A Calibration Method and System for Icing Sensors Based on Binocular Vision

Abstract

This invention relates to the field of aviation icing detection technology, specifically to a calibration method and system for icing sensors based on binocular vision. The binocular vision-based icing sensor calibration method includes the following steps: In an icing environment, the icing sensor continuously acquires icing information from its detection surface to obtain a real-time photoelectric response value x; in a non-icing environment, a binocular vision system continuously acquires images of the icing process on the detection surface of the icing sensor to obtain a real-time sequence of icing image pairs; based on the binocular vision principle, three-dimensional reconstruction is performed on the icing on the real-time detection surface of the icing sensor to obtain a three-dimensional image of the ice surface, and the real-time ice thickness value y is obtained; the real-time ice thickness value y is correlated with the real-time photoelectric response value x, and a functional correlation y=f(x) is obtained.

Description

Technical Field

[0001] This invention relates to the field of aviation icing detection technology, specifically to an icing sensor calibration method and system based on binocular vision. Background Technology

[0002] Icing sensors are commonly used components in aircraft to detect icing on parts such as wings, fuselages, and engine air intakes. They have functions such as icing warning and icing thickness detection. There are many types of icing sensors, among which fiber optic icing sensors are embedded inside the detection surface, without damaging the surface shape, and have advantages such as high detection accuracy, simple structure, and strong anti-interference ability.

[0003] Fiber optic icing sensors require calibration before deployment. The calibration method typically involves measuring the ice thickness using the icing sensor to obtain its photoelectric response curve. Simultaneously, other methods are used to measure the actual ice thickness. The icing sensor's response curve is then correlated with the actual ice thickness to complete the calibration. The calibration environment is usually an icing wind tunnel. During calibration, after setting the airflow and cloud field parameters, the wind tunnel is opened to induce icing on the sensor's detection surface. The icing sensor measures the response curve. Once a certain thickness is reached, the wind tunnel is closed, and the actual ice thickness is measured. The actual ice thickness is typically measured using a ruler. Because measuring the actual ice thickness with a ruler requires closing the wind tunnel, this calibration method generally only provides the final ice thickness value after the wind tunnel is closed, and cannot obtain the ice thickness values ​​during the intermediate icing process. Therefore, only the final ice thickness value is available for correlation with the icing sensor's photoelectric response curve. Since the rate of increase in ice thickness during the icing process is often variable, using only the final ice thickness value for calibration will result in low calibration accuracy. If the wind tunnel is closed multiple times during the icing process to measure the actual ice thickness, and then reopened to continue icing, although the actual ice thickness value during the intermediate icing process can be obtained, the ice patterns formed after closing and reopening the wind tunnel often differ, leading to calibration errors. Summary of the Invention

[0004] The purpose of this invention is to provide a calibration method for icing sensors based on binocular vision, so as to solve the problem of low accuracy of calibration results obtained by existing calibration methods.

[0005] To address the aforementioned problems, this invention proposes a calibration method for icing sensors based on binocular vision. The technical solution employed is as follows:

[0006] A method for calibrating an icing sensor based on binocular vision includes the following steps:

[0007] Step S1: In an icing environment, the icing sensor continuously collects icing information on its detection surface to obtain the real-time photoelectric response value x of the icing sensor; at the same time, a binocular vision system in a non-icing environment continuously collects images of the icing process on the detection surface of the icing sensor to obtain a real-time sequence of icing image pairs on the detection surface of the icing sensor.

[0008] Step S2: Combine the sequence of real-time icing images of the icing sensor detection surface and, based on the principle of binocular vision, perform three-dimensional reconstruction of the icing on the real-time icing sensor detection surface to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface.

[0009] Step S3: Based on the real-time three-dimensional image of the ice surface detected by the icing sensor, obtain the real-time ice thickness value y of the ice layer detected by the icing sensor.

[0010] Step S4: Correlate the real-time ice thickness value y of the icing sensor detection surface with the real-time photoelectric response value x of the icing sensor, and fit the function correlation y=f(x).

[0011] Further, in step S1, the step of continuously collecting icing information from the detection surface of the icing sensor in an icing environment to obtain the real-time photoelectric response value x of the icing sensor includes:

[0012] In an icy environment, ice forms on the detection surface of the icing sensor. The icing sensor continuously collects icing information from its detection surface to obtain the real-time response curve of the icing sensor to the icing information, and thus obtains the real-time photoelectric response value x of the icing sensor.

[0013] Further, in step S3, obtaining the real-time ice thickness value y of the ice layer on the ice-sensing surface based on the real-time three-dimensional image of the ice surface detected by the icing sensor includes:

[0014] Multiple sampling points are selected from the real-time three-dimensional image of the ice surface detected by the icing sensor to obtain the ice thickness values ​​of the multiple sampling points. The ice thickness values ​​of the multiple sampling points are then averaged to obtain the real-time ice thickness value y of the ice surface detected by the icing sensor.

[0015] Furthermore, the number of sampling points is greater than or equal to 30.

[0016] Further, in step S2, the step of combining the real-time icing image sequence of the icing sensor detection surface and, based on the principle of binocular vision, performing three-dimensional reconstruction of the icing on the real-time icing sensor detection surface to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface includes:

[0017] The surface feature points of the icing image sequence of the real-time icing sensor detection surface are extracted and matched to obtain the matched surface feature points. Based on the principle of binocular vision, the icing of the real-time icing sensor detection surface is reconstructed in three dimensions according to the matched surface feature points, the external parameters of the binocular camera and the internal parameters of the binocular camera, so as to obtain the three-dimensional image of the ice surface of the real-time icing sensor detection surface.

[0018] Furthermore, the step of performing three-dimensional reconstruction of the icing on the real-time icing sensor detection surface based on matched surface feature points, binocular camera external parameters, and binocular camera internal parameters includes:

[0019] A stereo matching algorithm is used to construct a local point cloud of the icing image of the real-time icing sensor detection surface. The spatial relationship between adjacent image pairs is obtained by matching the icing images of the real-time icing sensor detection surface, and a global point cloud of the real-time icing sensor detection surface is established, thereby realizing the three-dimensional reconstruction of the icing on the real-time icing sensor detection surface.

[0020] Furthermore, prior to step S1, the icing sensor is placed in the icing environment; the binocular vision system is placed outside the icing environment, i.e., in the non-icing environment.

[0021] Furthermore, the icing environment also includes an icing component to be tested, and the detection surface of the icing sensor is in the same direction and position as the icing surface of the icing component to be tested.

[0022] The present invention also provides a system for performing the above-described binocular vision-based icing sensor calibration method, comprising:

[0023] The data acquisition module is used to continuously acquire icing information of the icing sensor's detection surface in an icing environment to obtain the real-time photoelectric response value x of the icing sensor; at the same time, a binocular vision system in a non-icing environment is used to continuously acquire images of the icing process of the icing sensor's detection surface to obtain a real-time sequence of icing image pairs of the icing sensor's detection surface.

[0024] The ice surface 3D image acquisition module of the icing sensor detection surface is used to combine the real-time icing image pair sequence of the icing sensor detection surface and, based on the principle of binocular vision, to perform 3D reconstruction of the icing on the real-time icing sensor detection surface to obtain the real-time 3D image of the ice surface of the icing sensor detection surface.

[0025] The ice thickness value acquisition module of the icing sensor detection surface is used to obtain the real-time ice thickness value y of the icing sensor detection surface based on the real-time three-dimensional image of the ice surface of the icing sensor detection surface.

[0026] The fitting module is used to correlate the real-time ice thickness value y of the icing sensor detection surface with the real-time photoelectric response value x of the icing sensor, and fit the function correlation y=f(x).

[0027] Furthermore, the data acquisition module includes a photoelectric response value acquisition module for the icing sensor and an icing image pair sequence acquisition module for the detection surface of the icing sensor.

[0028] The photoelectric response value acquisition module of the icing sensor is used to continuously collect icing information on the detection surface of the icing sensor in an icing environment to obtain the real-time photoelectric response value x of the icing sensor.

[0029] The icing image pair sequence acquisition module of the icing sensor detection surface is used to continuously acquire images of the icing process of the icing sensor detection surface using a binocular vision system in a non-icing environment, so as to obtain a real-time icing image pair sequence of the icing sensor detection surface.

[0030] Beneficial Effects: This invention is an improved version. The proposed icing sensor calibration method based on binocular vision continuously acquires icing information from the sensor's detection surface in an icy environment, while simultaneously using a binocular vision system in a non-icy environment to continuously acquire images of the icing process on the sensor's detection surface. This allows for the acquisition of real-time icing thickness values ​​on the sensor's detection surface, meaning multiple true ice thickness values ​​are continuously obtained during the icing process under non-contact conditions. These values ​​are then used for calibrating the icing sensor. By fitting the relationship between the sensor's response value and the true ice thickness value, calibration accuracy is improved, solving the problem of previous calibration methods that only used the final ice thickness value as the true ice thickness. Furthermore, the proposed calibration method does not require repeatedly turning the icing environment on and off, allowing the acquisition of true ice thickness values ​​without shutting it down. This solves the calibration error problem caused by inconsistent ice patterns, further improving sensor calibration accuracy. It also saves significant time, reduces operating costs, and avoids the problems of repeatedly turning the icing environment on and off, and inconsistent ice patterns during these processes.

[0031] In step S3, obtaining the real-time ice thickness value y of the ice layer on the ice-sensing surface based on the real-time three-dimensional image of the ice surface detected by the icing sensor includes:

[0032] Multiple sampling points are selected from the real-time 3D image of the ice surface detected by the icing sensor to obtain the ice thickness values ​​of multiple sampling points. The ice thickness values ​​of multiple sampling points are then averaged to obtain the real-time ice thickness value y of the icing sensor detection surface, thereby improving the accuracy of the real-time ice thickness value y of the icing sensor detection surface.

[0033] The number of sampling points is greater than or equal to 30, which reduces the computational cost while ensuring the accuracy of the real-time ice thickness value y of the icing sensor detection surface. Attached Figure Description

[0034] Figure 1 This is a schematic flowchart of the icing sensor calibration method based on binocular vision according to the present invention. Detailed Implementation

[0035] As cited in the background art, the calibration results obtained by existing calibration methods have low accuracy. Therefore, this invention provides a calibration method for an icing sensor based on binocular vision, comprising the following steps: First, in an icing environment, the icing sensor continuously acquires icing information of its detection surface to obtain the real-time photoelectric response value x of the icing sensor, i.e., the photoelectric response value x of the icing sensor at different thicknesses; then, using a binocular vision system in a non-icing environment, it continuously acquires images of the icing process of the icing sensor's detection surface to obtain a real-time sequence of icing image pairs of the icing sensor's detection surface, i.e., the sequence of icing image pairs of the icing sensor's detection surface at different thicknesses; second, combining the real-time icing... This invention proposes a binocular vision-based icing sensor calibration method. It uses a sequence of images of the icing surface of an ice sensor and, based on binocular vision, performs 3D reconstruction of the ice surface in real-time, obtaining a 3D image of the ice surface. This image is then used to calculate the ice thickness value y. Next, based on the 3D image of the ice surface, the real-time ice thickness value y is obtained. Finally, the real-time ice thickness value y is correlated with the photoelectric response value x of the ice sensor, and a function correlation y=f(x) is fitted. This invention improves calibration accuracy by solving the problem that previous calibration methods only used the final ice thickness value as the true ice thickness. Furthermore, it eliminates the need to repeatedly turn the icing environment on and off, allowing the true ice thickness value to be obtained without turning it off, thus resolving calibration errors caused by inconsistent ice patterns and improving sensor calibration accuracy. At the same time, it saves a lot of time, reduces operating costs, and avoids the problems of repeatedly turning the icing environment on and off, as well as inconsistent icing patterns during repeated turning on and off.

[0036] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain this application, and should not be construed as limiting this application.

[0037] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0038] The following description, with reference to the accompanying drawings, illustrates an embodiment of the icing sensor calibration method and system based on binocular vision.

[0039] The following is combined with Figure 1 This application provides a detailed description of the icing sensor calibration method based on binocular vision.

[0040] Step S1: In an icing environment, the icing sensor continuously collects icing information on its detection surface to obtain the real-time photoelectric response value x of the icing sensor; simultaneously, a binocular vision system in a non-icing environment continuously collects images of the icing process on the detection surface of the icing sensor to obtain a real-time sequence of icing image pairs on the detection surface of the icing sensor.

[0041] Specifically, in an icing environment, the icing sensor continuously collects icing information from its detection surface to obtain the real-time photoelectric response value x of the icing sensor. This includes: in an icing environment, icing occurs on the detection surface of the icing sensor, the icing sensor continuously collects icing information from its detection surface, obtains the real-time response curve of the icing sensor to the icing information, and then obtains the real-time photoelectric response value x of the icing sensor.

[0042] Here, before step S1, the icing sensor is placed in the icing environment; the binocular vision system is placed outside the icing environment, i.e., in the non-icing environment. The icing component to be tested is also placed in the icing environment, and the detection surface of the icing sensor is in the same direction and position as the icing surface of the component to be tested.

[0043] This embodiment uses an application example in an open icing wind tunnel environment. An icing sensor and the corresponding icing component to be tested are placed at the exit of the icing wind tunnel. In this embodiment, the icing component to be tested is an airfoil model. The icing sensor is located inside the airfoil, with its detection surface at the leading edge of the airfoil. The detection surface of the icing sensor matches the arc of the leading edge of the airfoil, making it part of the leading edge; that is, the detection surface of the icing sensor is in the same direction and position as the leading edge of the airfoil. Thus, when the airfoil ices up, ice will also form on the detection surface of the icing sensor. A binocular vision system is set up outside the icing wind tunnel. The binocular vision system includes two cameras and a light source. The light source is a yellow light source. The two cameras are placed at a certain angle and aligned with the detection surface of the icing sensor on the leading edge of the airfoil.

[0044] During calibration, specific parameters such as wind speed, temperature, droplet size, and water flow rate are set. The icing wind tunnel is then activated to cause icing on the wing model and the icing sensor's detection surface. The icing sensor continuously records the icing information on its detection surface, obtaining its response curve and thus the real-time photoelectric response value *x*. Simultaneously, two cameras are used to capture real-time images of the icing process on the sensor's detection surface, specifically the shape of the ice surface. Here, during the operation of the icing wind tunnel, a non-contact binocular vision method is used to measure the ice thickness on the sensor's detection surface. Multiple real-time measurements of the ice thickness are achieved without shutting down the wind tunnel, obtaining multiple true ice thickness values ​​during a single wind tunnel operation for the icing sensor calibration.

[0045] It should be noted that two cameras scan the icing sensor's detection surface by horizontal translation, capturing images within a 60mm area of ​​the wing's leading edge. The scanning process lasts 1 second. Although the icing process is a non-steady dynamic process, the maximum icing velocity is approximately 0.1mm / s, and the changes in the ice surface are negligible within the scanning time. This allows for uninterrupted icing measurement throughout the entire process, obtaining a series of images. The icing sensor's detection surface includes a light-emitting end and a receiving end. The received signal is converted into a photoelectric response value, which changes with the ice thickness. For example, as the ice thickness increases, the photoelectric response values ​​at n time points are obtained, listed as [80, 136, 256, 335, 458, 537, 678, 790, 913, 1023, 1356, 1400…], containing a total of n values.

[0046] Step S2: Combine the sequence of real-time icing images of the icing sensor detection surface and, based on the principle of binocular vision, perform three-dimensional reconstruction of the icing on the real-time icing sensor detection surface to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface.

[0047] Specifically, by combining a sequence of real-time icing images of the icing sensor detection surface and based on the principle of binocular vision, a three-dimensional reconstruction of the icing on the real-time icing sensor detection surface is performed to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface. This includes: extracting and matching surface feature points from the sequence of real-time icing images of the icing sensor detection surface to obtain matched surface feature points; and based on the principle of binocular vision, using the matched surface feature points, external parameters of the binocular camera, and internal parameters of the binocular camera, a three-dimensional reconstruction of the icing on the real-time icing sensor detection surface is performed to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface.

[0048] Here, based on the matched surface feature points, the external parameters of the binocular camera, and the internal parameters of the binocular camera, the icing on the real-time icing sensor detection surface is reconstructed in three dimensions. This includes: constructing a local point cloud of the icing image of the real-time icing sensor detection surface using a stereo matching algorithm, obtaining the spatial relationship between adjacent image pairs through matching the icing image of the real-time icing sensor detection surface, establishing a global point cloud of the real-time icing sensor detection surface, and thus realizing the three-dimensional reconstruction of the icing on the real-time icing sensor detection surface.

[0049] For example, the real-time three-dimensional image of the ice surface detected by the icing sensor is determined as the three-dimensional image of the ice surface detected by the icing sensor at n time points. It should be noted that the n time points for determining the three-dimensional image of the ice surface detected by the icing sensor are the same as the n time points for determining the photoelectric response value in step S1.

[0050] Step S3: Based on the real-time three-dimensional image of the ice surface detected by the icing sensor, obtain the real-time ice thickness value y of the ice layer detected by the icing sensor.

[0051] Since the ice surface detected by the icing sensor is not flat but uneven, it is necessary to calculate the average ice thickness value of the icing sensor detection surface. Specifically, based on a real-time three-dimensional image of the ice surface detected by the icing sensor, the real-time ice thickness value y of the icing sensor detection surface is obtained, including: selecting multiple sampling points in the real-time three-dimensional image of the ice surface detected by the icing sensor, obtaining the ice thickness values ​​of multiple sampling points, and averaging the ice thickness values ​​of multiple sampling points to obtain the real-time ice thickness value y of the icing sensor detection surface. Here, the number of sampling points is greater than or equal to 30.

[0052] For example, in the three-dimensional images of the ice surface detected by the icing sensor at n time points in step S2, m sampling points (m ≥ 30) are selected within the range of the icing sensor detection surface for each time point. The ice thickness values ​​of the m sampling points are averaged to obtain the ice thickness value of the icing sensor detection surface at each time point. Using this method, the ice thickness values ​​of the icing sensor detection surface at n time points are obtained as [0.2, 0.5, 0.8, 1.1, 1.4, 1.7, 2.0, 2.3, 2.6, 2.9, 3.2, 3.5, 3.8, 4.1…], with a total of n values ​​in the list.

[0053] Step S4: Correlate the real-time ice thickness value y of the icing sensor detection surface with the real-time photoelectric response value x of the icing sensor, and fit the function correlation y=f(x).

[0054] For example, the ice thickness values ​​y of the icing sensor detection surface at n time points obtained in step S3 are correlated with the photoelectric response values ​​x of the icing sensor at n time points obtained in step S1, and the functional correlation y=f(x) is obtained by fitting.

[0055] This application also provides a system for performing the above-described binocular vision-based icing sensor calibration method, comprising:

[0056] The data acquisition module is used to continuously acquire icing information of the icing sensor's detection surface in an icing environment to obtain the real-time photoelectric response value x of the icing sensor; at the same time, a binocular vision system in a non-icing environment is used to continuously acquire images of the icing process of the icing sensor's detection surface to obtain a real-time sequence of icing image pairs of the icing sensor's detection surface.

[0057] The ice surface 3D image acquisition module of the icing sensor detection surface is used to combine the real-time icing image pair sequence of the icing sensor detection surface and, based on the principle of binocular vision, to perform 3D reconstruction of the icing on the real-time icing sensor detection surface to obtain the real-time 3D image of the ice surface of the icing sensor detection surface.

[0058] The ice thickness value acquisition module of the icing sensor detection surface is used to obtain the real-time ice thickness value y of the icing sensor detection surface based on the real-time three-dimensional image of the ice surface of the icing sensor detection surface.

[0059] The fitting module is used to correlate the real-time ice thickness value y of the icing sensor detection surface with the real-time photoelectric response value x of the icing sensor, and fit the function correlation y=f(x).

[0060] The data acquisition module includes a photoelectric response value acquisition module for the icing sensor and an icing image pair sequence acquisition module for the icing sensor detection surface. The photoelectric response value acquisition module is used to continuously collect icing information from the icing surface of the icing sensor in an icing environment to obtain the real-time photoelectric response value x of the icing sensor. The icing image pair sequence acquisition module is used to continuously collect images of the icing process of the icing sensor detection surface using a binocular vision system in a non-icing environment to obtain the real-time icing image pair sequence of the icing sensor detection surface.

[0061] Those skilled in the art will understand that the specific method of the above-described binocular vision-based icing sensor calibration system has been referenced above. Figure 1 The description of the icing sensor calibration method based on binocular vision has been detailed, so its repeated description will be omitted here.

[0062] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. The scope of patent protection of the present invention shall be determined by the claims. Similarly, any equivalent structural changes made based on the description and drawings of the present invention shall also be included within the scope of protection of the present invention.

Claims

1. A calibration method for an icing sensor based on binocular vision, characterized in that, Includes the following steps: Step S1: In an icing environment, the icing sensor continuously collects icing information on its detection surface to obtain the real-time photoelectric response value x of the icing sensor; at the same time, a binocular vision system in a non-icing environment continuously collects images of the icing process on the detection surface of the icing sensor to obtain a real-time sequence of icing image pairs on the detection surface of the icing sensor. Step S2: Combine the sequence of real-time icing images of the icing sensor detection surface and, based on the principle of binocular vision, perform three-dimensional reconstruction of the icing on the real-time icing sensor detection surface to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface. Step S3: Based on the real-time three-dimensional image of the ice surface detected by the icing sensor, obtain the real-time ice thickness value y of the ice layer detected by the icing sensor. Step S4: Correlate the real-time ice thickness value y of the icing sensor detection surface with the real-time photoelectric response value x of the icing sensor, and fit the function correlation y=f(x). In step S2, the step of combining the real-time icing image sequence of the icing sensor detection surface and, based on the principle of binocular vision, performing three-dimensional reconstruction of the icing on the real-time icing sensor detection surface to obtain a three-dimensional image of the ice surface of the real-time icing sensor detection surface includes: The surface feature points of the real-time icing sensor detection surface are extracted and matched to obtain the matched surface feature points. Based on the principle of binocular vision, the icing of the real-time icing sensor detection surface is reconstructed in three dimensions according to the matched surface feature points, the external parameters of the binocular camera and the internal parameters of the binocular camera, so as to obtain the three-dimensional image of the ice surface of the real-time icing sensor detection surface. The process of performing three-dimensional reconstruction of the icing on the real-time icing sensor detection surface based on matched surface feature points, binocular camera external parameters, and binocular camera internal parameters includes: A stereo matching algorithm is used to construct a local point cloud of the icing image of the real-time icing sensor detection surface. The spatial relationship between adjacent image pairs is obtained by matching the icing image of the real-time icing sensor detection surface, and a global point cloud of the real-time icing sensor detection surface is established, thereby realizing the three-dimensional reconstruction of the icing of the real-time icing sensor detection surface. Before step S1, the icing sensor is placed in the icing environment; the binocular vision system is placed outside the icing environment, i.e., in the non-icing environment. The icing environment also includes an icing component to be tested, and the detection surface of the icing sensor is in the same direction and position as the icing surface of the icing component to be tested.

2. The icing sensor calibration method based on binocular vision according to claim 1, characterized in that, In step S1, the step of continuously collecting icing information from the detection surface of the icing sensor in an icing environment to obtain the real-time photoelectric response value x of the icing sensor includes: In an icy environment, ice forms on the detection surface of the icing sensor. The icing sensor continuously collects icing information from its detection surface to obtain the real-time response curve of the icing sensor to the icing information, and thus obtains the real-time photoelectric response value x of the icing sensor.

3. The icing sensor calibration method based on binocular vision according to claim 1, characterized in that, In step S3, obtaining the real-time ice thickness value y of the ice layer on the ice-sensing surface based on the real-time three-dimensional image of the ice surface detected by the icing sensor includes: Multiple sampling points are selected from the real-time three-dimensional image of the ice surface detected by the icing sensor to obtain the ice thickness values ​​of the multiple sampling points. The ice thickness values ​​of the multiple sampling points are then averaged to obtain the real-time ice thickness value y of the ice surface detected by the icing sensor.

4. The icing sensor calibration method based on binocular vision according to claim 3, characterized in that, The number of sampling points is greater than or equal to 30.

5. A system for performing the icing sensor calibration method based on binocular vision as described in any one of claims 1-4, characterized in that, include: The data acquisition module is used to continuously acquire icing information of the icing sensor's detection surface in an icing environment to obtain the real-time photoelectric response value x of the icing sensor; at the same time, a binocular vision system in a non-icing environment is used to continuously acquire images of the icing process of the icing sensor's detection surface to obtain a real-time sequence of icing image pairs of the icing sensor's detection surface. The ice surface 3D image acquisition module of the icing sensor detection surface is used to combine the real-time icing image pair sequence of the icing sensor detection surface and, based on the principle of binocular vision, to perform 3D reconstruction of the icing on the real-time icing sensor detection surface to obtain the real-time 3D image of the ice surface of the icing sensor detection surface. The ice thickness value acquisition module of the icing sensor detection surface is used to obtain the real-time ice thickness value y of the icing sensor detection surface based on the real-time three-dimensional image of the ice surface of the icing sensor detection surface. The fitting module is used to correlate the real-time ice thickness value y of the icing sensor detection surface with the real-time photoelectric response value x of the icing sensor, and fit the function correlation y=f(x).

6. The icing sensor calibration system based on binocular vision method according to claim 5, characterized in that, The data acquisition module includes a photoelectric response value acquisition module for the icing sensor and an icing image pair sequence acquisition module for the detection surface of the icing sensor. The photoelectric response value acquisition module of the icing sensor is used to continuously collect icing information on the detection surface of the icing sensor in an icing environment to obtain the real-time photoelectric response value x of the icing sensor. The icing image pair sequence acquisition module of the icing sensor detection surface is used to continuously acquire images of the icing process of the icing sensor detection surface using a binocular vision system in a non-icing environment, so as to obtain a real-time icing image pair sequence of the icing sensor detection surface.

Images