. and All values ​​were directly extracted from the binary contour point array output by the Cannibal edge detection algorithm. Since there is a fixed calibration mapping relationship between the image resolution and the space of the physical experimental water tank, the width pixel value calculated through this difference equation can be used directly as a dimensionless feature for solving the hydrodynamic contraction ratio.

[0046] For example, assume that the resolution of the top-view projection image acquired by the high-definition area array camera is 1920×1080 pixels. Within the first region of interest at the water inlet end, the lateral mean of the set of pixel coordinates on the left inner wall is extracted. The horizontal mean of the set of pixel coordinates on the inner right wall is 320 pixels. The value is 1600 pixels. Substituting this into the one-dimensional coordinate difference equation, the inlet cross-section width is calculated to be 1600 minus 320, which equals 1280 pixels. Similarly, the outlet cross-section width is extracted and calculated to be 640 pixels in the second region of interest at the water flow separation end.

[0047] The system receives a side view image of the thin layer of water flow in the guide groove, performs grayscale threshold segmentation and contour connectivity analysis, extracts the extreme points of the vertical coordinates of the top boundary of the connectivity region to generate peak pixel coordinates, and extracts the set of horizontal straight line pixels at the solid-liquid interface to generate bottom pixel coordinates.

[0048] In the digital analysis of the longitudinal undulation morphology of thin-layer water flow, a high-speed industrial camera deployed outside the transparent glass observation window on the side of the water tank receives a side view image containing the standing wave motion morphology of the thin-layer water flow. To extract the effective contour of the water body from the complex laboratory background, the Otsu's method is used to perform adaptive grayscale thresholding on the side view image, transforming the multi-channel image into a binary image matrix with water as the foreground and blank background. Next, contour connected component analysis is performed on this binary image matrix, calculating the total pixel area of ​​each independent connected region, and retaining the main connected region with the largest area value as the physical object of the thin-layer water flow. This mathematically eliminates the interference of discrete high-brightness noise caused by water droplet splashing. Under the physical setting of a standard image coordinate system with the upper left corner as the origin and the downward direction as the positive vertical axis, the set of boundary pixels of the main connected region is traversed to find the point with the smallest vertical axis pixel coordinate value. This minimum extreme point accurately corresponds to the highest position of the standing wave bulge in the physical three-dimensional space. The two-dimensional coordinates of this point are extracted and the wave crest pixel coordinates are generated. Simultaneously, along the area below the water flow pattern, a high-contrast boundary line representing the physical substrate of the water tank is extracted from the original grayscale abrupt change region of the side view image. The coordinates of continuous pixels distributed along this horizontal boundary line are then extracted to generate the bottom pixel coordinates. By capturing these feature points, the absolute calibration of the dynamic bulge height of the thin-layer water flow within the pixel space is achieved.

[0049] Optionally, the generated head loss compensation coefficient includes:

[0050] Extract the inlet cross-sectional width value and the outlet cross-sectional width value, perform a division operation, and extract the dimensionless quotient value as the cross-sectional shrinkage ratio value;

[0051] When quantifying the squeezing effect of the flow channel's geometric boundary on the water flow, the inlet and outlet cross-sectional widths are read via a data bus. Because the thin-layer water flow undergoes forced morphological evolution and internal frictional energy loss when entering a narrow cross-section from a wide cross-section, this spatial contraction effect must be precisely digitized. Arithmetic logic operation instructions are invoked to perform floating-point division, using the inlet cross-sectional width as the dividend and the outlet cross-sectional width as the divisor. The formula for calculating the cross-sectional contraction ratio is defined as follows:

[0052] ,

[0053] in, The value representing the cross-sectional shrinkage ratio has the physical meaning of characterizing the degree of shrinkage in the three-dimensional geometric physical space of the guide groove in the direction of water flow. The value representing the inlet cross-section width is derived from the calculation result of the lateral pixel distance of the geometric boundary at the water inlet end; The value representing the outlet cross-sectional width is derived from the calculation of the lateral pixel distance of the geometric boundary at the water flow exit end. Since both the dividend and divisor use a uniform pixel as the physical dimension, the scale dependence of absolute spatial dimensions is completely eliminated through division, extracting a dimensionless quotient that reflects a purely relative proportional relationship. This dimensionless cross-sectional contraction ratio objectively reflects the relative strength of the constraint force exerted on the fluid by the external static physical boundary.

[0054] For example, if the inlet cross-sectional width of the upstream data stream is 1280 pixels and the outlet cross-sectional width is 640 pixels, the underlying calculation program uses a division operator to calculate a dimensionless quotient of 1280 divided by 640, which equals 2.0. This result of 2.0 is extracted and used as the cross-sectional contraction ratio, indicating that the channel contracts to half the inlet width at the outlet. This degree of contraction is the core external factor causing the loss of kinetic energy within the water flow.

[0055] The two-dimensional matrix index of the peak pixel coordinates and the bottom pixel coordinates is analyzed, the horizontal coordinate data is stripped and the vertical coordinate data is extracted, and a subtraction operation is performed to generate the difference in standing wave height.

[0056] When evaluating the dynamic response of thin-layer water flow caused by bottom friction and lateral contraction, it is necessary to quantify the absolute elevation change of the standing wave rise. The two-dimensional matrix sequence of wave crest pixel coordinates and bottom pixel coordinates stored in memory is retrieved. Since the potential energy characteristics of the standing wave are only related to spatial displacement along the direction of gravity, a matrix data dimension parsing operation is performed to remove the horizontal coordinate data from these two coordinate arrays, extracting only the vertical coordinate data along the direction perpendicular to gravity. In the standard computer vision two-dimensional matrix image coordinate system, the origin is fixed at the upper left corner of the image, and the vertical coordinate values ​​monotonically increase downwards. Therefore, the vertical coordinate data representing the bottom pixel coordinates, which is the physical reference of the channel bottom, must be numerically greater than the vertical coordinate data representing the wave crest pixel coordinates, which is the highest point of the water surface rise. The formula for the standing wave height difference is defined as:

[0057] ,

[0058] in, Representing the height difference of the standing wave bulge, its physical meaning is the absolute pixel height increment of the thin layer of water flow jumping upward in the guide groove due to lateral compression and longitudinal friction, which intuitively represents the scale of loss of kinetic energy converted into potential energy inside the water flow. The vertical coordinate data representing the bottom pixel coordinates is extracted from the vertical pixel value matrix of the horizontal straight line at the solid-liquid interface. The vertical coordinate data representing the pixel coordinates of the wave crest is extracted from the vertical pixel value matrix of the extreme points at the top boundary of the main connected domain of the water flow. Through this absolute subtraction operation, the global base translation error caused by the relative height of the camera installation is completely eliminated, generating the standing wave bulge height difference that corresponds only to the actual changes in the water head state.

[0059] For example, assume the peak pixel coordinates parsed from the underlying memory are a matrix array (600, 300), and the bottom pixel coordinates are represented as an array of arbitrary points on the horizontal line (X, 800). After performing data dimension parsing and stripping operations, the useless horizontal coordinate data 600 and X are discarded, and the vertical coordinate data 300 and 800 are precisely extracted. Substituting these into the standing wave height difference formula, the underlying operator performs a subtraction operation to obtain the standing wave bulge height difference as 800 minus 300, which equals 500 pixels.

[0060] The difference in standing wave height is used as the base and multiplied with the cross-sectional contraction ratio to generate a head loss compensation coefficient.

[0061] To establish a mathematical mapping relationship between the static geometric contraction ratio of the water flow channel and the dynamic standing wave response after kinetic energy loss, an adaptive energy loss evaluation index that does not require manual calibration needs to be constructed. The computational unit performs algebraic multiplication, using the difference in standing wave height, representing the dynamic response, as a base feature, and multiplies it by a scalar multiplication with the cross-sectional contraction ratio, representing the static constraint. The compensation coefficient mapping formula is defined as follows:

[0062] ,

[0063] in, Representing the head loss compensation coefficient, its physical meaning is to characterize the total energy loss composite equivalent caused by internal hydraulic friction and phase change of thin-layer water flow from entering the guide groove to the moment of ejection in a purely data-driven manner. This represents the difference in height between the standing waves; This represents the cross-sectional contraction ratio. The energy loss of the water flow is positively correlated with both the intensity of contraction and the standing wave performance. By multiplying these two factors, the kinetic energy loss effect under strong contraction conditions can be effectively amplified within the characteristic space, thereby generating a highly sensitive, non-manually-intervened head loss compensation coefficient. This purely data-driven characteristic parameter perfectly replaces the fixed friction constant in traditional empirical hydraulics, which heavily relies on expert estimation.

[0064] For example, based on the aforementioned logical operations, the extracted cross-sectional contraction ratio is known to be 2.0, and the standing wave bulge height difference is 500 pixels. The calculation unit retrieves the above parameters and performs a multiplication calculation, mapping 500 pixels to 2.0, i.e., 500 × 2.0 equals 1000. Thus, a head loss compensation coefficient of 1000, accurately reflecting the current instantaneous physical loss state, is generated. This data will be used as a correction variable to directly participate in the subsequent algebraic solution of the limiting flow velocity.

[0065] Optionally, the extraction of the pixel coordinates of the detachment point and the pixel coordinates of the water contact point includes:

[0066] The first and second video frames of adjacent time series are extracted using the video frames of the projected water tongue. The first video frame and the second video frame are then subjected to pixel-level grayscale subtraction to generate a binary motion difference matrix.

[0067] When tracking the transient spatial trajectory of a thin layer of water flow after it leaves a groove, the first and second video frames with strictly adjacent timestamps are extracted from the high-frequency data stream input from the high-speed camera. To eliminate the interference of light and color on the image matrix calculation, the program first converts the first and second video frames into single-channel grayscale matrices. Subsequently, a pixel-level grayscale subtraction operation is performed. The core physical purpose of this operation is to utilize the extremely short inter-frame time to remove the static experimental background in the image and separate the dynamic water kinetic energy coverage area where spatial displacement occurs. The pixel-level grayscale subtraction operation formula is defined as follows:

[0068] ,

[0069] in, It represents the absolute grayscale difference after differential processing of a specific coordinate point in a two-dimensional image matrix. Its physical meaning is the sudden change in optical reflection intensity caused by the water flow particle occupying that spatial position within a very short time interval. This represents the current grayscale value extracted from the corresponding pixel coordinate position in the second video frame; This represents the historical grayscale values ​​extracted from the same pixel coordinate position in the first video frame. (This is followed by a seemingly unrelated sentence about obtaining data containing a large number of pixels.) After obtaining the numerical difference grayscale matrix, an adaptive segmentation threshold is introduced to generate a well-defined binarized motion difference matrix. Addressing the limitation of traditional image processing where preset thresholds cannot adapt to varying laboratory lighting conditions, the threshold is dynamically obtained by extracting and calculating the mean square error of grayscale values ​​in static background regions without water flow in the first video frame. When the value is strictly greater than the adaptive segmentation threshold, the program assigns a maximum value representing the highlight at the corresponding coordinate position of the binary motion difference matrix; otherwise, it assigns a minimum value representing the static background.

[0070] For example, suppose the grayscale value of the first video frame at the coordinate point of 100 on the horizontal axis and 200 on the vertical axis is 45. A very short time later, splashing water particles pass through this spatial location, causing the grayscale value of the extracted second video frame at that point to jump to 205. Substituting into the pixel-level grayscale subtraction formula, the absolute grayscale difference is calculated to be |205-45|, which equals 160. At this point, the underlying algorithm calculates an adaptive segmentation threshold of 35 by analyzing the static background. Because 160 is much greater than 35, the program assigns the value of the binarized motion difference matrix at that coordinate point to be 255, accurately marking it as part of the moving water flow.

[0071] The binarized motion difference matrix is ​​traversed to obtain continuous bright pixel clusters. The coordinates of the initial divergence endpoints of the continuous bright pixel clusters are extracted to generate the coordinates of the detachment point pixels, and the coordinates of the final truncated endpoints of the continuous bright pixel clusters are extracted to generate the coordinates of the water-receiving point pixels.

[0072] After obtaining the binary motion difference matrix containing dynamic information, the eight-neighborhood region growing algorithm is used to perform a two-dimensional traversal of the entire matrix to find all connected regions in the highlighted state. To eliminate interference from isolated small water droplets caused by occasional splashes, the underlying algorithm calculates the total number of pixels contained in each connected region and extracts the core connected region with the largest total number of pixels as a continuous highlighted pixel cluster representing the main body of the water tongue movement. After locking this continuous highlighted pixel cluster, its edge geometric topological boundary needs to be analyzed along the fluid ejection direction. In the standard computer image coordinate system with the upper left corner as the origin, the moment the fluid leaves the guide groove is located at the position closest to the starting side of the entire parabolic trajectory, that is, it has the geometric feature of having the minimum horizontal pixel coordinate value. The program traverses all the outer boundary points of this continuous highlighted pixel cluster, retrieves and extracts the coordinate point with the minimum horizontal index value as the initial divergence endpoint coordinates, and then directly assigns them to generate the detachment point pixel coordinates. Similarly, at the moment the water tongue touches the water, it is at the farthest end of the horizontal displacement of the projectile trajectory and the lowest end of the gravity fall. The calculation program searches for and extracts the geometric limit boundary point that simultaneously satisfies the maximum horizontal index value and the maximum vertical index value in the continuous bright pixel cluster boundary. It uses this point as the end truncation endpoint coordinates and finally maps it to generate the pixel coordinates of the water touch point.

[0073] Optionally, the method further includes:

[0074] The reciprocal of the head loss compensation coefficient is multiplied by the pixel value of the horizontal displacement of the water tongue to generate a reliability score for the water tongue trajectory camera observation.

[0075] In the dynamic quantitative evaluation stage of optical image measurement data reliability, the output head loss compensation coefficient and water tongue horizontal displacement pixel value are retrieved. The head loss compensation coefficient reflects the severity of energy dissipation and flow turbulence within the water flow channel. A larger value indicates more severe and unstable water splashing during ejection, leading to increased dispersion of the landing point in the image matrix and reduced image measurement accuracy. The water tongue horizontal displacement pixel value characterizes the spatial span of the ejection trajectory on the two-dimensional observation plane. A larger value indicates a more extended parabola, less impact of image spatial resolution on the relative error of extracted coordinate points, and higher clarity and measurability of optical capture. To integrate these two heterogeneous physical indicators representing negative correlation measurement interference and positive correlation measurement accuracy, the reciprocal of the head loss compensation coefficient is first calculated, converting the error amplification factor into an accuracy gain factor. Then, an algebraic multiplication operation is performed, multiplying the calculated reciprocal value by the water tongue horizontal displacement pixel value. The reliability scoring formula is defined as follows:

[0076] ,

[0077] in, The value representing the credibility score of the water tongue trajectory camera observation is, in physical terms, the overall confidence level of the water tongue spatial coordinate data obtained by machine vision tracking technology under the current hydraulic conditions. The pixel value representing the horizontal displacement of the water tongue is calculated from the absolute difference between the horizontal coordinate component of the contact point and the horizontal coordinate component of the detachment point, and its unit of measurement is pixels. The head loss compensation coefficient is derived from the product of the difference in standing wave height and the cross-sectional area contraction ratio, and its unit of measurement is also pixels. Because it is a multiplier... With the denominator Both physical units are pixels. After multiplication and division, their units cancel each other out, resulting in a final output confidence score for the water tongue trajectory camera observation that is a purely dimensionless scalar. This dimensionless scalar provides an absolute data-driven evaluation basis for judging the validity of optical measurement data or for allocating weights in weighted fusion algorithms.

[0078] For example, in an experiment with a relatively stable thin-layer water flow and a relatively high initial projectile velocity, the horizontal displacement pixel value of the water tongue read from the memory register was 800 pixels, and the head loss compensation coefficient was 40 pixels. First, the reciprocal of the head loss compensation coefficient 40 was calculated to obtain 0.025. Then, a multiplication operation was performed, multiplying 800 by 0.025, i.e., 800 × 0.025 equals 20.0. The final output confidence score for the water tongue trajectory camera observation was 20.0. This high score indicates from a data perspective that the current water flow has a long projectile arc and a regular flow pattern, and the extracted pixel coordinates of the departure point and the landing point have significant physical reference value, indicating excellent data quality. For example, in another extreme test condition, due to the extremely high flow rate and sharp contraction of the cross-section at the inlet of the guide groove, violent hydraulic jump turbulence was triggered. The microprocessor read a head loss compensation coefficient as high as 500 pixels, and the water flow broke apart and scattered immediately after exiting the groove, with the measured horizontal displacement of the water tongue being only 150 pixels. The calculation program performed a reciprocal operation on 500, obtaining 0.002. Then, 150 and 0.002 were multiplied, i.e., 150 × 0.002 equals 0.3. At this time, the generated confidence score of the water tongue trajectory camera observation was only 0.3. The extremely low score accurately quantifies the risk of image resolution distortion caused by severe splashing and excessively short motion trajectories in the current flow field, providing a clear quantitative warning for reducing the trust weight of this frame of image data in the calculation process. Figure 5 As shown in the figure, the nonlinear negative correlation between the extracted confidence score and the final flow velocity observation error is revealed. As can be seen from the figure, as the confidence score increases, the relative error of the sample point velocity measurement shows an exponential decreasing trend accompanied by a sharp narrowing of the confidence interval. This proves the scientific validity and noise resistance effectiveness of using this pure image-derived index as a flow velocity weighted fusion factor from the perspective of physical data.

[0079] Optionally, the generated pixel values ​​for the horizontal displacement of the water tongue and the pixel values ​​for the vertical drop of the water tongue include:

[0080] The pixel coordinates of the detachment point and the pixel coordinates of the water contact point are decomposed into a two-dimensional matrix to generate the x-coordinate component of the detachment point, the y-coordinate component of the detachment point, and the x-coordinate component and the y-coordinate component of the water contact point.

[0081] When converting spatial anchor points into independent scalar parameters suitable for algebraic operations, the algorithm receives the pixel coordinates of the departure point and the pixel coordinates of the landing point. Since these two physical spatial coordinates are typically stored in computer memory as composite two-dimensional data tuples or matrix indexes, direct cross-dimensional distance calibration is not possible. Therefore, the underlying computation program must perform a two-dimensional matrix dimensional decomposition operation. Through addressing calls and index resolution rules, the computation program thoroughly deconstructs this composite data structure of the departure point pixel coordinates, extracting the x-coordinate component representing the horizontal axis spatial position and the y-coordinate component representing the vertical gravity axis spatial position. Similarly, the computation program performs the same index-level decomposition operation on the landing point pixel coordinates, accurately separating the x-coordinate and y-coordinate components of the landing point. This decomposition process is a necessary data reconstruction operation connecting the geometric features of image pixels with the one-dimensional motion equations of classical mechanics, ensuring the absolute numerical independence of the projectile trajectory in the horizontal and vertical orthogonal dimensions, and avoiding dimensional coupling pollution in subsequent computations.

[0082] The absolute value of the difference between the horizontal coordinate component of the water contact point and the horizontal coordinate component of the water separation point is calculated to generate the horizontal displacement pixel value of the water tongue. The absolute value of the difference between the vertical coordinate component of the water contact point and the vertical coordinate component of the water separation point is calculated to generate the vertical drop pixel value of the water tongue.

[0083] After obtaining the independent coordinate scalars in each orthogonal dimension, the microprocessor needs to accurately quantify the span of the macroscopic geometric trajectory traced by the projectile water jet in the two-dimensional observation plane. To eliminate the physical ambiguity of the sign of the coordinate difference caused by differences in the camera's installation orientation or the definition of the origin of the underlying image matrix coordinate system, and thus ensure that the output geometric span is always a non-negative scalar conforming to objective mechanical logic, the underlying arithmetic execution engine strictly calls the difference calculation instruction with an absolute value operator. First, the horizontal coordinate components of the contact point and the detachment point are substituted into the horizontal span equation, and the first-level absolute value operation of the difference is performed. The horizontal span equation is defined as follows:

[0084] ,

[0085] in, The pixel value representing the horizontal displacement of the water tongue has the physical meaning of the total absolute width of pixels spanned in the horizontal range direction when the thin layer of water flows in projectile motion. Represents the x-coordinate component of the water point; This represents the x-coordinate component of the breakaway point. Both are direct outputs from the dimensional decomposition operation of a two-dimensional matrix. Subsequently, the underlying arithmetic execution engine substitutes the y-coordinate components of the landing point and the breakaway point into the vertical span equation, performing the second-level absolute value operation of the difference. The vertical span equation is defined as follows:

[0086] ,

[0087] in, The pixel value representing the vertical drop of the water tongue has the physical meaning of the total absolute height difference of the pixels during the period from when the water flow leaves the bottom constraint of the guide groove to when it hits the boundary of the specified liquid surface. Represents the ordinate component of the water point; This represents the ordinate component of the departure point. Due to the mandatory absolute value constraint, regardless of whether the actual array index value of the water contact point is greater than or less than the array index value of the departure point, the final calculated pixel values ​​of the horizontal displacement and vertical drop of the water tongue are positive scalars that purely reflect the absolute amplitude of the projectile displacement. This provides the most core basic geometric input variables for solving the initial velocity based on the algebraic equation of the projectile trajectory.

[0088] For example, the underlying arithmetic execution engine retrieves the four independent components decomposed from the memory register: the horizontal coordinate component of the water contact point (890), the horizontal coordinate component of the water exit point (350), the vertical coordinate component of the water contact point (950), and the vertical coordinate component of the water exit point (420). First, a horizontal calculation is performed: 890 minus 350 equals 540. Taking the absolute value of 540 also results in 540, thus generating a horizontal displacement pixel value of 540 pixels for the water tongue. Second, a vertical calculation is performed: 950 minus 420 equals 530. Taking the absolute value of 530 also results in 530, thus generating a vertical drop pixel value of 530 pixels for the water tongue. These two objective values ​​rigorously quantify the two-dimensional macroscopic scale of the projectile water tongue.

[0089] Optionally, the actual projection angle of the generated water flow includes:

[0090] Receive the first digital tilt angle sensor data packet from the bottom surface of the flow guide groove and the second digital tilt angle sensor data packet from the experimental slope, perform data load decoding, and extract the component placement angle value and the slope tilt angle value;

[0091] In the stage of obtaining the physical angle reference that determines the initial direction of fluid projection, a sequence of digital signals sent from the experimental device hardware is received in parallel via the underlying hardware communication bus. The first digital tilt sensor is physically attached to the bottom surface of the guide groove to monitor the relative tilt angle of the groove to the slope; the second digital tilt sensor is fixed to the experimental slope reference structure to monitor the absolute tilt angle of the slope relative to the absolute horizontal plane. Upon receiving the underlying binary bitstream containing check bits, identifiers, and data bits—that is, the data packets for the first and second digital tilt sensors—the underlying driver immediately performs a data payload decoding operation. This decoding operation first verifies the data integrity and removes the frame header, frame tail, and cyclic redundancy check code, extracting the core data payload segment representing the original angle information. Subsequently, strictly following the resolution quantization factor specified by the sensor's underlying communication protocol, the hexadecimal integer payload data is converted into standard decimal floating-point numbers with actual physical dimensions, thereby extracting the component placement angle and slope tilt angle values. These two values ​​are the absolute angle starting reference for subsequent projection trajectory calculation.

[0092] The component placement angle value and the slope inclination angle value are added using a binary floating-point addition instruction to generate the actual water flow projection angle value.

[0093] After analyzing the local and global angle parameters, it is necessary to determine the comprehensive projection angle of the water flow relative to the absolute horizontal plane of nature at the instant it leaves the guide groove. The analyzed component placement angle and slope inclination angle values ​​are then loaded into the underlying arithmetic execution core, which calls the hardware-level binary floating-point addition instruction to perform a mathematical summation operation. The projection angle synthesis formula is defined as follows:

[0094] ,

[0095] in, It represents the actual projection angle of the water flow. Its physical meaning is the real physical angle between the tangent of the velocity vector of the water jet particle and the equipotential surface of the Earth's gravity at the instant it leaves the physical boundary constraint of the guide groove. The slope inclination angle value is derived from the data payload decoding result of the second digital tilt sensor data packet, and objectively characterizes the global tilt reference of the experimental slope. The component placement angle value is derived from the data payload decoding result of the first digital tilt sensor data packet, objectively representing the local elevation angle increment of the guide groove above the slope reference. Since physical spatial angles are vector additivity in the same plane, by executing binary floating-point addition instructions to sum these two values, the interference of manual measurement and experimental device assembly errors on the measurement of the initial fluid motion direction can be completely eliminated, generating the actual water flow projection angle value, providing an indispensable initial angle input for solving the fluid projection algebraic equation.

[0096] For example, the underlying arithmetic execution core retrieves the floating-point parameters extracted in the aforementioned steps: the component placement angle of 5.2 degrees and the slope inclination angle of 15.0 degrees. The arithmetic execution core executes a binary floating-point addition instruction, that is, it completes the floating-point arithmetic operation of adding 15.0 to 5.2, and outputs a high-precision sum of 20.2 degrees. This value of 20.2 degrees is then generated and assigned as the actual projection angle of the water flow. It accurately records, in a purely data-driven manner, that the thin layer of water flow is being projected into the air at an angle of 20.2 degrees to the absolute horizontal plane.

[0097] Optionally, the initial flow velocity value of the generated water flow includes:

[0098] Substitute the pixel values ​​of the horizontal displacement of the water tongue, the pixel values ​​of the vertical drop of the water tongue, and the actual projection angle of the water flow into a preset fluid projection mapping model to perform algebraic calculations and generate uncompensated velocity variables.

[0099] In the cross-dimensional mapping solution stage between physical mechanics theory and image pixel space, the underlying fluid projectile mapping model is invoked. To resolutely eliminate the uncertainty brought about by subjective human assumptions, this fluid projectile mapping model is constructed entirely based on the objective Newtonian classical kinematic projectile trajectory equations. To clarify the mapping process, the theoretical origins of the classical kinematic projectile trajectory equations are traced. In traditional manual physical measurements, the core parameters of the device include the horizontal distance. , lift height Slope angle and the slope angle of the jet faucet components The water flow has an initial velocity of The particle is ejected along the leading edge of the jet-propellant component. Under a simplified physical model that neglects air resistance, the particle undergoes projectile motion. Based on the kinematic equations of the projectile, its horizontal and vertical displacements satisfy the following mathematical relationships:

[0100] ,

[0101] ,

[0102] The classical physical formula for solving the initial velocity can be derived from the above system of simultaneous equations:

[0103] ,

[0104] in, Represents the initial absolute velocity of the water flow at the instant it leaves the device; The horizontal distance representing the water jet; The height of the jet of water; Represents the constant of natural gravitational acceleration; The slope angle represents the experimental slope relative to the absolute horizontal plane; This represents the slope angle of the inclined surface of the cantilever component itself. Specifically, when... At that time, the emission angle of the fluid particles At this point, the water flow is equivalent to projectile motion. Substituting the two-dimensional spatial geometric span and the initial launch direction (i.e., the pixel values ​​of the horizontal displacement and vertical drop of the water tongue) and the actual projection angle into this equation, and through inverse algebraic calculation, the ideal projection velocity undisturbed by friction is obtained. The algebraic calculation formula for the fluid projection mapping model is defined as follows:

[0105] ,

[0106] in, Represents the uncompensated velocity variable to be generated. Its physical meaning is the initial linear velocity scalar of the water tongue when it leaves the guide groove in an ideal vacuum without considering air resistance and water head energy loss. The unit is pixels per frame. Represents the pixel value of the horizontal displacement of the water tongue; Represents the pixel value of the vertical drop of the water tongue; This represents the actual projection angle of the water flow. This represents the pixel's gravitational acceleration constant. This pixel gravitational acceleration constant is not a value guessed based on human experience, but rather obtained through rigorous objective physical mapping and calculation. Its value is determined by dividing the actual physical gravitational acceleration by the physical size of a single pixel on the camera device, and then dividing by the square of the camera's frame rate. This objective algebraic operation generates an uncompensated velocity variable that purely follows the first principles of kinematics.

[0107] For example, given that the camera's frame rate is 100 frames per second, the physical size of a single pixel is 0.001 meters, and the underlying computational program precisely calculates the pixel gravitational acceleration constant to be 0.98 pixels per square frame, the microprocessor retrieves the previously calculated horizontal displacement of the water jet (500 pixels), vertical drop (200 pixels), and the actual water jet projection angle (0 degrees). Substituting these values ​​into the fluid projection mapping model formula, the core calculation yields the following internal term: The absolute value is 245000 / 400, which equals 612.5. After performing the square root operation, the uncompensated velocity variable is approximately 24.7 pixels per frame.

[0108] The uncompensated velocity variable is multiplied by the head loss compensation coefficient to generate a compensated transition velocity value.

[0109] The uncompensated velocity variable represents a theoretically derived value under ideal conditions. It does not account for the work done by boundary friction within the guide groove or the internal energy loss caused by forced cross-sectional contraction. Therefore, this internal loss needs to be incorporated into the velocity calculation. To address the issue of dimensional consistency and physical meaning alignment, a head loss compensation coefficient per pixel is extracted, and a unit normalization factor is introduced. Subsequently, a multiplication operation is performed, using this unit normalization factor to convert the head loss compensation coefficient into a dimensionless kinetic energy attenuation proportionality coefficient, thereby proportionally reducing the uncompensated velocity variable. The formula for this multiplication operation is defined as follows:

[0110] ,

[0111] in, The value representing the generated compensated transition velocity is, in physical terms, the actual residual linear velocity of the water flow after accurately subtracting the energy consumed by the hydraulic jump standing wave and cross-sectional contraction. Represents the uncompensated velocity variable; Represents the head loss compensation coefficient; The normalization factor, calibrated using standard hydraulic measuring instruments, represents the baseline percentage of velocity attenuation corresponding to the standing wave bulge per unit pixel. This multiplication operation transforms the static three-dimensional geometric physical resistance into a dimension-reduced attenuation correction for dynamic linear velocity at a purely algebraic level, generating a compensated transition velocity value with high real-world accuracy.

[0112] The compensated transition velocity value and the credibility score value of the water tongue trajectory camera observation are weighted and fused to generate the initial flow velocity value.

[0113] The instantaneous flow velocity captured in a single frame image is inevitably affected by random physical fluctuations in the flow field and observation noise from optical sensors. To obtain a macroscopically stable flow velocity scalar, a temporal memory smoothing mechanism is introduced in this final stage to perform weighted fusion calculations. The historical smoothed flow velocity variables cached in the previous computation cycle are called up, and the currently independently generated water tongue trajectory camera observation confidence score value is extracted as the core for dynamic weight allocation. By applying a normalized activation mapping, the water tongue trajectory camera observation confidence score value is mapped to a confidence weight scalar between 0 and 1. Subsequently, this confidence weight scalar is multiplied by the current compensated transition flow velocity value, and the remaining percentage weight is multiplied by the historical smoothed flow velocity variable to complete the final weighted sum mathematical fusion. The weighted fusion calculation formula is defined as follows:

[0114] ,

[0115] in, This represents the initial flow velocity value of the water, which is the final output high-frequency, high-precision velocity measurement result; This represents a confidence weight scalar generated by mapping the confidence score of the water tongue trajectory camera observations. This represents the current generated compensation transition velocity value; This represents the historical smoothed flow rate variable in the cache. This calculation completely transforms the absolute sharpness of optical observations into data confidence in velocity values, achieving self-closing-loop error cleanup in machine vision measurements.

[0116] For example, the compensated transition velocity value generated by the current operation is 20.0 pixels per frame, and the historical smoothed velocity variable cached in memory is 18.0 pixels per frame. The microprocessor extracts a very high confidence score value for the current water tongue trajectory camera observation, indicating that the trajectory captured in the current image frame is extremely clear and stable, and the confidence weight scalar generated by mapping is 0.9. Substituting into the fusion formula, the current high-confidence observation value occupies the dominant weight: 0.9 × 20.0 plus (1 minus 0.9) × 18.0 equals 18.0 plus 1.8, resulting in 19.8 pixels per frame. The microprocessor finally outputs an initial water flow velocity value of 19.8 pixels per frame. This result not only completes the dynamic compensation in the deep dimension of hydraulics, but also embeds an adaptive calibration and error correction mechanism based on visual tracking quality. Figure 6 As shown in the figure, the adaptive cleaning and noise reduction process of the multi-order algebraic solution model of the present invention for the observation data is demonstrated. As can be seen from the figure, the uncompensated velocity variable has a large discrete variance and system bias. However, after the head loss compensation dimensionality reduction and the weighted fusion of the credibility of the camera observation, the final generated initial flow velocity value not only returns to the objective real level, but its data fluctuation range is also extremely converged, which improves the anti-interference stability of machine vision velocity measurement under complex water flow conditions.

[0117] Based on the same inventive concept, the present invention also provides a device for measuring the flow velocity of water on a laboratory slope, such as... Figure 7 As shown, the device includes:

[0118] The image parameter acquisition module is used to acquire the inlet cross-sectional width and outlet cross-sectional width values ​​of the flow guide groove, receive the thin layer of water flowing through the flow guide groove, and extract the peak pixel coordinates and bottom pixel coordinates of the thin layer of water.

[0119] The loss compensation coefficient calculation module is used to perform a division operation between the inlet cross-section width value and the outlet cross-section width value to generate a cross-section contraction ratio value, perform a vertical coordinate difference calculation between the wave crest pixel coordinate and the bottom pixel coordinate to generate a standing wave bulge height difference value, and perform a multiplication operation between the standing wave bulge height difference value and the cross-section contraction ratio value to generate a head loss compensation coefficient.

[0120] The video frame parsing module is used to receive video frames of the thin layer of water flowing away from the outlet of the guide groove and projecting water tongue, perform inter-frame difference operation on the projecting water tongue video frames, and extract the pixel coordinates of the detachment point and the pixel coordinates of the water contact point.

[0121] The displacement pixel calculation module is used to perform horizontal and vertical difference calculations on the pixel coordinates of the detachment point and the pixel coordinates of the water contact point to generate the horizontal displacement pixel value and the vertical drop pixel value of the water tongue.

[0122] The projectile angle calculation module is used to obtain the component placement angle value of the bottom surface of the guide groove and the slope inclination angle value of the experimental slope, and perform an addition operation on the component placement angle value and the slope inclination angle value to generate the actual projectile angle value of the water flow.

[0123] The initial flow velocity calculation module is used to perform fluid projection algebra calculation on the pixel values ​​of the horizontal displacement of the water tongue, the pixel values ​​of the vertical drop of the water tongue, the actual projection angle of the water flow, and the head loss compensation coefficient to generate the initial flow velocity value of the water flow.

[0124] It should be noted that the functional division and information interaction between the various modules described above are logical, but in terms of physical implementation, they can be integrated on the same software platform or deployed in a distributed manner. The connections between them represent data flow and control flow, aiming to collaboratively achieve the objectives of this invention. The above descriptions are merely exemplary embodiments of this invention and should not be construed as limiting the scope of protection of this invention.

Claims

1. A method of measuring the flow rate of a laboratory hillslope water flow, characterized by, The method includes: Obtain the inlet cross-sectional width and outlet cross-sectional width values ​​of the flow guide groove, receive the thin layer of water flowing through the flow guide groove, and extract the peak pixel coordinates and bottom pixel coordinates of the thin layer of water. The inlet cross-section width value and the outlet cross-section width value are divided to generate the cross-section contraction ratio value. The peak pixel coordinate and the bottom pixel coordinate are calculated by the vertical coordinate difference to generate the standing wave bulge height difference value. The standing wave bulge height difference value and the cross-section contraction ratio value are multiplied to generate the head loss compensation coefficient. Receive video frames of the thin layer of water flowing out of the outlet of the guide groove and ejecting water jets; perform inter-frame difference operation on the video frames of the ejecting water jets and extract the pixel coordinates of the detachment point and the pixel coordinates of the water contact point. Perform horizontal and vertical difference calculations on the pixel coordinates of the detachment point and the pixel coordinates of the water contact point to generate the pixel values ​​of the horizontal displacement of the water tongue and the vertical drop of the water tongue. Obtain the component placement angle value of the bottom surface of the flow guide groove and the slope inclination angle value of the experimental slope. Perform an addition operation on the component placement angle value and the slope inclination angle value to generate the actual water flow projection angle value. The initial flow velocity value of the water flow is generated by performing fluid projection algebra solution on the pixel values ​​of the horizontal displacement of the water tongue, the pixel values ​​of the vertical drop of the water tongue, the actual projection angle of the water flow, and the head loss compensation coefficient.

2. The method of claim 1, wherein, The extraction of the peak pixel coordinates and bottom pixel coordinates of the thin-layer water flow includes: Receive the top-view projection image of the flow guide groove, extract the contour boundary lines of the inner walls on both sides, calculate the lateral pixel distance of the inner wall contour boundary lines at the water inlet and outlet ends, and generate the inlet cross-sectional width value and the outlet cross-sectional width value. The system receives a side view image of the thin layer of water flow in the guide groove, performs grayscale threshold segmentation and contour connectivity analysis, extracts the extreme points of the vertical coordinates of the top boundary of the connectivity region to generate peak pixel coordinates, and extracts the set of horizontal straight line pixels at the solid-liquid interface to generate bottom pixel coordinates.

3. The method for measuring the flow velocity of water on a laboratory slope according to claim 1, characterized in that, The generated head loss compensation coefficient includes: Extract the inlet cross-sectional width value and the outlet cross-sectional width value, perform a division operation, and extract the dimensionless quotient value as the cross-sectional shrinkage ratio value; The two-dimensional matrix index of the peak pixel coordinates and the bottom pixel coordinates is analyzed, the horizontal coordinate data is stripped and the vertical coordinate data is extracted, and a subtraction operation is performed to generate the difference in standing wave height. The difference in standing wave height is used as the base and multiplied with the cross-sectional contraction ratio to generate a head loss compensation coefficient.

4. The method for measuring the flow velocity of water on a laboratory slope according to claim 1, characterized in that, The extraction of the pixel coordinates of the detachment point and the pixel coordinates of the water contact point includes: The first and second video frames of adjacent time series are extracted using the video frames of the projected water tongue. The first video frame and the second video frame are then subjected to pixel-level grayscale subtraction to generate a binary motion difference matrix. The binarized motion difference matrix is ​​traversed to obtain continuous bright pixel clusters. The coordinates of the initial divergence endpoints of the continuous bright pixel clusters are extracted to generate the coordinates of the detachment point pixels, and the coordinates of the final truncated endpoints of the continuous bright pixel clusters are extracted to generate the coordinates of the water-receiving point pixels.

5. The method for measuring the flow velocity of water on a laboratory slope according to claim 1, characterized in that, The method further includes: The reciprocal of the head loss compensation coefficient is multiplied by the pixel value of the horizontal displacement of the water tongue to generate a reliability score for the water tongue trajectory camera observation.

6. The method for measuring the flow velocity of water on a laboratory slope according to claim 1, characterized in that, The generated horizontal displacement pixel value and vertical drop pixel value of the water tongue include: The pixel coordinates of the detachment point and the pixel coordinates of the water contact point are decomposed into a two-dimensional matrix to generate the x-coordinate component of the detachment point, the y-coordinate component of the detachment point, and the x-coordinate component and the y-coordinate component of the water contact point. The absolute value of the difference between the horizontal coordinate component of the water contact point and the horizontal coordinate component of the water separation point is calculated to generate the horizontal displacement pixel value of the water tongue. The absolute value of the difference between the vertical coordinate component of the water contact point and the vertical coordinate component of the water separation point is calculated to generate the vertical drop pixel value of the water tongue.

7. The method for measuring the flow velocity of water on a laboratory slope according to claim 1, characterized in that, The actual projectile angle of the generated water flow includes: Receive the first digital tilt angle sensor data packet from the bottom surface of the flow guide groove and the second digital tilt angle sensor data packet from the experimental slope, perform data load decoding, and extract the component placement angle value and the slope tilt angle value; The component placement angle value and the slope inclination angle value are added using a binary floating-point addition instruction to generate the actual water flow projection angle value.

8. The method for measuring the flow velocity of water on a laboratory slope according to claim 5, characterized in that, The initial flow velocity of the generated water flow includes: Substitute the pixel values ​​of the horizontal displacement of the water tongue, the pixel values ​​of the vertical drop of the water tongue, and the actual projection angle of the water flow into a preset fluid projection mapping model to perform algebraic calculations and generate uncompensated velocity variables. The uncompensated velocity variable is multiplied by the head loss compensation coefficient to generate a compensated transition velocity value. The compensated transition velocity value and the credibility score value of the water tongue trajectory camera observation are weighted and fused to generate the initial flow velocity value.

9. An apparatus for measuring the flow velocity of water on a laboratory slope, applied to a method for measuring the flow velocity of water on a laboratory slope as described in any one of claims 1-8, characterized in that, The device includes: The image parameter acquisition module is used to acquire the inlet cross-sectional width and outlet cross-sectional width values ​​of the flow guide groove, receive the thin layer of water flowing through the flow guide groove, and extract the peak pixel coordinates and bottom pixel coordinates of the thin layer of water. The loss compensation coefficient calculation module is used to perform a division operation between the inlet cross-section width value and the outlet cross-section width value to generate a cross-section contraction ratio value, perform a vertical coordinate difference calculation between the wave crest pixel coordinate and the bottom pixel coordinate to generate a standing wave bulge height difference value, and perform a multiplication operation between the standing wave bulge height difference value and the cross-section contraction ratio value to generate a head loss compensation coefficient. The video frame parsing module is used to receive video frames of the thin layer of water flowing away from the outlet of the guide groove and projecting water tongue, perform inter-frame difference operation on the projecting water tongue video frames, and extract the pixel coordinates of the detachment point and the pixel coordinates of the water contact point. The displacement pixel calculation module is used to perform horizontal and vertical difference calculations on the pixel coordinates of the detachment point and the pixel coordinates of the water contact point to generate the horizontal displacement pixel value and the vertical drop pixel value of the water tongue. The projectile angle calculation module is used to obtain the component placement angle value of the bottom surface of the guide groove and the slope inclination angle value of the experimental slope, and perform an addition operation on the component placement angle value and the slope inclination angle value to generate the actual projectile angle value of the water flow. The initial flow velocity calculation module is used to perform fluid projection algebra calculation on the pixel values ​​of the horizontal displacement of the water tongue, the pixel values ​​of the vertical drop of the water tongue, the actual projection angle of the water flow, and the head loss compensation coefficient to generate the initial flow velocity value of the water flow.

Images