CN-121982624-A - Video object recognition-based automatic underbridge space monitoring and early warning system

CN121982624ACN 121982624 ACN121982624 ACN 121982624ACN-121982624-A

Abstract

The invention belongs to the field of image processing, and relates to an automatic monitoring and early warning system for an under-bridge space based on video target identification, which comprises an image acquisition module, a video acquisition module and a video acquisition module, wherein the image acquisition module acquires video data and image data of the under-bridge space; the image transmission module transmits image data and video data to the under-bridge space foreign matter identification unit, the under-bridge space foreign matter identification unit identifies the video data and the image data based on a convolutional neural network combined with a visual transducer model to distinguish whether the foreign matter exists, the alarm unit sends out corresponding alarm prompts according to the acquired data, the alarm unit collects environment data and the image data of the under-bridge space, integrates the environment data and the image data to form three-dimensional data, calculates parameters, deploys the three-dimensional data, the image data and the environment data after training based on the convolutional neural network combined with the visual transducer model, identifies and processes the three-dimensional data, the image data and the environment data to identify whether the foreign matter exists, acquires alarm signals according to identification results, and carries out alarm after statistical information.

Inventors

PENG DEPIN
Shen Lingjie
ZHENG ZIANG
SUN KEQUAN

Assignees

绍兴市交投机电信息有限责任公司

Dates

Publication Date: 20260505
Application Date: 20251216

Claims (10)

1. The automatic under-bridge space monitoring and early warning system based on video target identification is characterized by comprising an image acquisition module, an image transmission module, an under-bridge space foreign matter identification unit and an alarm unit, wherein the image acquisition module acquires video data and image data of an under-bridge space, the image transmission module transmits the image data and the video data to the under-bridge space foreign matter identification unit, the under-bridge space foreign matter identification unit identifies and processes the video data and the image data based on a convolutional neural network and a visual transducer model to distinguish whether the under-bridge space foreign matter is contained or not, and the alarm unit sends out corresponding alarm prompts according to the data acquired by the under-bridge space foreign matter identification unit, and the automatic under-bridge space monitoring and early warning system comprises the following steps: collecting environment data and image data of the space under the bridge, integrating to form three-dimensional data and calculating parameters; based on the convolutional neural network, the vision transducer model is combined for training and then deployment, three-dimensional data, image data and environment data are identified and processed, and whether foreign matters exist or not is identified; And acquiring an alarm signal according to the identification result, and alarming after counting information.
2. The automatic monitoring and early warning system for the space under the bridge based on the video target recognition according to claim 1, wherein the steps of collecting the environmental data of the space under the bridge, integrating the environmental data with the image data to form three-dimensional data and calculating parameters comprise the following steps: Collecting image data and video data through an image collecting module; Acquiring environmental data of a space under a bridge, wherein the environmental data comprise the distance between an image acquisition module and the bridge, the angle formed by the image acquisition module and the bridge, the air humidity, the acquisition angle, the illumination intensity, the wind speed, the wind direction and the image temperature; integrating the image data, the video data and the environment data to form three-dimensional data; and optimizing the three-dimensional data, improving the image quality and the image acquisition quantity, and calculating the optimal parameters of image acquisition.
3. The automatic monitoring and early warning system for the space under the bridge based on the video target recognition according to claim 1, wherein the steps of collecting the environmental data of the space under the bridge, integrating the environmental data with the image data to form three-dimensional data and calculating parameters further comprise the following steps: Removing influence data, including water mist and illumination conditions, of the collected data, which are caused by external factors on the data collection quality; and calculating optimal parameters of the acquired data, including acquisition range, distance, angle and acquisition quantity.
4. The automatic monitoring and early warning system for the space under the bridge based on the video object recognition according to claim 1, wherein the recognition unit for the foreign matters in the space under the bridge recognizes and processes video data and image data based on a convolutional neural network in combination with a visual transducer model, and distinguishes whether the foreign matters are contained in the video data and the image data, and specifically comprises the following contents: Simulating environmental data of the space under the bridge through a bridge image simulation module, and integrating the environmental data with the image data to form three-dimensional data; The three-dimensional data is optimized through an image acquisition optimization module, and the acquisition quantity and the acquisition quality of image acquisition are increased; the optimized three-dimensional data is identified through an identification and confirmation module, and whether foreign matters are contained in the three-dimensional data is determined; Identifying the foreign matters through an alarm information statistics module, determining the types, the volumes and unknown parameters of the foreign matters, and determining an alarm mode and an alarm period according to the danger coefficients of the foreign matters; And the alarm time statistics module is used for collecting alarm signals and counting alarm times and alarm time information in a continuous alarm period in real time.
5. The automatic monitoring and early warning system for the space under the bridge based on the video object recognition according to claim 4, wherein the recognition confirming module is used for recognizing the optimized three-dimensional data and determining whether the optimized three-dimensional data contains foreign matters or not, and the system comprises the following contents: Three-dimensional data processed by the image acquisition optimization module are acquired and identified by combining a vision transducer model based on a convolutional neural network; If the foreign matter is primarily judged, positioning through the attention weight output by the model, selecting a plurality of image areas with highest weight as foreign matter candidate areas, and acquiring three-dimensional coordinates and contour features of the areas; And (3) performing secondary identification on the foreign object candidate region, comparing the foreign object candidate region with the three-dimensional data, judging whether the position of the foreign object exists in the space under the bridge, if so, judging that the foreign object exists in the space under the bridge, and otherwise, eliminating the foreign object.
6. The automatic monitoring and early warning system for the space under the bridge based on the video target recognition according to claim 4 is characterized in that the alarm information statistics module is used for recognizing the foreign matters, determining the types, the volumes and the unknown parameters of the foreign matters and determining the alarm mode and the alarm time period according to the danger coefficient of the foreign matters, the alarm time statistics module is used for collecting alarm signals and counting the alarm times and the alarm time information in the continuous alarm time period in real time, and the system comprises the following contents: When the identification confirmation module confirms that the data contains foreign matters, the identification confirmation module sends confirmation information to the acquisition signal receiving module to finish acquisition of the alarm signal; Comparing the image data containing the foreign matters with historical data in a database, and triggering an alarm confirmation module to acquire signals if the difference exists; The acquisition signal comprises an acquisition alarm time period and alarm times, and is sent to an alarm time statistics module, and is sent to an information processing module after statistics.
7. The automatic monitoring and early warning system for under-bridge space based on video object recognition according to claim 6, wherein the comparing the image data containing the foreign matters with the historical data in the database comprises the following contents: collecting historical data containing foreign matters in the space under the bridge; Preprocessing historical data to remove data influenced by external factors; collecting the image data which is identified to contain the foreign matters, and comparing the image data with the preprocessed historical data; And alarming according to whether the comparison result is different, if yes, triggering an alarm confirmation module to acquire signals, and if no, not acquiring signals.
8. The automatic monitoring and early warning system for the space under the bridge based on the video target recognition according to claim 6, wherein the acquisition signal comprises an acquisition alarm time period and an alarm frequency, and is sent to an alarm time statistics module, and is sent to an information processing module after statistics, and the system comprises the following contents: Counting the times, interval time and alarm occurrence time period of data containing foreign matters according to the collected alarm times and alarm time period information; Recording alarm times and alarm time period information, and generating a classification record table according to the alarm times and the alarm time period information; And when foreign matter alarm information continuously appears in the classification record table and the preset times are reached, determining to send out an alarm signal.
9. The automatic monitoring and early warning system for the under-bridge space based on video target recognition according to claim 1 is characterized in that the image acquisition optimization module comprises image optimization and acquisition quantity calculation, wherein the image optimization comprises image denoising, data in image data, which influence data acquisition quality due to external factors, are removed, the external factors comprise water mist and illumination conditions, and the acquisition quantity calculation comprises calculation of optimal parameters of image data acquisition, including the range, distance, angle and image quantity of acquired images.
10. An automatic monitoring and early warning device for an under-bridge space based on video target recognition is characterized by comprising an image acquisition device, an under-bridge space recognition device and an alarm device, wherein the image acquisition device is in communication connection with the under-bridge space recognition device and is used for acquiring environment video data and image data of the under-bridge space, the under-bridge space recognition device is used for processing the video data and the image data acquired by the image acquisition device, determining whether foreign matters exist in the video data and the image data and determining parameters of the foreign matters, calculating the risk coefficient of the foreign matters according to the determined parameters, and controlling the alarm device, and the alarm device comprises an early warning device and an alarm device and is used for sending an alarm signal when the under-bridge space recognition device determines that the under-bridge space is in a dangerous state.

Description

Video object recognition-based automatic underbridge space monitoring and early warning system Technical Field The invention belongs to the field of image processing, and particularly relates to an automatic under-bridge space monitoring and early warning system based on video target recognition. Background In recent years, with the development of computer technology, the traditional infrastructure monitoring maintenance mode has been changed from a passive reactive monitoring mode to an active early warning monitoring mode, and the problem of foreign matter accumulation in the space under the bridge gradually attracts attention of related functional departments. Because the space under the bridge is influenced by the factors of the bridge, the natural environment and the like, the phenomenon of collapse or accumulation easily occurs at different time points or spaces, thereby easily causing traffic jam and even influencing the use safety of the bridge. Therefore, it is urgently needed to design a monitoring system capable of monitoring the space state under the bridge in real time, so as to improve the self, management and maintenance level of the bridge and provide technical support for safety monitoring of the space under the bridge. The prior art has the automatic monitoring early warning system of bridge foreign matter, can accurately detect whether there is the foreign matter to pile up in the space under the bridge through the cooperation of infrared sensor, rain sensor and wind speed sensor three sensor, even also can detect the environmental information in space under the bridge under the normal condition of no foreign matter, when the processor judges that there is the foreign matter to pile up in space under the bridge and pile up to certain height according to the information of infrared sensor, rain sensor and wind speed sensor received, send warning information to monitor terminal, make the monitoring personnel can in time look for warning information and can obtain the processing as early as possible. But this scheme belongs to traditional infrastructure control maintenance mode, belongs to passive reactive's monitoring mode, and this kind of mode can only gather the environmental information in underbridge space, and after the information acquisition of the same day, can only provide the environmental information of the same day, can not carry out the information comparison according to historical data, can not accurately reflect the state in underbridge space, can't prejudge the underbridge space hidden danger simultaneously, and the alarming mode is comparatively single, is inconvenient for staff in time analysis data and in time handle. Therefore, there is a need for an automatic under-bridge space monitoring and early warning system based on video object recognition to solve the above problems. Disclosure of Invention The invention provides an automatic under-bridge space monitoring and early warning system based on video target identification. The automatic monitoring and early warning system for the space under the bridge based on the video target identification comprises an image acquisition module, an image transmission module, a space under the bridge foreign matter identification unit and an alarm unit, wherein the image acquisition module acquires video data and image data of the space under the bridge, the image transmission module transmits the image data and the video data to the space under the bridge foreign matter identification unit, the space under the bridge foreign matter identification unit identifies and processes the video data and the image data based on a convolutional neural network and a visual transducer model to distinguish whether the space under the bridge foreign matter is contained or not, and the alarm unit sends out corresponding alarm prompts according to the data acquired by the space under the bridge foreign matter identification unit, and the automatic monitoring and early warning system comprises the following steps: collecting environment data and image data of the space under the bridge, integrating to form three-dimensional data and calculating parameters; based on the convolutional neural network, the vision transducer model is combined for training and then deployment, three-dimensional data, image data and environment data are identified and processed, and whether foreign matters exist or not is identified; And acquiring an alarm signal according to the identification result, and alarming after counting information. Preferably, the collecting environmental data of the space under the bridge, integrating the environmental data with the image data, forming three-dimensional data and calculating parameters includes the following contents: Collecting image data and video data through an image collecting module; Acquiring environmental data of a space under a bridge, wherein the environmental data comprise the distance between an image acquisition mo