Single-camera water intersection scene reconstruction and real-time flow measurement method

Abstract

The invention discloses a single-camera water intersection scene reconstruction and real-time flow measurement method, and relates to the technical field of water intersection scene reconstruction and flow measurement, and the single-camera water intersection scene reconstruction and real-time flow measurement method comprises the following operation steps: S1, data acquisition; s2, boundary identification and feature extraction; s3, feature auxiliary extraction; s4, performing three-dimensional reconstruction; and S5, real-time flow measurement is carried out. According to the invention, when the optical flow method is utilized to extract the feature points such as water ripples and ripples on the water flow surface, if the breakpoint areas of the water ripples and ripples under the conditions of illumination change and water surface reflection are detected to be larger than the upper limit of a repairable threshold, the upward flow throwing device is remotely controlled to throw the floating balls with bright colors, and the floating balls flow along with the water ripples and ripples, so that the floating balls can be quickly thrown. Therefore, after the floating ball is captured by the camera, the optical flow method extracts and tracks the floating ball as a feature point, so that the difficulty of extracting the feature point by the optical flow method is conveniently reduced, the extraction accuracy is greatly improved, and accurate real-time flow measurement is facilitated.

Description

Technical Field

[0001] The invention relates to the technical field of water intersection scene reconstruction and flow measurement, and in particular to a single-camera water intersection scene reconstruction and real-time flow measurement method. Background Art

[0002] Single-camera water intersection scene reconstruction and real-time flow measurement is a comprehensive technology involving computer vision, image processing and fluid mechanics. It is mainly used to reconstruct the dynamics of water flow and estimate the flow rate through a single camera in water intersection scenarios (such as rivers or canal flows).

[0003] In actual use, the existing single-camera water intersection scene reconstruction and real-time flow measurement technology often has the extraction of feature points (such as ripples) greatly affected by the reflection of sunlight, which makes three-dimensional reconstruction more difficult.

[0004] Therefore, in view of this, the existing structure and deficiencies are studied and improved, and a single-camera water intersection scene reconstruction and real-time flow measurement method is proposed. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides a single-camera water intersection scene reconstruction and real-time flow measurement method, which solves the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: a single-camera water intersection scene reconstruction and real-time flow measurement method, the single-camera water intersection scene reconstruction and real-time flow measurement method comprising the following operating steps: S1. Data collection: Use a single high-resolution camera to continuously shoot the water flow scene to obtain continuous video data of the water flow; S2, boundary recognition and feature extraction: Use edge detection algorithm to identify water flow boundaries, and use optical flow method to extract feature points of water flow surface from video data. The edge detection algorithm is as follows: Use histogram equalization, contrast stretching and other methods to enhance the contrast between water flow and background; Use edge detection operators to calculate the gradient of each pixel in the image. The gradient represents the rate and direction of change of the pixel value. Enhance the pixels representing the edge in the gradient image and suppress other areas; According to the preset threshold, the gradient image is binarized to extract obvious edge pixels; Use morphological operations and Hough transform methods to connect edge segments into continuous boundaries; Use morphological operations and non-maximum suppression methods to refine coarse edges into single-pixel wide boundaries; Extract shape features, color features, and texture features of the edge, and use a machine learning algorithm or a rule-based method to classify the edge as a water flow boundary or a non-water flow boundary based on the extracted features; The specific operation of the optical flow method is as follows: The feature points on the water surface are usually areas with violent movement or complex texture. The optical flow vectors in these areas have modulus or direction changes. The modulus and direction of the optical flow vector are calculated: According to the optical flow vector Calculate the magnitude and direction of the optical flow vector for each pixel: The modulus of the optical flow vector is ,in The modulus represents the speed of pixel movement. Represents the square root operator symbol; Direction ,in The direction represents the direction of pixel movement, is the square tangent function, used to Ratio calculation angle; Set a threshold based on the optical flow vector modulus or direction change, select pixels with violent motion or complex texture as feature points, track the extracted feature points, and obtain their motion trajectory in the time series; S3, feature-assisted extraction: When it is detected that the influence of illumination changes and water surface reflection factors exceeds the preset threshold, resulting in inaccurate results of the optical flow method for extracting water surface feature points, a floating ball is dropped upstream through a fixed-point drop device pre-installed upstream of the shooting area, and the floating ball is extracted and tracked as a feature point on the water surface; S4. 3D reconstruction: Use structured light or stereo vision to reconstruct the three-dimensional structure of the water flow scene using the spatial position and temporal changes of feature points, and use optical flow to calculate the water flow velocity based on the temporal displacement of feature points; S5, Real-time flow measurement: The real-time flow rate in the area can be calculated in real time by using the real-time water flow velocity calculated above and the cross-sectional area of ​​the river after three-dimensional reconstruction.

[0007] Furthermore, in step S1, the camera shooting position needs to fully cover the water flow area, and the continuous video data shot by the camera needs to be filtered to remove noise, and white balance and brightness adjustment must be performed.

[0008] Furthermore, in step S2, the feature points include but are not limited to water ripples and waves.

[0009] Furthermore, in step S2, the edge detection operator includes a Sobel operator, a Prewitt operator, and a Roberts operator.

[0010] Furthermore, in step S2, the optical flow vector is a two-dimensional vector that describes the displacement of a pixel in an image between two frames. It represents the displacement in the horizontal direction. Indicates the vertical position.

[0011] Furthermore, in step S3, the float is brightly colored and can be clearly distinguished from the surrounding environment, and is made of degradable materials.

[0012] Furthermore, in step S3, the fixed-point delivery device is controlled by remote control technology.

[0013] Furthermore, in step S5, the real-time flow calculation formula is as follows: .

[0014] Furthermore, the is the flow rate of the river, is the Manning roughness coefficient, is the cross-sectional area of ​​the river channel, is the ratio of wetted perimeter to wetted area, The slope of the water surface.

[0015] The present invention provides a single-camera water intersection scene reconstruction and real-time flow measurement method, which has the following beneficial effects: 1. This single-camera water scene reconstruction and real-time flow measurement method uses the optical flow method to extract feature points such as water ripples and ripples on the water surface. If the breakpoint area of ​​the water ripples and ripples detected under the condition of illumination change and water surface reflection is greater than the upper limit of the repairable threshold, the upstream delivery device is remotely controlled to deliver a brightly colored floating ball, and the floating ball flows with the water ripples and ripples. After the floating ball is captured by the camera, the optical flow method is used to extract and track the floating ball as a feature point, thereby reducing the difficulty of extracting feature points by the optical flow method and greatly improving the extraction accuracy, so as to accurately measure the flow in real time. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 The present invention is a schematic diagram of the overall process of a single-camera water intersection scene reconstruction and real-time flow measurement method. DETAILED DESCRIPTION

[0017] The following embodiments of the present invention are described in further detail in conjunction with the accompanying drawings and examples. The following examples are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0018] like Figure 1 As shown, the present invention provides a technical solution: a single-camera water intersection scene reconstruction and real-time flow measurement method, the single-camera water intersection scene reconstruction and real-time flow measurement method comprises the following operating steps: S1. Data collection: Use a single high-resolution camera to continuously shoot the water flow scene to obtain continuous video data of the water flow; The camera shooting position must fully cover the water flow area, and the continuous video data captured by the camera must be filtered to remove noise, and white balance and brightness adjustment must be performed; S2, boundary recognition and feature extraction: Use edge detection algorithm to identify water flow boundaries, and use optical flow method to extract feature points of water flow surface from video data. Feature points include but are not limited to water ripples and ripples. The edge detection algorithm is as follows: Use histogram equalization, contrast stretching and other methods to enhance the contrast between water flow and background; Use edge detection operators to calculate the gradient of each pixel in the image. The gradient represents the rate and direction of change of the pixel value. Among them, edge detection operators include Sobel operator, Prewitt operator, and Roberts operator; Enhance the pixels representing the edge in the gradient image and suppress other areas; According to the preset threshold, the gradient image is binarized to extract obvious edge pixels; Use morphological operations (such as dilation, erosion, closing operations) and Hough transform methods to connect edge fragments into continuous boundaries; Use morphological operations (such as skeleton extraction) and non-maximum suppression methods to refine coarse edges into single-pixel wide boundaries; Extract shape features (such as length, curvature, direction), color features, and texture features of the edge, and use machine learning algorithms (such as support vector machines, random forests) or rule-based methods to classify the edge as a flow boundary or a non-flow boundary based on the extracted features; The specific operation of the optical flow method is as follows: The feature points on the water surface are usually areas with violent movement or complex texture. The optical flow vectors in these areas have modulus or direction changes. The modulus and direction of the optical flow vector are calculated: According to the optical flow vector Calculate the magnitude and direction of the optical flow vector for each pixel: The modulus of the optical flow vector is ,in The modulus represents the speed of pixel movement. Represents the square root operator symbol; Direction ,in The direction represents the direction of pixel movement, is the square tangent function, used to Ratio calculation angle; Set a threshold based on the optical flow vector modulus or direction change, select pixels with violent motion or complex texture as feature points, track the extracted feature points, and obtain their motion trajectory in the time series; Among them, the optical flow vector is a two-dimensional vector that describes the displacement of a pixel in an image between two frames. It represents the displacement in the horizontal direction. It indicates the position in the vertical direction; S3, feature-assisted extraction: When it is detected that the influence of illumination changes and water surface reflection factors exceeds the preset threshold, resulting in inaccurate results of the optical flow method for extracting water surface feature points, a floating ball is dropped upstream through a fixed-point drop device pre-installed upstream of the shooting area, and the floating ball is extracted and tracked as a feature point on the water surface; The floating ball is brightly colored and can be clearly distinguished from the surrounding environment. It is made of degradable materials, and the fixed-point delivery device is controlled by remote control technology. S4. 3D reconstruction: Use structured light or stereo vision to reconstruct the three-dimensional structure of the water flow scene using the spatial position and temporal changes of feature points, and use optical flow to calculate the water flow velocity based on the temporal displacement of feature points; The feature point used by the buoy does not participate in the reconstruction process of the three-dimensional structure. The camera can obtain the geometric composition of the water flow area by moving and switching the perspective, and the images of the same area from different perspectives can be verified with each other to reconstruct the three-dimensional structure. S5. Real-time flow measurement: The real-time flow rate in the area is calculated in real time by using the real-time water flow velocity calculated above and the cross-sectional area of ​​the river channel after three-dimensional reconstruction; The real-time traffic calculation formula is as follows:

[0019] in, is the flow rate of the river (unit: cubic meters per second), is the Manning roughness coefficient, is the area of ​​the river cross section (unit: square meters), is the ratio of wetted perimeter to wetted area, The slope of the water surface; When the optical flow method cannot accurately extract the characteristic points (water ripples and waves) on the water surface due to changes in lighting and reflections on the water surface, the floating ball flowing with the water surface is captured as the characteristic point, and the moving speed of the floating ball is the water flow velocity.

[0020] In summary, when using the single-camera water intersection scene reconstruction and real-time flow measurement method, first use a single high-resolution camera to continuously shoot the water flow scene to obtain continuous video data of the water flow. In order to reduce the influence of lighting and water surface reflection, the high-resolution camera can use an infrared camera. The camera shooting position needs to fully cover the water flow area, and the continuous video data shot by the camera needs to be filtered to remove noise, and white balance and brightness adjustment are performed to enhance the image. Use edge detection algorithm to identify water flow boundaries, and use optical flow method to extract feature points of water flow surface from video data. Feature points include but are not limited to water ripples and ripples. The edge detection algorithm is as follows: The contrast between the water flow and the background is enhanced by using methods such as histogram equalization and contrast stretching. The gradient of each pixel in the image is calculated using an edge detection operator. The gradient represents the rate and direction of change of the pixel value. Enhance the pixels representing the edge in the gradient image and suppress other areas; binarize the gradient image according to the preset threshold and extract obvious edge pixels; use morphological operations (such as dilation, erosion, closing operation) and Hough transform method to connect edge fragments into continuous boundaries; use morphological operations (such as skeleton extraction) and non-maximum suppression method to refine the coarse edges into single-pixel wide boundaries; extract the shape features (such as length, curvature, direction), color features, and texture features of the edge, and use machine learning algorithms (such as support vector machines, random forests) or rule-based methods to classify the edges as water flow boundaries or non-water flow boundaries according to the extracted features; The specific operation of the optical flow method is as follows: The feature points on the water surface are usually areas with violent movement or complex texture. The optical flow vectors in these areas have modulus or direction changes. The modulus and direction of the optical flow vector are calculated: According to the optical flow vector Calculate the magnitude and direction of the optical flow vector for each pixel: The modulus of the optical flow vector is ,in The modulus represents the speed of pixel movement. Represents the square root operator; the direction is ,in The direction represents the direction of pixel movement, is the square tangent function, used to Ratio calculation angle; Set a threshold based on the optical flow vector modulus or direction change, select pixels with violent motion or complex texture as feature points, track the extracted feature points, and obtain their motion trajectory in the time series; When it is detected that the influence of illumination changes and water surface reflection factors exceeds the preset threshold (such as when water ripples have breakpoints and their discontinuity is greater than the upper limit that can be inferred to be filled), resulting in inaccurate results of the optical flow method for extracting water surface feature points, a floating ball is released upstream through remote control using a fixed-point release device pre-installed upstream of the shooting area, and the floating ball is extracted and tracked as a feature point on the water surface; the floating ball is brightly colored and can be clearly distinguished from the surrounding environment, and the floating ball is made of degradable materials to prevent environmental pollution; Use structured light or stereo vision to reconstruct the three-dimensional structure of the water flow scene using the spatial position and temporal changes of feature points, and use optical flow to calculate the water flow velocity based on the temporal displacement of feature points; The real-time flow rate in the area is calculated in real time by using the real-time water flow velocity calculated above and the cross-sectional area of ​​the river channel after three-dimensional reconstruction; The real-time traffic calculation formula is as follows:

[0021] in, is the flow rate of the river (unit: cubic meters per second), is the Manning roughness coefficient, is the area of ​​the river cross section (unit: square meters), is the ratio of wetted perimeter to wetted area, The slope of the water surface.

[0022] When the present invention utilizes the optical flow method to extract characteristic points on the water surface, such as water ripples and waves, if it is detected that the breakpoint area of ​​the water ripples and waves due to changes in illumination and reflection on the water surface is greater than the upper limit of the repairable threshold, a delivery device upstream is remotely controlled to deliver a brightly colored float, and the float flows along with the water ripples and waves. After the float is captured by the camera, the optical flow method is used to extract and track the float to serve as a characteristic point, thereby conveniently reducing the difficulty of extracting characteristic points using the optical flow method and greatly improving the extraction accuracy, so as to accurately measure the flow in real time.

[0023] The embodiments of the present invention are given for the purpose of illustration and description, and are not intended to be exhaustive or to limit the invention to the disclosed forms. Many modifications and variations will be apparent to those of ordinary skill in the art. The embodiments are selected and described in order to better illustrate the principles and practical applications of the present invention and to enable those of ordinary skill in the art to understand the present invention and thereby design various embodiments with various modifications suitable for specific uses.

Claims

1. A single-camera water intersection scene reconstruction and real-time flow measurement method, characterized by: The single-camera water intersection scene reconstruction and real-time flow measurement method comprises the following steps: S1. Data collection: Use a single high-resolution camera to continuously shoot the water flow scene to obtain continuous video data of the water flow; S2, boundary recognition and feature extraction: Use edge detection algorithm to identify water flow boundaries, and use optical flow method to extract feature points of water flow surface from video data. The edge detection algorithm is as follows: Use histogram equalization, contrast stretching and other methods to enhance the contrast between water flow and background; Use edge detection operators to calculate the gradient of each pixel in the image. The gradient represents the rate and direction of change of the pixel value. Enhance the pixels representing the edge in the gradient image and suppress other areas; According to the preset threshold, the gradient image is binarized to extract obvious edge pixels; Use morphological operations and Hough transform methods to connect edge segments into continuous boundaries; Use morphological operations and non-maximum suppression methods to refine coarse edges into single-pixel wide boundaries; Extract shape features, color features, and texture features of the edge, and use a machine learning algorithm or a rule-based method to classify the edge as a water flow boundary or a non-water flow boundary based on the extracted features; The specific operation of the optical flow method is as follows: The feature points on the water surface are usually areas with violent movement or complex texture. The optical flow vectors in these areas have modulus or direction changes. The modulus and direction of the optical flow vector are calculated: According to the optical flow vector Calculate the magnitude and direction of the optical flow vector for each pixel: The modulus of the optical flow vector is ,in The modulus represents the speed of pixel movement. Represents the square root operator symbol; Direction ,in The direction represents the direction of pixel movement, is the square tangent function, used to Ratio calculation angle; Set a threshold based on the optical flow vector modulus or direction change, select pixels with violent motion or complex texture as feature points, track the extracted feature points, and obtain their motion trajectory in the time series; S3, feature-assisted extraction: When it is detected that the influence of illumination changes and water surface reflection factors exceeds the preset threshold, resulting in inaccurate results of the optical flow method for extracting water surface feature points, a floating ball is dropped upstream through a fixed-point drop device pre-installed upstream of the shooting area, and the floating ball is extracted and tracked as a feature point on the water surface; S4. 3D reconstruction: Use structured light or stereo vision to reconstruct the three-dimensional structure of the water flow scene using the spatial position and temporal changes of feature points, and use optical flow to calculate the water flow velocity based on the temporal displacement of feature points; S5. Real-time flow measurement: The real-time flow rate in the area can be calculated in real time by using the real-time water flow velocity calculated above and the cross-sectional area of ​​the river after three-dimensional reconstruction.

2. A single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1, characterized in that: In step S1, the camera shooting position needs to fully cover the water flow area, and the continuous video data shot by the camera needs to be filtered to remove noise, and white balance and brightness adjustment must be performed.

3. The single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1 is characterized by: In step S2, the feature points include but are not limited to water ripples and waves.

4. The single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1 is characterized by: In step S2, the edge detection operator includes a Sobel operator, a Prewitt operator, and a Roberts operator.

5. The single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1, characterized in that: In step S2, the optical flow vector is a two-dimensional vector that describes the displacement of a pixel in an image between two frames. It represents the displacement in the horizontal direction. Indicates the vertical position.

6. The single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1, characterized in that: In step S3, the float is brightly colored and can be clearly distinguished from the surrounding environment, and the float is made of degradable materials.

7. The single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1, characterized in that: In step S3, the fixed-point delivery device is controlled by remote control technology.

8. The single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 1, characterized in that: In step S5, the real-time flow calculation formula is as follows: 。 9. A single-camera water intersection scene reconstruction and real-time flow measurement method according to claim 8, characterized in that: Said is the flow rate of the river, is the Manning roughness coefficient, is the cross-sectional area of ​​the river channel, is the ratio of wetted perimeter to wetted area, The slope of the water surface.

Images