Weak pressure measurement method based on stable buoyancy system and image centroid algorithm

Abstract

The application discloses a weak pressure measurement method based on a stable buoyancy system and an image centroid algorithm, and belongs to the technical field of precise mechanics measurement. The method first constructs a self-stable buoyancy rotation platform supported by high-viscosity silicone oil and an optical lever amplification light path, and eliminates mechanical friction by means of the difference between the gravity moment and the buoyancy moment; a weak pressure to be measured is applied to make the platform produce a slight rotation angle, the rotation angle is amplified into linear displacement of a reflected light spot through the optical lever principle, and a video is collected; red channel extraction and multi-frame superposition and averaging are carried out on the video frames to suppress liquid surface fluctuation noise; a threshold segmentation and a morphological operation are used to lock a light spot connected domain, a weighted pixel centroid algorithm is used to calculate the light spot center coordinates; and the light spot displacement is converted into the pressure value to be measured according to a calibrated linear response model. The application realizes high-precision measurement of weak pressure in the order of micro-newton, and the measurement resolution is better than 1 muN.

Description

Technical Field

[0001] This invention belongs to the field of precision mechanical measurement technology, specifically relating to a weak pressure measurement method based on a stable buoyancy system and an image centroid algorithm. Background Technology

[0002] In the field of precision mechanical measurement at the micro-Newton level, traditional mechanical sensors are limited by the static friction of physical structures and the nonlinear deformation of mechanical supports, resulting in a significant dead zone when measuring extremely weak forces. In recent years, non-contact measurement schemes based on the optical lever principle have become a research hotspot for weak pressure measurement technology due to their high sensitivity. By combining the optical lever principle with camera measurement, the small deformation is converted into pixel changes, and the measurement results are obtained using a sub-pixel spot extraction algorithm. However, the reflector and the object being measured are mechanically fixed, and the influence of support friction on the lower limit of measurement is not eliminated. To address the defects caused by the mechanical fixed connection, existing technologies use a latex spherical membrane as the sensitive element and a position detection sensor to receive the spot through laser reflection; and use an air buoyancy platform for micro-force calibration, using fluid buoyancy to eliminate mechanical contact. However, the above two methods have the following defects: (1) Using a latex spherical membrane as the sensitive element, since the latex spherical membrane is an elastic element, the material hysteresis effect limits the further improvement of measurement linearity; (2) Using an air buoyancy platform for micro-force calibration does not combine optical lever amplification with image centroid positioning. Furthermore, in measurement schemes combining fluid dynamics and optics, minute fluctuations in the liquid surface can cause high-frequency random drift of the reflected light spot. Existing technologies lack mechanisms to compensate for these fluid fluctuations in the time domain. Simultaneously, the light spot displacement and the pressure being measured are influenced by various factors, and there is a lack of a mathematical model to linearly invert changes in the image's spatial coordinates into physical force values.

[0003] Therefore, providing a weak pressure measurement method that can combine the self-stabilizing physical properties of fluids and utilize image processing algorithms to compensate for dynamic fluctuations in both the time and spatial domains has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0004] This invention provides a weak pressure measurement method based on a stable buoyancy system and an image centroid algorithm, characterized by the following steps:

[0005] Step S100: Construct a self-stabilizing buoyancy corner platform and optical lever amplification optical path. Place the buoyancy platform in high viscosity silicone oil, adjust the counterweight so that the center of gravity of the platform system is located below the center of buoyancy to form a self-stabilizing structure, and turn on the laser light source so that the beam is reflected by the coated plane mirror and projected onto the industrial camera photosensitive chip.

[0006] Step S200: Apply the weak pressure to be tested to the load-bearing plate to make the buoyancy angle platform generate a small angle. The angle is amplified into the linear displacement of the reflected light spot by the optical lever principle. The displacement video of the reflected light spot is collected in real time by an industrial camera to obtain the original video sequence.

[0007] Step S300: Perform channel separation on each frame of the RGB image of the original video sequence, extract the grayscale image of the red channel, and perform pixel-level superposition and averaging on the grayscale images of the red channel of N consecutive frames to obtain a denoised grayscale image, where N is 10 to 20 frames.

[0008] Step S400: Perform grayscale threshold segmentation on the denoised grayscale image to obtain an initial binary image. Perform morphological opening and closing operations on the initial binary image in sequence to eliminate isolated noise points and fill the holes inside the light spot to obtain the connected domain of the light spot. Calculate the center coordinates of the light spot using the weighted pixel centroid algorithm to achieve sub-pixel level positioning accuracy.

[0009] Step S500: Substitute the displacement of the center coordinates of the light spot into the linear response model obtained by pre-calibration using standard weights to calculate the pressure value to be measured, thus completing the weak pressure measurement.

[0010] Further, step S100 includes:

[0011] Step S110: Place the liquid container on a horizontal support base and inject high-viscosity silicone oil into the container. The kinematic viscosity of the silicone oil is 500 cSt to 1000 cSt.

[0012] Step S120: Place the buoyancy-rotating platform in silicone oil. The upper part of the platform is equipped with a load-bearing plate and a coated plane mirror. The lower part of the platform is connected to a counterweight through a connecting rod. Adjust the mass and position of the counterweight so that the overall center of gravity of the platform system is located vertically below its geometric center of buoyancy, forming a self-stabilizing structure.

[0013] Step S130: Turn on the helium-neon laser source, adjust the position and angle of the source so that the beam is reflected by the coated plane mirror and projected onto the photosensitive chip of the industrial camera to form a clear spot image, thus completing the construction of the optical lever amplification optical path.

[0014] Further, step S300 includes:

[0015] Step S310: Perform channel separation on each frame of the RGB image of the original video sequence, extract the grayscale image of the red channel, and filter out ambient background light interference by utilizing the high matching between the laser wavelength and the photosensitive range of the red channel.

[0016] Step S320: Perform pixel-level superposition averaging on N consecutive frames of red channel grayscale images, with N ranging from 10 to 20 frames, to smooth the high-frequency random jitter of the light spot spatial position caused by the slight fluctuation of the liquid surface, and obtain a denoised grayscale image.

[0017] Further, step S400 includes:

[0018] Step S410: Obtain the grayscale value of each pixel within the connected region of the light spot. and pixel coordinates .

[0019] Step S420: Calculate the centroid coordinates of the light spot using the following formula. :

[0020] ;

[0021] in For pixels grayscale value at that location These are pixel coordinates.

[0022] Furthermore, in step S420, pixels in high grayscale areas are given greater weight through grayscale weighting, achieving a positioning accuracy on the order of 0.1 pixels.

[0023] Further, in step S510, the system calibration coefficient C is obtained through pre-calibration using standard weights, and the linear response model of the system is: ,in k is the system's restoring torque coefficient, obtained through calibration with standard weights; L is the reflected light path length of the optical lever; and l is the horizontal distance from the pressure application point to the platform's rotation center. denoted as the linear displacement of the light spot on the image sensor, and F is the pressure value to be measured.

[0024] Step S520: Convert the change in the center coordinates of the light spot calculated in step S400 into a linear displacement. Substitute The pressure value F acting on the load-bearing plate 22 is calculated.

[0025] Further, in step S510, the calibration process is as follows: standard weights of known mass are placed sequentially on the load-bearing plate, and the spot displacement corresponding to each weight is recorded. The system calibration coefficient C is determined by linear regression.

[0026] Furthermore, in step S200, the industrial camera acquires the displacement video of the reflected light spot in real time at a frame rate of not less than 30fps.

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] (1) By constructing a fluid self-stabilizing buoyancy angle platform, the measuring device is floated in high viscosity silicone oil, and the counterweight is adjusted so that the center of gravity of the system is located below the center of buoyancy. The difference between the gravitational torque and the buoyancy torque is used as the restoring torque, which completely eliminates the dead zone phenomenon caused by the static friction of the physical structure and the nonlinear deformation of the mechanical support in traditional mechanical sensors, and realizes frictionless sensing of micro-Newton level weak pressure.

[0029] (2) By employing an image preprocessing strategy that combines red channel extraction with multi-frame superposition averaging, effective compensation is provided for the spot displacement drift caused by liquid surface fluctuations in the temporal dimension. Red channel extraction utilizes the high matching between the wavelength of the helium-neon laser and the photosensitive range of the red channel of the image sensor to filter out ambient background light interference at the source of the algorithm. Multi-frame superposition averaging reduces the standard deviation of background noise by more than 40% through pixel-level temporal smoothing, overcoming the deficiency of existing optical measurement schemes that lack a temporal compensation mechanism for fluid fluctuations.

[0030] (3) By using the weighted pixel centroid algorithm to perform sub-pixel-level positioning of the connected domain of the spot, and using the pixel gray value as the weight to calculate the center coordinate of the spot, a positioning accuracy of 0.1 pixels was achieved. Without changing the hardware configuration of the industrial camera, the equivalent magnification of the system was increased several times, breaking through the resolution limit of hardware imaging.

[0031] (4) A dual-layer noise suppression mechanism was constructed through the synergistic effect of physical layer damping characteristics and algorithm layer temporal denoising. High-viscosity silicone oil suppresses the high-frequency vibration amplitude of the liquid surface at the physical level, while multi-frame superposition averaging smooths the residual low-frequency fluctuations at the algorithm level. The coupling of the two effectively improves the stability and signal-to-noise ratio of the measurement system.

[0032] (5) By calibrating with standard weights, a linear response model is established, simplifying the relationship between the light spot displacement and the pressure to be measured into a linear relationship. The linearity of the measurement reaches 0.992, which simplifies the operation logic of precision measurement and realizes the automatic inversion of image spatial coordinate changes to physical force values.

[0033] The present invention will now be further described with reference to the accompanying drawings. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the overall flow of the measurement method in an embodiment of the present invention.

[0035] Figure 2 This is a schematic diagram of a physical measurement system with self-stabilizing characteristics in an embodiment of the present invention.

[0036] Figure 3 This is a comparison chart of the effects of each stage of the image processing algorithm in the embodiments of the present invention.

[0037] In the diagram: silicone oil 1, counterweight 21, load-bearing plate 22, coated plane mirror 23. Detailed Implementation

[0038] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0039] This invention provides a weak pressure measurement method based on a stable buoyancy system and an image centroid algorithm. The method eliminates mechanical friction by constructing a fluid self-stabilizing buoyancy platform, uses the optical lever amplification principle to convert minute pressure into observable light spot displacement, combines multi-frame superposition denoising and weighted pixel centroid algorithm to achieve sub-pixel level displacement capture, and finally converts the light spot displacement into the pressure value to be measured through a calibrated linear response model.

[0040] First, construct a physical measurement system with self-stabilizing properties. (Refer to...) Figure 2 A liquid container is placed on a horizontal support base and filled with silicone oil 1 with a kinematic viscosity of 500 cSt to 1000 cSt. A buoyancy-driven rotating platform is placed in the silicone oil 1. The upper part of the platform has a load-bearing plate 22 and a coated plane mirror 23, while the lower part is connected to a counterweight 21 via a connecting rod. The mass and position of the counterweight 21 are adjusted so that the overall center of gravity of the platform system is vertically below its geometric center of buoyancy. This configuration ensures that when the load-bearing plate 22 is subjected to a small pressure F and deflects, the system can automatically restore balance based on the difference between the gravitational torque and the buoyancy torque, without the need for external power, forming a self-stabilizing structure. A helium-neon laser light source is turned on, and its position and angle are adjusted so that the beam is reflected by the coated plane mirror 23 and projected onto the photosensitive chip of an industrial camera, forming a clear spot image, thus completing the construction of the optical lever amplification path. The amplification factor of the optical lever is proportional to the length L of the reflected light path.

[0041] Based on the aforementioned physical measurement system, referring to Figure 1 This measurement method includes the following steps performed in sequence.

[0042] Step S100: Acquire video of the displacement of the reflected light spot under the pressure to be tested.

[0043] Reference Figure 1 Step S100 includes S110 and S120.

[0044] In step S110, a weak pressure F is applied to the load-bearing plate 22, causing a slight rotation of the buoyancy-driven rotating platform. As the platform rotates, the coated plane mirror 23 rotates synchronously with the platform, changing the direction of the reflected light path and causing the position of the reflected light spot on the industrial camera's image sensor to move accordingly. Through the principle of optical levers, the slight rotation of the platform is magnified into a linear displacement of the reflected light spot.

[0045] Step S120: Turn on the industrial camera and acquire real-time video of the displacement of the reflected light spot at a frame rate of not less than 30fps to obtain the original video sequence under the pressure to be tested. The original video sequence contains all positional changes of the light spot during the stress process.

[0046] Step S200: Perform image preprocessing on the original video sequence, separate and extract the red channel image from the color video frames, and suppress random noise introduced by liquid surface fluctuations by multi-frame superposition averaging.

[0047] Reference Figure 1 Step S200 includes S210 and S220.

[0048] Step S210: Channel separation is performed on each frame of the RGB image in the original video sequence to extract the grayscale image of the red channel. Since the laser source uses a helium-neon laser with a wavelength of approximately 632.8nm, which is highly matched with the light-sensing range of the red channel of the image sensor, the extraction of the red channel can effectively filter out ambient background light interference, including incandescent light interference in the laboratory and natural astigmatism, and enhance the contrast between the target spot and the background.

[0049] Step S220 involves pixel-level overlay averaging of N consecutive frames of the red channel grayscale image, where N is between 10 and 20 frames. The minute fluctuations on the liquid surface mainly originate from vibrations in the laboratory foundation and airflow disturbances. These fluctuations cause high-frequency, random displacement drift of the reflected light spot on the image sensor, directly superimposed on the effective measurement signal. Through multi-frame overlay averaging, the high-frequency random jitter of the light spot's spatial position caused by the weak fluctuations on the liquid surface is significantly smoothed, reducing the background noise standard deviation by more than 40%, resulting in a denoised grayscale image.

[0050] Reference Figure 3 (a) is a grayscale image after multi-frame overlay processing. The reflected light spots are clearly visible in the background, but some low-brightness speckle noise and ambient background light still remain in the image.

[0051] Step S300: Perform binarization and morphological operations on the denoised grayscale image to lock the connected components of the light spot.

[0052] Reference Figure 1 Step S300 includes S310 and S320.

[0053] Step S310: Set a grayscale threshold to perform threshold segmentation on the denoised grayscale image. Pixels with grayscale values ​​higher than the threshold are classified as light spot regions, and pixels with grayscale values ​​lower than the threshold are classified as background, thus obtaining an initial binary image. The grayscale threshold is set based on the actual contrast between the light spots and the background.

[0054] Step S320: Morphological opening and closing operations are performed sequentially on the initial binary image using a 3×3 operator. The opening operation involves erosion followed by dilation to eliminate isolated noise points and edge speckle around the spot. The closing operation involves dilation followed by erosion to fill the holes inside the spot caused by local occlusion or uneven reflection, ultimately obtaining a connected region of the spot with a clear outline and continuous internal structure.

[0055] Reference Figure 3 (b) is a binary image after thresholding and morphological processing. The light spot appears as a connected region with a clear outline, providing pure target features for subsequent center localization.

[0056] Step S400: Calculate the center coordinates of the light spot using the weighted pixel centroid algorithm to achieve sub-pixel level positioning accuracy.

[0057] Reference Figure 1 Step S400 includes S410 and S420.

[0058] Step S410: Set a grayscale threshold to perform threshold segmentation on the denoised grayscale image to obtain an initial binary image;

[0059] Step S420: Perform morphological opening and closing operations sequentially on the initial binary image, using a 3×3 operator to eliminate isolated noise points and edge speckle around the light spot, fill in the internal holes of the light spot, and obtain a clearly defined connected region of the light spot; obtain the gray value of each pixel within the connected region of the light spot. and pixel coordinates Calculate the centroid coordinates of the light spot using the following formula.

[0060]

[0061] in For pixels grayscale value at that location These are pixel coordinates. By using grayscale weighting, the algorithm assigns greater weight to pixels in high grayscale areas, achieving sub-pixel-level precise analysis of the center position of the light spot. The positioning accuracy reaches the level of 0.1 pixels, which significantly improves the equivalent magnification of the system without changing the hardware configuration of the industrial camera.

[0062] Reference Figure 3 (c) The position marked by the crosshair is the sub-pixel center coordinate calculated by the weighted pixel centroid algorithm. Comparing the above stages, it can be seen that this method can accurately and stably lock the spot core from a complex background with dynamic fluctuations, ensuring extremely high accuracy in displacement capture.

[0063] Step S500: Convert the displacement of the center coordinates of the light spot into the pressure value to be measured according to the calibrated linear response model.

[0064] Reference Figure 1 Step S500 includes S510 and S520.

[0065] Step S510: Obtain the system calibration coefficients through pre-calibration using standard weights. The linear response model of the system is as follows: ,in k is the system's restoring torque coefficient, obtained through calibration with standard weights; L is the reflected light path length of the optical lever; and l is the horizontal distance from the pressure application point to the platform's rotation center. Let F be the linear displacement of the light spot on the image sensor, and F be the pressure value to be measured. The calibration process is as follows: standard weights of known mass are placed sequentially on the load receiving plate 22, and the light spot displacement corresponding to each weight is recorded. The system calibration coefficient C is determined by linear regression.

[0066] Step S520: Convert the change in the center coordinates of the light spot calculated in step S400 into a linear displacement. Substitute The pressure value F acting on the load-bearing plate 22 is calculated, and the weak pressure measurement is completed.

[0067] The physical system described above, along with steps S100 to S500, constitutes a complete closed loop for weak pressure measurement. The physical system eliminates mechanical friction and provides stable support through a fluid buoyancy platform. S100 amplifies the pressure to be measured using an optical lever, converting it into an observable video of the light spot displacement. S200 and S300 suppress fluid fluctuation noise in the temporal domain through channel-specific extraction and multi-frame stacking denoising. S400 achieves sub-pixel-level displacement capture in the spatial domain using a weighted pixel centroid algorithm. S500 inverts the image spatial coordinate changes into physical force values ​​using a calibrated linear response model. The damping characteristics of the physical layer and the temporal denoising of the algorithm layer work synergistically—high-viscosity silicone oil suppresses high-frequency vibration amplitude, and multi-frame stacking averaging smooths residual low-frequency fluctuations; the coupling of these two elements achieves dual-layer suppression of fluid fluctuation noise.

[0068] The method will be described below with reference to specific embodiments.

[0069] Example 1

[0070] In the physical measurement system constructed in this embodiment, the liquid container is filled with silicone oil with a kinematic viscosity of 1000 cSt. The buoyancy-driven rotating platform adjusts the counterweights to position the overall center of gravity 10 mm below the center of buoyancy. The laser source is a helium-neon laser with a reflected light path length L of 1.5 m. The industrial camera uses a CMOS sensor with a resolution of 1920×1080, a frame rate of 30 fps, and a pixel size of 5.5 μm.

[0071] During the calibration process, standard weights ranging from 10mg to 500mg were used for progressive loading, and the corresponding spot displacement was recorded for each weight. Linear regression results show that the system response linearity reaches 0.992 within the measurement range of 0.1 μN to 100 μN, and the measurement resolution is better than 1 μN.

[0072] During the actual measurement, a weak pressure F was applied to the load-bearing plate 22, and an industrial camera captured the displacement video of the reflected light spot at a frame rate of 30fps. The algorithm continuously extracted 10 frames of red channel grayscale images and averaged them, reducing the standard deviation of background noise by 42%. After thresholding and morphological opening and closing operations by a 3×3 operator, the grayscale images after denoising showed clear outlines and internal continuity of the light spot's connected domains. The center coordinates of the light spot, located by the weighted pixel centroid algorithm, remained stable at the sub-pixel level. The displacement of the light spot was then measured. Substitution The pressure value to be measured is obtained, and the measurement is completed.

Claims

1. A weak pressure measurement method based on a stable buoyancy system and an image centroid algorithm, characterized in that, Includes the following steps: Step S100: Construct a self-stabilizing buoyancy corner platform and optical lever amplification optical path. Place the buoyancy platform in high viscosity silicone oil, adjust the counterweight so that the center of gravity of the platform system is located below the center of buoyancy to form a self-stabilizing structure, and turn on the laser light source so that the beam is reflected by the coated plane mirror and projected onto the industrial camera photosensitive chip. Step S200: Apply the weak pressure to be tested to the load-bearing plate to make the buoyancy angle platform generate a small angle. The angle is amplified into the linear displacement of the reflected light spot by the optical lever principle. The displacement video of the reflected light spot is collected in real time by an industrial camera to obtain the original video sequence. Step S300: Perform channel separation on each frame of the RGB image of the original video sequence, extract the grayscale image of the red channel, and perform pixel-level superposition and averaging on the grayscale images of the red channel of N consecutive frames to obtain a denoised grayscale image, where N is 10 to 20 frames. Step S400: Perform grayscale threshold segmentation on the denoised grayscale image to obtain an initial binary image. Perform morphological opening and closing operations on the initial binary image in sequence to eliminate isolated noise points and fill the holes inside the light spot to obtain the connected domain of the light spot. Calculate the center coordinates of the light spot using the weighted pixel centroid algorithm to achieve sub-pixel level positioning accuracy. Step S500: Substitute the displacement of the center coordinates of the light spot into the linear response model obtained by pre-calibration using standard weights to calculate the pressure value to be measured, thus completing the weak pressure measurement.

2. The method according to claim 1, characterized in that, Step S100 includes: Step S110: Place the liquid container on a horizontal support base and inject high-viscosity silicone oil into the container. The kinematic viscosity of the silicone oil is 500 cSt to 1000 cSt. Step S120: Place the buoyancy-rotating platform in silicone oil. The upper part of the platform is equipped with a load-bearing plate and a coated plane mirror. The lower part of the platform is connected to a counterweight through a connecting rod. Adjust the mass and position of the counterweight so that the overall center of gravity of the platform system is located vertically below its geometric center of buoyancy, forming a self-stabilizing structure. Step S130: Turn on the helium-neon laser source, adjust the position and angle of the source so that the beam is reflected by the coated plane mirror and projected onto the photosensitive chip of the industrial camera to form a clear spot image, thus completing the construction of the optical lever amplification optical path.

3. The method according to claim 1, characterized in that, Step S300 includes: Step S310: Perform channel separation on each frame of the RGB image of the original video sequence, extract the grayscale image of the red channel, and filter out ambient background light interference by utilizing the high matching between the laser wavelength and the photosensitive range of the red channel. Step S320: Perform pixel-level superposition averaging on N consecutive frames of red channel grayscale images, with N ranging from 10 to 20 frames, to smooth the high-frequency random jitter of the light spot spatial position caused by the slight fluctuation of the liquid surface, and obtain a denoised grayscale image.

4. The method according to claim 1, characterized in that, Step S400 includes: Step S410: Obtain the grayscale value of each pixel within the connected region of the light spot. and pixel coordinates ; Step S420: Calculate the centroid coordinates of the light spot using the following formula. : ； in For pixels grayscale value at that location These are pixel coordinates.

5. The method according to claim 4, characterized in that, In step S420, pixels in high grayscale areas are given greater weight through grayscale weighting, achieving a positioning accuracy on the order of 0.1 pixels.

6. The method according to claim 1, characterized in that, Step S500 includes: Step S510: Obtain the system calibration coefficient C through pre-calibration using standard weights. The linear response model of the system is: ,in k is the system's restoring torque coefficient, obtained through calibration with standard weights; L is the reflected light path length of the optical lever; and l is the horizontal distance from the pressure application point to the platform's rotation center. denoted as the linear displacement of the light spot on the image sensor, and F is the pressure value to be measured. Step S520: Convert the change in the center coordinates of the light spot calculated in step S400 into a linear displacement. Substitute The pressure value F acting on the load-bearing plate 22 is calculated.

7. The method according to claim 6, characterized in that, In step S510, the calibration process is as follows: using standard weights of known mass, place them sequentially on the load-bearing plate, and record the spot displacement corresponding to each weight. The system calibration coefficient C is determined by linear regression.

8. The method according to claim 1, characterized in that, In step S200, the industrial camera captures the displacement video of the reflected light spot in real time at a frame rate of not less than 30fps.

Images