阅读说明：本技术 一种基于深度学习的水位高度测量系统 (Water level height measurement system based on deep learning ) 是由 丁勇 袁羽 孙英豪 崔何亮 于 2020-12-13 设计创作，主要内容包括：本发明公开一种基于深度学习的水位高度测量系统,其特征在于,包含三个模块：1、水位尺模块,水位尺模块由浮子、刻度板、连接杆和立杆组成。刻度板安装在构筑物上,浮子固定于刻度板附近,浮子中间开一个通道与立杆串在一起,立杆通过连接杆固定在刻度板旁边；2、图像采集模块,刻度板上端口正前方设置一摄像机,用于采集水位尺的水面图像,采集到的水位尺图像通过网络传输到网络服务器；3、图像处理模块,图像处理模块预置神经网络算法,用于识别红色小球位置,再通过红色小球位置计算出水位高度。本发明可以实现全自动的城市内涝和河流,水库水位实时监测,并且水位尺的抗环境影响能力强,可以满足恶劣天气下的水位监测,消除了刻度板倒影对水位识别的不利影响。(The invention discloses a water level height measuring system based on deep learning, which is characterized by comprising three modules: 1. the water level gauge module consists of a floater, a scale plate, a connecting rod and a vertical rod. The scale plate is arranged on a structure, the floater is fixed near the scale plate, a channel is formed in the middle of the floater and is connected with the vertical rod in series, and the vertical rod is fixed beside the scale plate through a connecting rod; 2. the image acquisition module is provided with a camera right in front of an upper port of the scale plate and used for acquiring a water surface image of the water level gauge, and the acquired water level gauge image is transmitted to the network server through a network; 3. and the image processing module is used for presetting a neural network algorithm and is used for identifying the position of the red ball and then calculating the height of the water level through the position of the red ball. The invention can realize full-automatic urban waterlogging and river and reservoir water level real-time monitoring, has strong environmental impact resistance of the water level gauge, can meet the water level monitoring in severe weather, and eliminates the adverse effect of scale plate reflection on water level identification.)

1. The utility model provides a water level height measurement system based on degree of depth study which characterized in that: the device comprises 9 parts, namely a hard plastic cylinder (1), a red ball (2), a vertical rod (3), a connecting rod (4), a structure (5), a scale plate (6), a horizontal plane (7), a camera (8) and a network server (9); wherein: the hard plastic cylinder (1) and the red small ball (2) form a floater, and the red small ball (2) is tangent to the horizontal plane (7); the floater is connected with the scale plate (6) through the upright stanchion (3) and the connecting rod (4), and the connecting rod (4) is parallel to the horizontal plane (7); the scale plate (6) is arranged on the structure (5); the lower half part of the scale plate (6) is inserted below the horizontal plane (7), the upper half part of the scale plate is exposed out of the horizontal plane (7), and the top end of the scale plate is marked with an elevation by a leveling instrument; the camera (8) is opposite to the measuring surface of the scale plate (6), and the included angle between the shooting center line of the camera and the perpendicular line on the surface of the scale plate (6) is not more than +/-60 degrees; the images of the red small ball (2) and the scale plate (6) shot by the camera (8) are transmitted to a network server (9) through a network, and a convolutional neural network algorithm is arranged in the network server (9).

2. The deep learning-based water level height measuring system according to claim 1, wherein: the water gauge comprises a water gauge module, an image acquisition module and an image processing module which are sequentially connected.

3. The deep learning-based water level height measuring system according to claim 1, wherein: the water level gauge module comprises a hard plastic cylinder (1), a red small ball (2), a vertical rod (3), a pair of connecting rods (4) and a scale plate (6).

4. The deep learning-based water level height measuring system according to claim 1, wherein: the hard plastic cylinder (1) and the red small ball (2) form a floater; wherein: the red small ball (2) forms a cylindrical hole with a proper radius in the vertical direction through the center of the ball, the hard plastic cylinder (1) forms a hollowed cylindrical hole with the same radius at the center, and the upright rod (3) passes through the cylindrical holes on the hard plastic cylinder (1) and the red small ball (2) without contacting; the upper surface cylindrical hole of the hard plastic cylinder (1) is aligned with the lower surface cylindrical hole of the red small ball (2), and the hard plastic cylinder (1) is arranged on the lower surface of the red small ball (2) through structural glue to provide buoyancy for the red small ball (2).

5. The deep learning-based water level height measuring system according to claim 1, wherein: given that the volume of the hard plastic cylinder (1) is V, the density of the hard plastic cylinder (1) is rho, and the mass of the floater formed by the hard plastic cylinder (1) and the red small ball (2) is M, the red small ball (2) can be kept tangent to the surface of the water (7) only by ensuring that M = rho V.

6. The deep learning-based water level height measuring system according to claim 1, wherein: the upper part of the upright rod (3) is connected with the connecting rod (4) on the horizontal plane (7) by welding, the lower part of the upright rod (3) is connected with the connecting rod (4) below the horizontal plane (7) by welding, and the upright rod (3) and the horizontal plane (7) are kept in a vertical state.

7. The deep learning-based water level height measuring system according to claim 1, wherein: connecting rod (4) on horizontal plane (7) link together through the welding with the upper surface of scale plate (6), and in a similar way, connecting rod (4) under horizontal plane (7) link together through the welding with the lower surface of scale plate (6), and connecting rod (4) are parallel with horizontal plane (7).

8. The deep learning-based water level height measuring system according to claim 1, wherein: the scale plate (6) is connected with the structure (5) through welding, wherein the upper surface and the lower surface of the scale plate (6) are parallel to the horizontal plane (7), the scale plate (6), the connecting rod (4) and the upright rod (3) form a frame whole, and the stability of the frame whole in water flow can be ensured through a welding mode.

9. The deep learning-based water level height measuring system according to claim 1, wherein: the image acquisition module includes a camera (8), is fixed in scale plate (6) dead ahead, and its contained angle of shooing central line and scale plate (6) surface perpendicular is no longer than 60 degrees, avoids at the shooting in-process, and the water gauge image that leads to shooing because of the angle is too big produces serious distortion, influences the judged result.

10. The deep learning-based water level height measuring system according to claim 1, wherein: the image acquisition module sends and uploads the water level gauge image shot by the camera (8) to the network server (9).

11. The deep learning-based water level height measuring system according to claim 1, wherein: the image processing module comprises a network server (9) and is used for receiving and sending the image of the water level gauge, extracting the position of the red ball (2) in the image through a preset convolutional neural network algorithm, calculating the height of the water level, and storing the data after the image processing into a database of the network server (9) so as to facilitate the query of a user.

12. The water level height measuring system based on deep learning of claim 1, wherein: the method comprises the following steps:

step 1: installing a scale plate (6), a hard plastic cylinder (1) and a red ball (2) to a place needing to be monitored; the top end of the scale plate (6) is used for marking out the elevation by a level gauge;

step 2: the image acquisition module takes a picture of the scale plate (6) and the red small ball (2);

and step 3: the water gauge image obtained by photographing is sent and uploaded to a network server (9);

and 4, step 4: the image processing module receives the sent images of the scale plate (6) and the red ball (2), stores the images into a network server (9), then performs image recognition on the images of the scale plate (6) and the red ball (2), recognizes the position of the red ball (2) through a convolutional neural network, and knows that the diameter of the ball is d, the horizontal line of the waist of the red ball (2) passing through the center of a circle corresponds to the total number m of complete E on the scale plate (6), the ratio of the incomplete E to the complete E is n, and the top of the scale plate (6) is marked with an elevation h through a level; final water level reading x = h-d/2-5 (m + n) and storing the result in the network server (9);

and 5: and according to the obtained water level reading, if the water level reading exceeds a threshold value set by the network server (9), carrying out real-time monitoring alarm.

Technical Field

The invention relates to a water level height measuring system based on deep learning, and belongs to the field of water level monitoring.

Background

Currently, water resource and water safety issues have become one of the important factors affecting social, economic and ecological development, wherein water level data is a very critical hydrological data that can reflect these issues.

There are many methods for detecting water level in the water level station, the personal safety of monitoring personnel needs to be considered when the water level is manually measured, and the real-time performance of data is poor; the existing automatic monitoring aspect has a water level detector installed on site, and the water level height cannot be correctly identified due to reflection formed by the water surface under the interference of light. The other method is to install an image water level gauge and a camera aligned with the water level gauge, the image of the water level gauge is collected through the camera and is identified by a computer after being transmitted in a wireless or wired mode, and the method also has the problem that the interference of light forms a reflection on the water surface and the height of the water level cannot be identified correctly.

With the development of science and technology, it is a necessary trend that these traditional water level measuring methods are replaced by automatic intelligent water level monitoring methods. Based on the above, the water level height measuring system based on deep learning is designed and developed by comprehensively utilizing the image recognition technology, and the water level height is calculated by processing the water level gauge image through the image recognition technology. The water level gauge consists of a floater and a scale plate, and the problem of identifying the scales of the scale plate is ingeniously converted into the problem of identifying the position of a red small ball. The acquired water level gauge picture is transmitted to the network server through the network, and then the position of the small ball is identified through the convolutional neural network and the water level height is calculated, so that the remote real-time automatic monitoring of the water level can be realized, and the problem that the water level height is adversely affected by reflection of the water level gauge is solved.

Disclosure of Invention

In view of the deficiencies of the prior art as set forth in the background, a deep learning based water level height measurement system is now provided. A water gauge for measuring water level height based on deep learning, comprising: float, scale plate, connecting rod and pole setting. The floater consists of a hard plastic cylinder and a red small ball, the red small ball is fixed on the hard plastic cylinder, the floater is connected with the vertical rod in series, and the vertical rod is connected with the scale plate through a connecting rod; the front surface of the scale plate is used as a measuring surface for acquiring images by the camera.

Furthermore, the water discharging quality of the whole hard plastic cylinder of the floater is equal to the quality of the floater, so that the floater can keep the red small ball tangent to the water surface, and the water level height can be calculated by only identifying the position of the red small ball when the water level is identified. The vertical rod penetrates through the center of the floater, and the central hole of the floater is ensured to be large enough, so that the friction between the vertical rod and the floater can be ignored, and the water level measurement is more reliable; the floater is fixed beside the scale plate through a connecting rod.

A water level height measuring system based on deep learning comprises a water level gauge module, an image acquisition module and an image processing module. The water level gauge module is characterized in that the graduated plate is arranged at a place to be monitored and is vertical to the ground, the lower end of the graduated plate enters the water surface, and the top end of the graduated plate is used for marking out the elevation by using a level gauge; the image acquisition module is used for photographing the water level gauge and transmitting the photographed water level gauge to the network server; and the image processing module is used for receiving the sent water level gauge image, storing the water level gauge image into the network server, carrying out image recognition on the water level gauge image, extracting the position of a red ball in the water level gauge image through a convolutional neural network, and storing the result into the network server. Furthermore, the image acquisition module selects a camera, and the included angle between the shooting center line of the camera and the perpendicular line on the surface of the scale plate is not more than +/-60 degrees.

A water level monitoring method based on computer image recognition comprises the following technical scheme:

step 1: installing a scale plate at a place needing to be monitored; and the top end of the scale plate is used for marking out the elevation by using a level gauge.

Step 2: the image acquisition module shoots the water gauge.

And step 3: and transmitting and uploading the water level gauge image obtained by photographing to a network server through a network.

And 4, step 4: the image processing module receives the sent water level gauge image, stores the water level gauge image in a network server, then carries out image recognition on the water level gauge image, recognizes the position of a red small ball in the water level gauge through a convolutional neural network, the diameter of the known small ball is d, the horizontal line of the waist of the small ball passing through the center of a circle corresponds to the total number m of complete E on the scale plate, the ratio of the incomplete E to the complete E is n, and the top end of the scale plate is marked with an elevation h through a level gauge; the final water level reading x = h-d/2-5 (m + n) and stores the result in the database of the network server.

And 5: and carrying out real-time monitoring alarm according to the obtained water level reading.

The invention has the beneficial effects that: the invention can realize full-automatic urban waterlogging and river and real-time monitoring of reservoir water level; the water level gauge has strong environmental impact resistance, can meet the requirement of monitoring in severe weather, and can eliminate the adverse effect of scale plate reflection on water level identification.

Drawings

FIG. 1 is a schematic view of the installation of the water gauge of the present invention;

FIG. 2 is a front view of the water gauge of the present invention;

FIG. 3 is a side view of the water gauge of the present invention;

FIG. 4 is a top view of the water gauge of the present invention;

fig. 5 is a large scale view of the water gauge of the present invention.

Detailed Description

The present invention is further illustrated by the following figures and specific examples, which are to be understood as illustrative only and not as limiting the scope of the invention, which is to be given the full breadth of the appended claims and any and all equivalent modifications thereof which may occur to those skilled in the art upon reading the present specification.

The water level height measuring system based on deep learning comprises a hard plastic cylinder (1), red balls (2), upright rods (3), connecting rods (4), a structure (5), a scale plate (6), a horizontal plane (7), a camera (8) and a network server (9);

the floater consists of a hard plastic cylinder (1) and a red ball (2), and the lower end of the red ball can be ensured to be tangent to the water surface only by ensuring that M = rho V if the volume of the hard plastic cylinder (1) is V and the mass of the hard plastic cylinder and the red ball is M;

the water level gauge module comprises a hard plastic cylinder (1), a red small ball (2), a vertical rod (3), a connecting rod (4) and a graduated plate (6), wherein the graduated plate is installed in a place to be monitored and is vertical to the ground, and the top end of the graduated plate is used for marking out the elevation by using a level gauge;

the image acquisition module comprises a camera (8) for photographing the water level gauge and uploading the water level gauge image to a network server (9);

the image processing module comprises a cloud end (9) and is used for receiving the sent image of the water level gauge, storing the image into the network server (9), carrying out image recognition on the image of the water level gauge, extracting the position of the image red ball (2) through a convolutional neural network, and storing the result into a database of the network server (9).

Wherein, a water level gauge of water level height measurement system based on degree of depth study, its technical scheme as follows: the device comprises a hard plastic cylinder (1), a red ball (2), a vertical rod (3), a connecting rod (4) and a scale plate (6), wherein the scale plate (6) faces towards a camera (8) and serves as a measuring surface for acquiring images, and the other surface back to the camera (8) is called as an installation surface;

as shown in figure 1, the lower half part of the scale plate (6) is vertically inserted into water, and the upper half part is exposed out of the water surface. The top end of the water level is used for marking out the elevation. The camera (8) is over against the measuring surface of the scale plate (6), and the included angle between the shooting center line of the camera and the perpendicular line on the surface of the scale plate (6) is not more than +/-60 degrees. The specific measurement method is as follows:

step 1, installing a scale plate (6), a hard plastic cylinder (1) and a red small ball (2) to a place needing to be monitored; the top end of the scale plate (6) is used for marking out the elevation by a level gauge;

step 2, the image acquisition module takes a picture of the scale plate (6) and the red small ball (2);

step 3, the image of the water level ruler obtained by photographing is transmitted and uploaded to a network server (9) through a network;

step 4, the image processing module receives the sent water level gauge image, stores the water level gauge image in a network server (9), then performs image recognition on the water level gauge image, recognizes the position of the red ball (2) through a convolutional neural network, the diameter of the known ball is d, the horizontal line passing through the center of a circle of the waist of the red ball (2) corresponds to the total number m of the complete E on the scale plate (6), the ratio of the incomplete E to the complete E is n, and the top end of the scale plate (6) is marked by a leveling instrument to form an elevation h; the final water level reading x = h-d/2-5 (m + n) and stores the result in a database of the network server (9);

and 5, carrying out real-time monitoring alarm according to the obtained water level reading.

The invention only needs one water level gauge, can quickly lay the water level gauge to a place to be monitored, lays a camera (8) opposite to a graduated plate (6), transmits the camera to a network server (9) through a network, receives the camera by the network server (9), displays and applies the camera after image recognition, is different from the prior automatic monitoring water level gauge and recognition method, has high recognition precision requirement of a computer on the water level gauge, is difficult to keep the stability of a shot image under the influence of natural factors and in severe weather environment so as to lead the computer to be difficult to recognize, solves the problem by a water level height measuring system based on deep learning, and has no characters on the whole graduated plate (6) and only red small balls (2) beside the water level gauge. In the embodiment, the standard diameter of the red small ball (2) is 0.2m, the total number m of the complete E on the scale plate (6) is only needed to be identified, the horizontal line passing through the center of the circle of the waist of the red small ball (2) corresponds to the horizontal line, the ratio of the incomplete E to the complete E is n, the distance from the top end of the scale plate (6) to the diameter of the waist of the red small ball (2) is a, the height h printed by the top end of the scale plate (6) through the level gauge is h, and the final water level reading x = h-0.1-5(m + n).

It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential attributes thereof. The present embodiments are therefore to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference sign in a claim should not be construed as limiting the claim concerned.

Furthermore, it should be understood that although the present description refers to embodiments, not every embodiment may contain only a single embodiment, and such description is for clarity only, and those skilled in the art should integrate the description, and the embodiments may be combined as appropriate to form other embodiments understood by those skilled in the art.