CN118135480A - Visual image processing method and system for electromechanical construction of tunnel - Google Patents
Visual image processing method and system for electromechanical construction of tunnel Download PDFInfo
- Publication number
- CN118135480A CN118135480A CN202410204540.5A CN202410204540A CN118135480A CN 118135480 A CN118135480 A CN 118135480A CN 202410204540 A CN202410204540 A CN 202410204540A CN 118135480 A CN118135480 A CN 118135480A
- Authority
- CN
- China
- Prior art keywords
- frame
- abnormal event
- frames
- key
- video data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000010276 construction Methods 0.000 title claims abstract description 90
- 230000000007 visual effect Effects 0.000 title claims abstract description 34
- 238000003672 processing method Methods 0.000 title claims abstract description 24
- 230000002159 abnormal effect Effects 0.000 claims abstract description 102
- 238000001514 detection method Methods 0.000 claims abstract description 57
- 238000004458 analytical method Methods 0.000 claims abstract description 37
- 238000012545 processing Methods 0.000 claims abstract description 34
- 238000000034 method Methods 0.000 claims abstract description 21
- 238000010801 machine learning Methods 0.000 claims abstract description 18
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 13
- 238000013144 data compression Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 24
- 238000012360 testing method Methods 0.000 claims description 13
- 230000004044 response Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 8
- 239000013598 vector Substances 0.000 claims description 7
- 238000013145 classification model Methods 0.000 claims description 6
- 238000012423 maintenance Methods 0.000 claims description 6
- 238000003062 neural network model Methods 0.000 claims description 5
- 238000007781 pre-processing Methods 0.000 claims description 5
- 238000012800 visualization Methods 0.000 claims description 5
- 238000013500 data storage Methods 0.000 claims description 3
- 238000012544 monitoring process Methods 0.000 abstract description 20
- 238000007726 management method Methods 0.000 description 11
- 230000008859 change Effects 0.000 description 7
- 238000007906 compression Methods 0.000 description 7
- 230000006835 compression Effects 0.000 description 7
- 230000008569 process Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000009430 construction management Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000012216 screening Methods 0.000 description 3
- 230000003068 static effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000033001 locomotion Effects 0.000 description 2
- 230000007246 mechanism Effects 0.000 description 2
- 239000000203 mixture Substances 0.000 description 2
- 238000011056 performance test Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000013136 deep learning model Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000009826 distribution Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010191 image analysis Methods 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 230000009191 jumping Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000002265 prevention Effects 0.000 description 1
- 238000007637 random forest analysis Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
Landscapes
- Image Analysis (AREA)
Abstract
The application relates to a visual image processing method and a visual image processing system for electromechanical construction of a tunnel, which belong to the technical field of electromechanical construction monitoring, wherein the visual image processing method comprises the following steps: acquiring video data of a tunnel electromechanical construction area in real time; extracting continuous frame sequences according to video data, and carrying out inter-frame difference analysis; detecting and marking key frames and non-key frames based on a machine learning algorithm according to an inter-frame difference analysis result; inputting each key frame into a pre-trained abnormal event detection model respectively to obtain an abnormal event detection result; responding to the abnormal event detection result to determine that the abnormal event exists, and determining the corresponding key frame as an abnormal event frame; and carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data. The method and the system can effectively identify and optimize redundant data in the monitoring video, and simultaneously ensure the safety and engineering quality of the electromechanical construction of the tunnel.
Description
Technical Field
The application relates to the technical field of electromechanical construction monitoring, in particular to a visual image processing method and system for tunnel electromechanical construction.
Background
With development of tunnel engineering technology, the automation and informatization levels of electromechanical construction are continuously improved, and an image monitoring system plays an increasingly important role in the electromechanical construction of tunnels. Through the monitoring video, constructors can monitor the state of the construction site in real time, construction safety is ensured, and construction quality is improved.
However, existing tunnel electromechanical construction image monitoring systems often face some technical challenges in processing and analyzing surveillance video data. For example, surveillance videos often contain a large number of duplicate, useless, or other redundant images, which not only occupy a large amount of memory space, but also consume a large amount of computing resources during data transmission and processing, which is particularly apparent at certain stages of tunnel construction, such as during non-operational periods, where the surveillance video may only capture static scenes for a long period of time. For another example, in the prior art, a continuous recording manner is often adopted when monitoring video data is processed, so that importance and urgency of a scene are not distinguished, and the response is not timely enough when an abnormal situation is found, so that quick response to a key event cannot be realized.
Therefore, how to effectively identify and optimize redundant data in the monitoring video, and simultaneously detect key events of a construction site rapidly, ensure the safety and engineering quality of the electromechanical construction of the tunnel, and become the current problem to be solved urgently.
Disclosure of Invention
In order to effectively identify and optimize redundant data in a monitoring video and simultaneously ensure the safety and engineering quality of the electromechanical construction of a tunnel, the application provides a visual image processing method and a visual image processing system for the electromechanical construction of the tunnel.
In a first aspect, the application provides a visual image processing method for electromechanical construction of a tunnel, which adopts the following technical scheme:
a visual image processing method for tunnel electromechanical construction comprises the following steps:
Acquiring video data of a tunnel electromechanical construction area in real time;
extracting a continuous frame sequence according to the video data, and carrying out inter-frame difference analysis;
Detecting and marking key frames and non-key frames based on a machine learning algorithm according to an inter-frame difference analysis result;
inputting each key frame into a pre-trained abnormal event detection model to obtain an abnormal event detection result;
Responding to the abnormal event detection result to determine that an abnormal event exists, and determining a corresponding key frame as an abnormal event frame;
and carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data.
By adopting the technical scheme, the video data of the tunnel electromechanical construction site is analyzed and processed frame by frame, redundant data in the monitoring video can be effectively identified and optimized, the storage resource use is balanced, meanwhile, the reservation requirement of key information is met, and the key event of the construction site is detected rapidly, so that construction management staff can respond in time, the safety and engineering quality of the tunnel electromechanical construction are ensured, and the construction safety management level is improved.
Optionally, the step of extracting a continuous sequence of frames from the video data and performing inter-frame difference analysis includes:
Dividing the video data into a sequence of successive frames;
carrying out gray scale processing on every two continuous frames in the frame sequence;
Separating foreground objects from the two continuous frames based on a dynamic background model;
according to the foreground object of each two continuous frames, calculating to obtain corresponding pixel differences;
and obtaining an inter-frame difference analysis result according to the pixel difference corresponding to each two continuous frames.
By adopting the technical scheme, the pixel change between two continuous frames is quantized by utilizing the inter-frame difference analysis, so that basis is provided for the subsequent identification of the significant change between the continuous frames.
Optionally, the step of detecting and marking the key frames and the non-key frames based on the machine learning algorithm based on the inter-frame difference analysis result includes:
Obtaining pixel differences corresponding to each two continuous frames according to the inter-frame difference analysis result;
Respectively judging whether the pixel difference corresponding to each two continuous frames exceeds a preset threshold value, and if so, determining that the two continuous frames are candidate key frames; if not, marking the two continuous frames as non-key frames;
extracting the characteristics of the candidate key frames and generating corresponding characteristic vectors;
And inputting the feature vector corresponding to each candidate key frame into a pre-trained machine learning classification model, and determining and marking the key frames and the non-key frames according to the model output result.
By adopting the technical scheme, a preliminary screening mechanism is provided by utilizing inter-frame difference analysis, candidate key frames can be rapidly and roughly identified, and the reduction of the number of frames needing further analysis is facilitated, so that the detection efficiency of the key frames is improved; and then, the key frame detection is carried out through the machine learning classification model, so that the intelligent level of visual image processing of the electromechanical construction of the tunnel is improved, the key frame can be identified more accurately and efficiently, and support is provided for application such as construction monitoring, abnormal event detection and the like.
Optionally, the method further includes a training step of the abnormal event detection model, the training step including:
acquiring a sample data set; the sample data set comprises a plurality of image data pre-marked with abnormal events or normal events;
preprocessing the sample data set, and dividing the sample data set into a training set and a testing set;
inputting the training set into a pre-constructed neural network model for training to obtain the abnormal event detection model;
and testing the abnormal event detection model based on the test set, and correcting parameters of the model to obtain the trained abnormal event detection model.
By adopting the technical scheme, a proper neural network model is selected and trained, and the model is tested and parameter corrected, so that an efficient and accurate abnormal event detection model is constructed, the model analyzes an input key frame, judges whether the key frame has an abnormal event according to the characteristics learned during training, and quickly and accurately identifies the abnormal condition so as to prompt a manager to take corresponding measures in time, thereby improving the safety and management efficiency of a construction site.
Optionally, after the step of determining the corresponding key frame as the abnormal event frame, the method further includes:
And generating abnormal event prompt information according to the abnormal event frame and the corresponding time stamp, and sending the abnormal event prompt information to a maintenance terminal.
By adopting the technical scheme, the intelligent abnormal event detection and the real-time communication response are combined, so that the response speed and accuracy of the abnormal event are improved, potential safety hazards can be found and treated in time, the safety of constructors is ensured, and the smooth progress of construction is ensured.
Optionally, the image processing method further includes:
Content classification is carried out according to the images corresponding to the key frames, and classification labels of the images are determined;
Establishing an index for the image corresponding to each key frame according to the classification label of the image;
and responding to the query element input by the user, determining an image set matched with the query element according to the index, and outputting the image set as a query result.
By adopting the technical scheme, based on automatic image content classification and index establishment, the management and the retrieval of the visual images of the tunnel electromechanical construction become simpler and more efficient, the processing speed and the accuracy of image data are improved, the experience of the user for retrieving the required information is improved, and powerful data support is provided for decision support.
In a second aspect, the application provides a visual image processing system for electromechanical construction of a tunnel, which adopts the following technical scheme:
a tunnel electromechanical construction visualization image processing system, comprising:
The video data acquisition module is used for acquiring video data of the electromechanical construction area of the tunnel in real time;
the inter-frame difference analysis module is used for extracting continuous frame sequences according to the video data and carrying out inter-frame difference analysis;
The frame detection module is used for detecting and marking key frames and non-key frames based on a machine learning algorithm according to the inter-frame difference analysis result;
the abnormal event detection result generation module is used for respectively inputting each key frame into a pre-trained abnormal event detection model to obtain an abnormal event detection result;
the abnormal event frame determining module is used for determining a corresponding key frame as an abnormal event frame in response to the abnormal event detection result as the abnormal event;
And the video data storage module is used for carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data.
Optionally, the image processing system further comprises:
The image classification module is used for classifying the content according to the image corresponding to the key frame and determining the classification label of the image;
The index establishing module is used for establishing an index for the image corresponding to each key frame according to the classification label of the image;
And the query result output module is used for responding to the query element input by the user, determining an image set matched with the query element according to the index and outputting the image set as a query result.
In a third aspect, the present application provides a computer device, which adopts the following technical scheme:
A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method according to the first aspect when executing the computer program.
In a fourth aspect, the present application provides a computer readable storage medium, which adopts the following technical scheme:
a computer readable storage medium storing a computer program capable of being loaded by a processor and executing any one of the methods of the first aspect.
In summary, the present application includes at least one of the following beneficial technical effects: the video data of the tunnel electromechanical construction site is analyzed and processed frame by frame, redundant data in the monitoring video can be effectively identified and optimized, the storage resource use is balanced, meanwhile, the reservation requirement of key information is met, and the key event of the construction site is detected rapidly, so that construction manager can respond in time, the safety and engineering quality of the tunnel electromechanical construction are ensured, and the construction safety management level is improved.
Drawings
Fig. 1 is a schematic flow chart of a visual image processing method for tunnel electromechanical construction according to one embodiment of the present application.
Fig. 2 is a second flow diagram of a visual image processing method for tunnel electromechanical construction according to one embodiment of the present application.
Fig. 3 is a third flow diagram of a visual image processing method for tunnel electromechanical construction according to one embodiment of the present application.
Fig. 4 is a fourth flowchart of a visual image processing method for tunnel electromechanical construction according to one embodiment of the present application.
Fig. 5 is a fifth flowchart of a visual image processing method for tunnel electromechanical construction according to one embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings 1 to 5 and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the application.
The embodiment of the application discloses a visual image processing method for tunnel electromechanical construction.
Referring to fig. 1, a visual image processing method for electromechanical construction of a tunnel includes:
step S101, video data of a tunnel electromechanical construction area is obtained in real time;
Specifically, a camera or other video acquisition equipment can be utilized to continuously monitor the electromechanical construction area of the tunnel, and video streams can be captured in real time.
The method comprises the steps of acquiring video data, acquiring the video data, wherein the video data is acquired by a video processing system, and acquiring the video data according to the video data, wherein the video data is acquired by the video processing system, so that the video data is acquired by the video processing system;
step S102, extracting continuous frame sequences according to video data, and carrying out inter-frame difference analysis;
The video data is divided into a series of continuous frames, dynamic changes in the scene, such as moving construction equipment or personnel, are analyzed by comparing pixel differences between the continuous frames, so that dynamic contents in the video are rapidly identified, a basis is provided for detection of key frames, and the calculated amount of subsequent processing steps is reduced;
Step S103, detecting and marking key frames and non-key frames based on a machine learning algorithm according to the inter-frame difference analysis result;
Wherein, the key frame refers to a frame containing important construction activities, potential dangers or significant changes, and the non-key frame refers to a frame with less information, such as repeated images, useless images or other redundant images;
Specifically, a machine learning algorithm is applied to identify and mark key frames and non-key frames in the video by utilizing a preliminary judgment result of the inter-frame difference analysis so as to extract key information from a large amount of video data, thereby effectively improving the efficiency of monitoring the electromechanical construction of the tunnel;
Step S104, each key frame is respectively input into a pre-trained abnormal event detection model to obtain an abnormal event detection result;
The key frames are sent into an abnormal event detection model which is trained in advance through a large amount of marking data, and the model can identify abnormal events in the construction process, such as safety accidents or equipment faults, so that the identification capacity of potential risks is improved, and timely countermeasures are facilitated;
Step S105, determining the corresponding key frame as an abnormal event frame in response to the abnormal event detection result being that an abnormal event exists;
when the output result of the model is that an abnormal event is detected, the corresponding key frame is marked as an abnormal event frame, so that construction managers can quickly locate and review the abnormal event, and the method is very important for accident investigation and future risk prevention;
And step S106, carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data.
Specifically, data compression refers to reducing the storage space occupied by data through a compression algorithm, and in the field of video monitoring, it generally means reducing the size of a video file by reducing redundant information so as to save storage space and bandwidth, including lossless compression and lossy compression; in the embodiment of the present application, the non-key frames are typically subjected to lossy compression to save storage space.
Wherein, the high-quality storage means a mode of storing data with high fidelity, so as to ensure the integrity and accuracy of the data; for video data, high quality storage is stored using a preset compression ratio, a preset resolution, and a preset frame rate, which may be preset according to monitoring standard requirements, typically a lower compression ratio and a higher resolution and frame rate, so as to maintain detailed information and definition of video content.
It will be appreciated that in order to optimize storage space, non-key frames are reduced in data size by compression algorithms, while key frames and abnormal event frames are stored in a high quality format to preserve detailed information, thereby reducing the occupation of storage resources by duplicate, useless or other redundant images, ensuring the integrity and availability of key information.
In the embodiment, the video data of the tunnel electromechanical construction site is analyzed and processed frame by frame, redundant data in the monitoring video can be effectively identified and optimized, the storage resource use is balanced, meanwhile, the reservation requirement of key information is met, and the key event of the construction site is detected rapidly, so that construction management staff can respond in time, the safety and engineering quality of the tunnel electromechanical construction are ensured, and the construction safety management level is improved.
Referring to fig. 2, as an embodiment of step S102, the steps of extracting a continuous frame sequence from video data and performing inter-frame difference analysis include:
step S201, dividing video data into continuous frame sequences;
The continuous video stream is divided into independent static image frames so as to be convenient for frame-by-frame processing, and a basis is provided for subsequent image analysis;
Step S202, gray scale processing is carried out on every two continuous frames in the frame sequence;
The gray level image has only one color channel, and the color image usually has three colors of red, green and blue, so that the data volume is reduced, the complexity of data processing is simplified and the calculation speed of the subsequent steps is increased by converting the color frame into the gray level frame;
step S203, separating foreground objects from two continuous frames based on the dynamic background model;
specifically, the dynamic background model (e.g., gaussian mixture model) can detect and separate foreground objects, i.e., dynamically changing parts, such as moving people, vehicles, construction equipment, etc., from each current frame, and ignore static background by focusing on dynamic content in the video, thereby improving the detection accuracy of key information.
The logic principle of dynamic background modeling is to analyze a video frame sequence and establish a statistical model to represent typical characteristics (such as color, texture and the like) of the background; for example, in a Gaussian Mixture Model (GMM), the color value of each pixel is assumed to be mixed by a plurality of gaussian distributions, reflecting the change in time sequence of the point, in such a way that the model can accommodate small changes in the background while marking as foreground those pixels that are significantly different from the background model.
Step S204, according to the foreground object of every two continuous frames, calculating to obtain the corresponding pixel difference;
Wherein the pixel difference can determine the degree of change between two frames, which can be achieved by comparing the changes in pixel values;
Step S205, according to the pixel difference corresponding to every two continuous frames, obtaining the inter-frame difference analysis result.
In the above embodiment, the inter-frame difference analysis is used to quantify the pixel change between two consecutive frames, thereby providing a basis for subsequent identification of significant changes between consecutive frames.
Referring to fig. 3, as an embodiment of step S103, the step of detecting and marking key frames and non-key frames based on a machine learning algorithm according to the inter-frame difference analysis result includes:
Step S301, obtaining pixel differences corresponding to every two continuous frames according to the inter-frame difference analysis result;
the pixel difference can quantify the change between frames, and provides a preliminary screening basis for the subsequent key frame detection;
Step S302, judging whether the pixel difference corresponding to each two continuous frames exceeds a preset threshold value or not respectively, and if not, jumping to step S303; if yes, jump to step S304;
Wherein the preset threshold may be preconfigured according to historical experience, applying thresholding to determine which variance changes are significant, thereby identifying meaningful motions or changes;
Step S303, marking two continuous frames as non-key frames;
Step S304, determining that two continuous frames are candidate key frames;
Specifically, if the pixel difference exceeds a preset threshold, it indicates that a significant change occurs between the two frames, i.e., both frames can be marked as candidate key frames, thereby ensuring that important changes are not missed; if the pixel difference does not exceed the preset threshold, the two frames are possibly repeated or inconspicuous images, and the non-key frames are effectively filtered, so that the calculated amount of subsequent processing is reduced;
it will be appreciated that candidate keyframes are a series of frames that identify significant changes or movements, but not all of these frames contain valuable information to the construction manager, and thus require further keyframe detection and marking steps to identify truly meaningful keyframes.
Step S305, extracting the characteristics of the candidate key frames and generating corresponding characteristic vectors;
Wherein for each candidate key frame, extracting features that characterize its content, such as color histograms, texture features, edge features, etc., and combining these features into a feature vector, thereby providing a machine learning model with a numerical representation that can be used for classification, key information of the frame can be captured and used for further analysis;
Step S306, inputting the feature vector corresponding to each candidate key frame into a pre-trained machine learning classification model, and determining and marking the key frames and the non-key frames according to the model output result.
The machine learning classification model can be selected from a support vector machine, a random forest, a deep neural network and the like, and can be based on a large amount of training data, for example, a large amount of tunnel electromechanical construction monitoring videos are collected, key frames and non-key frames are marked manually, and a training data set is constructed; the model can judge whether each candidate key frame is a real key frame according to the learned rule, thereby improving the accuracy of key frame detection.
In the above embodiment, a preliminary screening mechanism is provided by using inter-frame difference analysis, so that candidate key frames can be rapidly and roughly identified, and the number of frames needing further analysis is reduced, thereby improving the detection efficiency of the key frames; and then, the key frame detection is carried out through the machine learning classification model, so that the intelligent level of visual image processing of the electromechanical construction of the tunnel is improved, the key frame can be identified more accurately and efficiently, and support is provided for application such as construction monitoring, abnormal event detection and the like.
Referring to fig. 4, as an embodiment of the abnormal event detection model in step S104, a training step of the abnormal event detection model is further included, the training step including:
Step S401, a sample data set is acquired; the sample data set comprises a plurality of image data pre-marked with abnormal events or normal events;
Specifically, enough image data is collected in advance, and the image data should cover various scenes of the electromechanical construction site of the tunnel, including normal working scenes and various abnormal events such as security violations, equipment faults, fires and the like; moreover, each image data needs to be marked in advance by a professional, and marking information comprises event types (normal or abnormal), abnormal types and the like;
It can be appreciated that the high-quality sample data set is the basis for training an accurate model, and by covering various scenes, the model can learn more features, so that generalization capability and accuracy of the model are improved;
step S402, preprocessing a sample data set, and dividing the sample data set into a training set and a testing set;
The preprocessing step comprises the steps of size adjustment, normalization, denoising, enhancement and the like of the image, and then the processed data set is divided into a training set and a testing set; image preprocessing can improve the speed and performance of model training, and reasonable data set division is helpful for evaluating the generalization capability of the model on unknown data.
Step S403, inputting a training set into a pre-constructed neural network model for training to obtain an abnormal event detection model;
For an image classification task, a Convolutional Neural Network (CNN) model can be selected, the model can effectively extract spatial hierarchy characteristics in an image, an image classification frame pre-training model such as ResNet, inception or VGG can be selected, and parameter fine adjustment is performed on the basis; by using a pre-trained CNN model, rich features learned over a large dataset can be utilized to help improve the recognition capabilities of the model in a particular construction scenario.
Step S404, testing the abnormal event detection model based on the test set, and correcting parameters of the model to obtain the trained abnormal event detection model.
Performing performance test on the trained model by using the divided test set, wherein the performance test comprises indexes such as calculation accuracy, recall rate, accuracy and the like; and correcting the model parameters based on the test result, such as adjusting the learning rate, increasing or decreasing training rounds, changing the network structure and the like, so as to improve the detection performance of the model.
It can be understood that the model test and the parameter correction are helpful to ensure that the model has higher accuracy and reliability in practical application, reduce the situations of false alarm and missing report, and improve the recognition capability of abnormal events.
In the embodiment, a proper neural network model is selected and trained, and the model is tested and parameter corrected, so that an efficient and accurate abnormal event detection model is constructed, the model analyzes an input key frame, judges whether the key frame has an abnormal event according to the characteristics learned during training, and rapidly and accurately identifies the abnormal condition, so that management personnel can be reminded of taking corresponding measures in time, and the safety and management efficiency of a construction site are improved.
As a further embodiment of the image processing method, after the step of determining that the corresponding key frame is an abnormal event frame, the method further includes:
And generating abnormal event prompt information according to the abnormal event frame and the corresponding time stamp, and sending the abnormal event prompt information to the maintenance terminal.
When an abnormal event frame is detected, the system generates abnormal event prompt information, wherein the information comprises an abnormal event type, occurrence time and the like; the prompt information can be sent to a maintenance terminal through a network, and the maintenance terminal can be a computer system of a monitoring center, mobile equipment of field maintenance personnel or other communication equipment.
In the embodiment, the intelligent abnormal event detection and the real-time communication response are combined, so that the response speed and accuracy of the abnormal event are improved, potential safety hazards can be found and processed in time, the safety of constructors is ensured, and the smooth progress of construction progress is ensured.
Referring to fig. 5, as a further embodiment of the image processing method, the image processing method further includes:
Step S501, classifying contents according to images corresponding to the key frames, and determining classification labels of the images;
Wherein each key frame image may be content classified using a trained deep learning model that analyzes the image's features and assigns them to predefined categories, such as workers, machinery, tools, etc.;
It will be appreciated that by automated content classification, the primary elements in the image can be quickly and accurately identified, thereby reducing the effort of manual annotation and providing a basis for subsequent indexing and retrieval.
Step S502, establishing an index for the image corresponding to each key frame according to the classification label of the image;
Wherein, the index can be a database or other data structure, wherein key information of the image and corresponding classification labels are recorded; by establishing the index, the retrieval process is more efficient, and the user can directly find out related images through the classification labels without checking each frame of images one by one, so that the retrieval speed and the user experience are greatly improved.
Step S503, responding to the query element input by the user, determining an image set matched with the query element according to the index, and outputting the image set as a query result.
When a user inputs a query element, the retrieval system returns an image set matched with the query element according to the established index, and the user can quickly obtain the required image information, so that the method is very convenient in terms of project management, progress tracking, safety monitoring and the like.
In the embodiment, based on automatic image content classification and index establishment, the management and the retrieval of the visual images of the tunnel electromechanical construction become simpler and more efficient, the processing speed and the accuracy of image data are improved, the experience of the user for retrieving the required information is improved, and powerful data support is provided for decision support.
The embodiment of the application also discloses a visual image processing system for the electromechanical construction of the tunnel.
A tunnel electromechanical construction visualization image processing system, comprising:
The video data acquisition module is used for acquiring video data of the electromechanical construction area of the tunnel in real time;
The inter-frame difference analysis module is used for extracting continuous frame sequences according to the video data and carrying out inter-frame difference analysis;
The frame detection module is used for detecting and marking key frames and non-key frames based on a machine learning algorithm according to the inter-frame difference analysis result;
The abnormal event detection result generation module is used for respectively inputting each key frame into a pre-trained abnormal event detection model to obtain an abnormal event detection result;
The abnormal event frame determining module is used for determining a corresponding key frame as an abnormal event frame in response to the abnormal event detection result as the abnormal event;
and the video data storage module is used for carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data.
In the embodiment, the video data of the tunnel electromechanical construction site is analyzed and processed frame by frame, redundant data in the monitoring video can be effectively identified and optimized, the storage resource use is balanced, meanwhile, the reservation requirement of key information is met, and the key event of the construction site is detected rapidly, so that construction management staff can respond in time, the safety and engineering quality of the tunnel electromechanical construction are ensured, and the construction safety management level is improved.
As a further embodiment of the image processing system, the image processing system further includes:
the image classification module is used for classifying the content according to the images corresponding to the key frames and determining classification labels of the images;
The index establishing module is used for establishing an index for the image corresponding to each key frame according to the classification label of the image;
and the query result output module is used for responding to the query elements input by the user, determining an image set matched with the query elements according to the index and outputting the image set as a query result.
In the embodiment, based on automatic image content classification and index establishment, the management and the retrieval of the visual images of the tunnel electromechanical construction become simpler and more efficient, the processing speed and the accuracy of image data are improved, the experience of the user for retrieving the required information is improved, and powerful data support is provided for decision support.
The visual image processing system for the tunnel electromechanical construction can realize any one of the visual image processing methods for the tunnel electromechanical construction, and the specific working process of each module in the visual image processing system for the tunnel electromechanical construction can refer to the corresponding process in the embodiment of the method.
In several embodiments provided by the present application, it should be understood that the methods and systems provided may be implemented in other ways. For example, the system embodiments described above are merely illustrative; for example, a division of a module is merely a logical function division, and there may be another division manner in actual implementation, for example, multiple modules may be combined or may be integrated into another system, or some features may be omitted or not performed.
The embodiment of the application also discloses computer equipment.
Computer equipment comprises a memory, a processor and a computer program which is stored in the memory and can run on the processor, wherein the processor realizes the tunnel electromechanical construction visualization image processing method when executing the computer program.
The embodiment of the application also discloses a computer readable storage medium.
A computer-readable storage medium storing a computer program capable of being loaded by a processor and executing any one of the tunnel electromechanical construction visualization image processing methods as described above.
Wherein a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device; program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
In the foregoing embodiments, the descriptions of the embodiments are focused on, and for those portions of one embodiment that are not described in detail, reference may be made to the related descriptions of other embodiments.
The foregoing description of the preferred embodiments of the application is not intended to limit the scope of the application in any way, including the abstract and drawings, in which case any feature disclosed in this specification (including abstract and drawings) may be replaced by alternative features serving the same, equivalent purpose, unless expressly stated otherwise. That is, each feature is one example only of a generic series of equivalent or similar features, unless expressly stated otherwise.
Claims (10)
1. The visual image processing method for the electromechanical construction of the tunnel is characterized by comprising the following steps of:
Acquiring video data of a tunnel electromechanical construction area in real time;
extracting a continuous frame sequence according to the video data, and carrying out inter-frame difference analysis;
Detecting and marking key frames and non-key frames based on a machine learning algorithm according to an inter-frame difference analysis result;
inputting each key frame into a pre-trained abnormal event detection model to obtain an abnormal event detection result;
Responding to the abnormal event detection result to determine that an abnormal event exists, and determining a corresponding key frame as an abnormal event frame;
and carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data.
2. The method for processing a visual image of a tunnel electromechanical construction according to claim 1, wherein the step of extracting a continuous sequence of frames from the video data and performing inter-frame difference analysis comprises:
Dividing the video data into a sequence of successive frames;
carrying out gray scale processing on every two continuous frames in the frame sequence;
Separating foreground objects from the two continuous frames based on a dynamic background model;
according to the foreground object of each two continuous frames, calculating to obtain corresponding pixel differences;
and obtaining an inter-frame difference analysis result according to the pixel difference corresponding to each two continuous frames.
3. The method for processing a visual image of tunnel electromechanical construction according to claim 2, wherein the step of detecting and marking the key frames and the non-key frames based on the machine learning algorithm according to the result of the inter-frame difference analysis comprises:
Obtaining pixel differences corresponding to each two continuous frames according to the inter-frame difference analysis result;
Respectively judging whether the pixel difference corresponding to each two continuous frames exceeds a preset threshold value, and if so, determining that the two continuous frames are candidate key frames; if not, marking the two continuous frames as non-key frames;
extracting the characteristics of the candidate key frames and generating corresponding characteristic vectors;
And inputting the feature vector corresponding to each candidate key frame into a pre-trained machine learning classification model, and determining and marking the key frames and the non-key frames according to the model output result.
4. The method for processing a visual image of tunnel electromechanical construction according to claim 1, further comprising a training step of the abnormal event detection model, the training step comprising:
acquiring a sample data set; the sample data set comprises a plurality of image data pre-marked with abnormal events or normal events;
preprocessing the sample data set, and dividing the sample data set into a training set and a testing set;
inputting the training set into a pre-constructed neural network model for training to obtain the abnormal event detection model;
and testing the abnormal event detection model based on the test set, and correcting parameters of the model to obtain the trained abnormal event detection model.
5. The method for processing a visual image of tunnel electromechanical construction according to claim 1, further comprising, after the step of determining that the corresponding key frame is an abnormal event frame:
And generating abnormal event prompt information according to the abnormal event frame and the corresponding time stamp, and sending the abnormal event prompt information to a maintenance terminal.
6. A tunnel electromechanical construction visualization image processing method as defined in any one of claims 1 to 5, further comprising:
Content classification is carried out according to the images corresponding to the key frames, and classification labels of the images are determined;
Establishing an index for the image corresponding to each key frame according to the classification label of the image;
and responding to the query element input by the user, determining an image set matched with the query element according to the index, and outputting the image set as a query result.
7.A visual image processing system for electromechanical construction of a tunnel, comprising:
The video data acquisition module is used for acquiring video data of the electromechanical construction area of the tunnel in real time;
the inter-frame difference analysis module is used for extracting continuous frame sequences according to the video data and carrying out inter-frame difference analysis;
The frame detection module is used for detecting and marking key frames and non-key frames based on a machine learning algorithm according to the inter-frame difference analysis result;
the abnormal event detection result generation module is used for respectively inputting each key frame into a pre-trained abnormal event detection model to obtain an abnormal event detection result;
the abnormal event frame determining module is used for determining a corresponding key frame as an abnormal event frame in response to the abnormal event detection result as the abnormal event;
And the video data storage module is used for carrying out data compression storage on non-key frames in the video data, and carrying out high-quality storage on the key frames and the abnormal event frames to obtain the processed video data.
8. The visual image processing system for tunnel boring machine construction of claim 7, further comprising:
The image classification module is used for classifying the content according to the image corresponding to the key frame and determining the classification label of the image;
The index establishing module is used for establishing an index for the image corresponding to each key frame according to the classification label of the image;
And the query result output module is used for responding to the query element input by the user, determining an image set matched with the query element according to the index and outputting the image set as a query result.
9. A computer device, characterized by: comprising a memory, a processor and a computer program stored on said memory and executable on said processor, said processor implementing the method according to any of claims 1 to 6 when said program is executed.
10. A computer-readable storage medium, characterized by: a computer program stored which can be loaded by a processor and which performs the method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410204540.5A CN118135480A (en) | 2024-02-24 | 2024-02-24 | Visual image processing method and system for electromechanical construction of tunnel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410204540.5A CN118135480A (en) | 2024-02-24 | 2024-02-24 | Visual image processing method and system for electromechanical construction of tunnel |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118135480A true CN118135480A (en) | 2024-06-04 |
Family
ID=91234791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410204540.5A Pending CN118135480A (en) | 2024-02-24 | 2024-02-24 | Visual image processing method and system for electromechanical construction of tunnel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118135480A (en) |
-
2024
- 2024-02-24 CN CN202410204540.5A patent/CN118135480A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101942808B1 (en) | Apparatus for CCTV Video Analytics Based on Object-Image Recognition DCNN | |
| CN112183313B (en) | SlowFast-based power operation field action identification method | |
| KR20200058260A (en) | Apparatus for CCTV Video Analytics Based on Object-Image Recognition DCNN and Driving Method Thereof | |
| CN111382623A (en) | Live broadcast auditing method, device, server and storage medium | |
| CN112163572A (en) | Method and device for identifying object | |
| CN112507860A (en) | Video annotation method, device, equipment and storage medium | |
| CN113642474A (en) | Hazardous area personnel monitoring method based on YOLOV5 | |
| CN110096945B (en) | Indoor monitoring video key frame real-time extraction method based on machine learning | |
| KR20190079047A (en) | A supporting system and method that assist partial inspections of suspicious objects in cctv video streams by using multi-level object recognition technology to reduce workload of human-eye based inspectors | |
| CN113128412A (en) | Fire trend prediction method based on deep learning and fire monitoring video | |
| CN112597928B (en) | Event detection method and related device | |
| CN114758271A (en) | Video processing method, device, computer equipment and storage medium | |
| CN112257545A (en) | Violation real-time monitoring and analyzing method and device and storage medium | |
| CN117197713A (en) | Extraction method based on digital video monitoring system | |
| CN117351271A (en) | Fault monitoring method and system for high-voltage distribution line monitoring equipment and storage medium thereof | |
| CN115035328A (en) | Converter image increment automatic machine learning system and establishment training method thereof | |
| CN114332071A (en) | Video anomaly detection method based on foreground information enhancement | |
| CN117272215B (en) | Intelligent community safety management method and system based on data mining | |
| CN115309871B (en) | Industrial big data processing method and system based on artificial intelligence algorithm | |
| CN118135480A (en) | Visual image processing method and system for electromechanical construction of tunnel | |
| CN115909144A (en) | Method and system for detecting abnormity of surveillance video based on counterstudy | |
| CN115131686A (en) | Intelligent power inspection method based on active learning and semi-supervised learning | |
| CN114373116A (en) | Method, device, equipment and product for screening operation and maintenance site images | |
| CN113518205A (en) | Video patrol processing method based on AI analysis | |
| CN112418055A (en) | Scheduling method based on video analysis and personnel trajectory tracking method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |