CN-116778678-B - Tunnel water burst machine vision unattended monitoring and early warning method, device and system

CN116778678BCN 116778678 BCN116778678 BCN 116778678BCN-116778678-B

Abstract

The invention discloses a visual unattended monitoring and early warning method, device and system for a tunnel water inrush machine, which comprise the steps of calculating optical flow vectors of all pixel points in a face image, determining the position of the water inrush points, the number of the water inrush points and the optical flow vectors at the water inrush points according to the optical flow vectors of all the pixel points, judging whether the current water inrush points disappear according to the change of the optical flow vectors at the water inrush points, judging whether water inrush flows exist according to the optical flow vectors at the water inrush points, determining the optical flow vectors at the water inrush flows, determining the water inrush quantity according to the optical flow vectors at the water inrush flows, and judging whether early warning is carried out according to the water inrush quantity. And the movement characteristic of the water flow of the water burst through the tunnel is utilized, and the light flow method is used for monitoring the water burst range of the tunnel, so that the real-time monitoring of the water burst of the tunnel is realized.

Inventors

CHENG SHUAI
WANG JIANGHAO
AN ZHELI
HAN ZILI
Gao Chenglu
YUAN ZHENYU

Assignees

山东大学
中国国家铁路集团有限公司
中国铁道科学研究院集团有限公司

Dates

Publication Date: 20260505
Application Date: 20230629

Claims (7)

1. The machine vision unattended monitoring and early warning method for the sudden water burst of the tunnel is characterized by comprising the following steps of: Acquiring a face image to be detected, and calculating optical flow vectors of all pixel points in the face image; Determining the position and the number of the surging points and the optical flow vectors at the surging points according to the optical flow vectors of all the pixel points in the face image, determining the position of the surging points in the face image according to the comparison result of the optical flow vectors of all the pixel points in the face image and the threshold value of the surging point identification, thereby obtaining all the surging points and the number of the surging points in the face image, and determining the optical flow vectors at the surging points according to the optical flow vectors of the pixel points at the surging points; Judging whether the current surging point disappears according to the change of the optical flow vector at the surging point so as to update the number of the surging points, wherein the process of judging whether the current surging point disappears comprises judging whether the current surging point disappears according to whether the change of the optical flow vector at the surging point is larger than the threshold value of the surging point, and judging whether the current surging point disappears when the current surging point is smaller than the threshold value of the surging point; judging whether a gushing water flow exists according to the optical flow vector at the gushing point, and determining the optical flow vector at the gushing water flow; Determining the water burst quantity according to the optical flow vector at the water burst position, and judging whether to perform early warning according to the water burst quantity; The process for determining the water burst quantity comprises the steps of calculating the water burst quantity according to an optical flow vector at the water burst position, the acquired video frame number, the distance between a video acquisition device and a face and the range angle of the image of the face acquired by the video acquisition device, specifically, taking the optical flow vector as the instantaneous speed of the movement of a pixel point, multiplying the flow speed by the time, and then carrying out proportional conversion to obtain the water burst quantity, wherein the specific formula is as follows: Wherein: for the water quantity of the sudden water gushing, As the correlation coefficient(s), As an optical flow vector, For the number of frames of the video, For the distance between the video device and the face, And acquiring the range angle of the image for the video monitoring lens.
2. The method for visually unattended monitoring and early warning of a tunnel water inrush machine according to claim 1, wherein a dense optical flow method is used for solving a two-frame estimation algorithm based on spatial gradients, so as to estimate optical flow vectors of all pixel points in a face image.
3. The method for visually unattended monitoring and early warning of a tunnel water inrush machine according to claim 1, wherein the process of judging whether to perform early warning comprises sending out an early warning of water inrush change if the water quantity change at the water inrush current is greater than a water inrush current threshold, otherwise not sending out an early warning.
4. Tunnel suddenly gushes water machine vision unmanned on duty monitoring early warning device, its characterized in that includes: the data acquisition module is configured to acquire a face image to be detected and calculate optical flow vectors of all pixel points in the face image; The surging point identification module is configured to determine the position and the number of the surging points and the optical flow vectors at the surging points according to the optical flow vectors of the pixel points in the face image, determine the surging points in the face image according to the comparison result of the optical flow vectors of the pixel points in the face image and the threshold value of the surging point identification, so as to obtain the positions and the number of all the surging points in the face image, and determine the optical flow vectors at the surging points according to the optical flow vectors of the pixel points at the surging points; A surging point updating module configured to determine whether the current surging point disappears according to the change of the optical flow vector at the surging point, thereby updating the number of the surging points, wherein the process of determining whether the current surging point disappears comprises the steps of determining whether the current surging point disappears according to whether the change of the optical flow vector at the surging point is larger than the threshold value of the surging point existence, and determining whether the current surging point disappears when the current surging point is smaller than the threshold value of the surging point existence; a gushing water flow identification module configured to determine whether there is a gushing water flow from the optical flow vector at the gushing point and determine the optical flow vector at the gushing water flow; the water quantity and early warning module is configured to determine the water quantity of the sudden water according to the optical flow vector at the sudden water flow and judge whether to perform early warning according to the water quantity of the sudden water; The process for determining the water burst quantity comprises the steps of calculating the water burst quantity according to an optical flow vector at the water burst position, the acquired video frame number, the distance between a video acquisition device and a face and the range angle of the image of the face acquired by the video acquisition device, specifically, taking the optical flow vector as the instantaneous speed of the movement of a pixel point, multiplying the flow speed by the time, and then carrying out proportional conversion to obtain the water burst quantity, wherein the specific formula is as follows: Wherein: for the water quantity of the sudden water gushing, As the correlation coefficient(s), As an optical flow vector, For the number of frames of the video, For the distance between the video device and the face, And acquiring the range angle of the image for the video monitoring lens.
5. The tunnel water burst machine vision unattended monitoring and early warning system is characterized by comprising a video acquisition module and the tunnel water burst machine vision unattended monitoring and early warning device disclosed in claim 4; The video acquisition module is used for acquiring video streams of the tunnel face and the side wall and sending the video streams to the visual unattended monitoring and early warning device of the tunnel water burst machine; the tunnel water inrush machine vision unattended monitoring and early warning device is used for carrying out real-time monitoring and early warning on the tunnel water inrush condition according to the face image in the video stream.
6. The tunnel water burst machine vision unattended monitoring and early warning system is characterized in that the video acquisition module comprises a video monitoring lens, a rotating bearing and a rotating base, wherein the video monitoring lens is fixed on the rotating bearing to enable the video acquisition module to rotate at 110 degrees in a vertical tangent plane, and the rotating bearing is connected with the rotating base to enable the video acquisition module to rotate at 360 degrees in a horizontal tangent plane.
7. The visual unattended monitoring and early warning system for the tunnel water burst machine is characterized in that the video acquisition module is arranged on a tunnel surrounding rock through a bracket, and the height of the bracket is adjustable; And the video acquisition module is provided with a light supplementing lamp for improving illumination intensity.

Description

Tunnel water burst machine vision unattended monitoring and early warning method, device and system Technical Field The invention relates to the technical field of monitoring, prevention and control of tunnel water inrush disasters, in particular to a visual unattended monitoring and early warning method, device and system for a tunnel water inrush machine. Background The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art. In the tunnel construction process, disasters such as tunnel gushing water and the like are often caused by crossing environments such as mountain canyon areas, strong water-rich areas, fault fracture zones, karst cave and the like, so that monitoring and early warning are indispensable steps in tunnel gushing water disaster treatment. The monitoring and early warning of the water burst of the tunnel is a key and precondition for realizing the active prevention and control of disasters. The existing tunnel water inrush disaster monitoring and detecting method is mainly characterized in that sensors are manually arranged for monitoring, and the detection mode often has some defects, such as water inrush possibly generated at any time in a tunnel, damage to monitoring personnel, large manual monitoring error in a dim high-risk environment, and easy unsatisfied actual requirements on accuracy. In order to overcome the problems, a machine vision recognition technology is gradually introduced into tunnel water burst monitoring, but at present, machine vision recognition can only acquire a fuzzy video of the interior of a tunnel, and due to the lack of a related algorithm, workers cannot accurately find water burst points from the video. Disclosure of Invention In order to solve the problems, the invention provides a visual unattended monitoring and early warning method, device and system for a tunnel water inrush machine, which are used for monitoring a tunnel water inrush range by utilizing a light flow method through the movement characteristics of water inrush flow, judging the position, the quantity and the water inrush quantity of the water inrush by calculating the light flow vector moving in an image, and realizing the real-time monitoring of the tunnel water inrush by judging whether early warning prompt in the tunnel is carried out or not. In order to achieve the above purpose, the present invention adopts the following technical scheme: in a first aspect, the invention provides a machine vision unattended monitoring and early warning method for tunnel water burst, which comprises the following steps: Acquiring a face image to be detected, and calculating optical flow vectors of all pixel points in the face image; Determining the position of the surging points, the number of the surging points and the optical flow vectors at the surging points according to the optical flow vectors of all the pixel points in the face image; judging whether the current surge points disappear according to the change of the optical flow vectors at the surge points, thereby updating the number of the surge points; judging whether a gushing water flow exists according to the optical flow vector at the gushing point, and determining the optical flow vector at the gushing water flow; and determining the water burst quantity according to the optical flow vector at the water burst position, and judging whether to perform early warning according to the water burst quantity. Alternatively, a solution of a two-frame estimation algorithm based on spatial gradients is performed using a dense optical flow method, so as to estimate the optical flow vector of each pixel point in the face image. As an alternative embodiment, the surge point positions in the face image are determined according to the comparison result of the optical flow vectors of the pixel points in the face image and the surge point recognition threshold, so as to obtain all the surge point positions and the number in the face image, and the optical flow vector at the surge point is determined according to the optical flow vectors of the pixel points at the surge point. Alternatively, determining whether the current surge point is vanishing includes determining whether the current surge point is vanishing based on whether the optical flow vector change at the surge point is greater than a surge point presence threshold, and if the current surge point is less than the latter, the surge point is vanishing. In an alternative embodiment, the process of determining the water burst quantity comprises the step of calculating the water burst quantity according to an optical flow vector at the water burst position, the acquired video frame number, the distance between the video acquisition equipment and the face and the range angle of the face image acquired by the video acquisition equipment. As an alternative implementation mode, the process of judging whether to per