CN-116665106-B - Urban expressway sound barrier hidden danger analysis method based on video analysis

Abstract

The invention belongs to the field of video identification, and discloses a hidden danger analysis method for an urban expressway sound barrier based on video analysis, which comprises the following steps of S1, shooting the sound barrier at the roadside of an urban expressway to obtain a patrol video containing the sound barrier; S2, acquiring video frames for image recognition from the inspection video, and storing the acquired video frames in a set to be recognized, S3, respectively inputting each video frame in the set to be recognized into a pre-trained neural network model for recognition, and acquiring a recognition result of the video frames, S4, and storing the recognition result in a result set. The invention can greatly reduce the probability of missed detection, thereby timely and comprehensively finding the potential safety hazard of the sound barrier when the potential safety hazard exists in the sound barrier.

Inventors

• REN HONGMEI
• HU XIAO
• ZHANG SHUJUAN
• MAO WEISHENG
• YUE WENWEN
• WANG YUN
• ZHANG DEJIA

Assignees

• 上海市政养护管理有限公司
• 上海城建城市运营(集团)有限公司
• 上海城建智慧城市运营管理有限公司

Dates

Publication Date

20260512

Application Date

20230602

Claims (6)

1. 1. The urban expressway sound barrier hidden danger analysis method based on video analysis is characterized by comprising the following steps of: s1, shooting a sound barrier at the roadside of an urban expressway to obtain a patrol video containing the sound barrier; S2, acquiring a video frame for image recognition from the inspection video, and storing the acquired video frame into a set to be recognized; S3, respectively inputting each video frame in the set to be identified into a pre-trained neural network model for identification, and obtaining an identification result of the video frame; S4, storing the identification result into a result set; Shooting a sound barrier at the roadside of an urban expressway to obtain a patrol video containing the sound barrier, wherein the patrol video comprises: Shooting a sound barrier at the roadside of an urban expressway through a vehicle provided with a camera to obtain a patrol video containing the sound barrier, and recording the speed information of the vehicle by adopting a fixed acquisition period in the shooting process to obtain a set of speed information; The speed information includes a speed and a recording time; before obtaining the video frame for image recognition from the inspection video, the method further comprises: continuously numbering video frames in the inspection video according to shooting time, wherein the number is larger when the shooting time is later; the method for acquiring the video frame for image recognition from the patrol video comprises the following steps: The method comprises the steps of obtaining a first frame of video frame in a patrol video for the first time, wherein the first frame of video frame is used for carrying out image recognition; The process of the nth acquisition is as follows, n is more than or equal to 2: the video frames for image recognition, which are obtained in the n-1 th time, are rapidly recognized to obtain the distance between the acquisition camera and the sound barrier ; Acquiring the number of the video frame obtained for image recognition at the n-1 th time And the shooting time ; Acquiring recording time and distance shooting time Recent speed ; Calculating the number of the video frame for image recognition from the inspection video for the nth time by adopting the following function : Wherein, the The standard distance is indicated as such, Representing the calculated coefficients of the calculation, , The standard speed is indicated and the speed of the vehicle is indicated, Indicating the set constant value of the current value, , And The lower limit value and the upper limit value of the number change amount are respectively indicated.
2. 2. The method for analyzing hidden danger of urban expressway sound barrier based on video analysis according to claim 1, wherein the video frames for image recognition obtained in the n-1 th time are rapidly recognized to obtain the distance between the acquisition camera and the sound barrier Comprising: By using Representing the video frame obtained for image recognition at the n-1 th time; For a pair of The process of performing the quick recognition is as follows: Respectively to Calculating each pixel point in the image data to obtain an edge coefficient of each pixel point; At the position of Deleting pixel points with edge coefficients smaller than the set edge coefficient threshold value to obtain a plurality of connected regions, and storing the obtained connected regions in a set ; Respectively obtain A length and a width of each communication region in (a); Communicating areas with the length smaller than the set length threshold value are gathered from the collection Deleted to obtain a collection ; Communicating regions with widths smaller than a set width threshold value from the collection Deleted to obtain a collection ; Respectively acquiring sets An area of each of the communication regions; acquiring a collection based on an area A communication region of the upright post belonging to the sound barrier; Calculating the image distance of the communication areas of the adjacent upright posts belonging to the sound barrier; Obtaining distance between camera and sound barrier based on image distance calculation 。
3. 3. The method for analyzing hidden danger of sound barrier in urban expressway based on video analysis according to claim 1, wherein the recording time is acquired from the shooting time Recent speed Comprising: the recording time in the set of the speed information is ordered according to the sequence from the early to the late to obtain the set ; Separately calculate Each recording time and The absolute value of the difference between them; taking the speed corresponding to the recording moment corresponding to the absolute value of the minimum difference value as the speed 。
4. 4. The method for analyzing hidden danger of urban expressway sound barrier based on video analysis according to claim 1, wherein each video frame in the set to be identified is respectively input into a pre-trained neural network model for identification, and an identification result of the video frame is obtained, comprising: preprocessing each video frame in the set to be identified respectively to obtain the video frame to be identified; and inputting the video frames to be identified into a pre-trained neural network model for identification, and obtaining the identification result of the video frames.
5. 5. The method for analyzing hidden danger of sound barrier in urban expressway based on video analysis according to claim 1, wherein the recognition result is that the sound barrier in the video frame has hidden danger or has no hidden danger.
6. 6. The method for analyzing hidden danger of sound barrier in urban expressway based on video analysis according to claim 5, wherein when the recognition result is that the sound barrier in the video frame has a hidden danger, the recognition result further includes a type of hidden danger existing in the sound barrier in the video frame.

Description

Urban expressway sound barrier hidden danger analysis method based on video analysis Technical Field The invention relates to the field of video identification, in particular to an urban expressway sound barrier hidden danger analysis method based on video analysis. Background The sound barrier is a specially designed acoustic baffle that stands between the noise source and the sound receiving point. Because some urban expressways are closer to buildings, in order to reduce the influence of noise on buildings closer to the urban expressways, sound barriers are usually arranged on two sides of the urban expressways. The sound barrier can be influenced by factors such as temperature difference, rainwater, strong wind, road vibration and the like in the using process, so that the conditions such as deformation, rust, loss of parts and the like can occur to the structure of the sound barrier, and potential safety hazards can be generated. Therefore, the sound barrier needs to be inspected regularly, and the sound barrier with potential safety hazard is repaired or replaced in time. The traditional inspection mode is to observe the sound barrier through human eyes and judge whether the sound barrier has potential safety hazard or not, but the inspection mode is easy to leak inspection due to human eye fatigue. Disclosure of Invention The invention aims to disclose an urban expressway sound barrier hidden danger analysis method based on video analysis, which solves the problem of how to reduce the probability of missed detection when a sound barrier is inspected. In order to achieve the above purpose, the present invention provides the following technical solutions: the invention provides an urban expressway sound barrier hidden danger analysis method based on video analysis, which comprises the following steps: s1, shooting a sound barrier at the roadside of an urban expressway to obtain a patrol video containing the sound barrier; S2, acquiring a video frame for image recognition from the inspection video, and storing the acquired video frame into a set to be recognized; S3, respectively inputting each video frame in the set to be identified into a pre-trained neural network model for identification, and obtaining an identification result of the video frame; S4, storing the identification result into a result set. Preferably, shooting a sound barrier at the roadside of an urban expressway, and obtaining a patrol video containing the sound barrier, wherein the patrol video comprises: Shooting a sound barrier at the roadside of an urban expressway through a vehicle provided with a camera, obtaining a patrol video containing the sound barrier, recording the speed information of the vehicle by adopting a fixed acquisition period in the shooting process, and obtaining a set of speed information. Preferably, the speed information includes a speed and a recording timing. Preferably, before acquiring the video frame for image recognition from the inspection video, the method further comprises: and continuously numbering video frames in the inspection video according to shooting time, wherein the number is larger when the shooting time is later. Preferably, the method for obtaining the video frame for image recognition from the inspection video comprises the following steps: The method comprises the steps of obtaining a first frame of video frame in a patrol video for the first time, wherein the first frame of video frame is used for carrying out image recognition; The process of the nth acquisition is as follows, n is more than or equal to 2: the video frames for image recognition, which are obtained in the n-1 th time, are rapidly recognized to obtain the distance between the acquisition camera and the sound barrier ; Acquiring the number of the video frame obtained for image recognition at the n-1 th timeAnd the shooting time; Acquiring recording time and distance shooting timeRecent speed; Calculating the number of the video frame for image recognition from the inspection video for the nth time by adopting the following function: Wherein, the The standard distance is indicated as such,Representing the calculated coefficients of the calculation,,The standard speed is indicated and the speed of the vehicle is indicated,Indicating the set constant value of the current value,,AndThe lower limit value and the upper limit value of the number change amount are respectively indicated. Preferably, the video frame for image recognition obtained in the n-1 th time is rapidly recognized to obtain the distance between the acquisition camera and the sound barrierComprising: By using Representing the video frame obtained for image recognition at the n-1 th time; For a pair of The process of performing the quick recognition is as follows: Respectively to Calculating each pixel point in the image data to obtain an edge coefficient of each pixel point; At the position of Deleting pixel points with edge coefficients smaller than the s