A construction site safety risk non-sensing inspection method and system

Abstract

The present application relates to the technical field of construction safety detection, in particular to a construction site safety risk non-inductive inspection method and system. The method comprises the following steps: reading historical inspection image data of a preset inspection route, and marking risks on the historical inspection image data; establishing a risk identification template according to the historical inspection image data with risk marks; receiving real-time inspection image data of the preset inspection route; comparing the real-time inspection image data with the risk identification template to obtain a risk identification result, effectively reducing the inspection operation process of construction personnel and reducing the complexity of inspection operation; sending the risk identification result to the person in charge, and converting the real-time inspection image data into a storage hash value, storing the risk identification result and the storage hash value, avoiding tampering with the inspection result by others, and improving the supervision effectiveness of the construction site safety inspection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of construction safety detection, and particularly relates to a construction site safety risk non-sensing inspection method and system. BACKGROUND

[0002] In the field of building engineering and some labor-intensive production and manufacturing fields, the coverage rate of mechanical automation is still low, and a large amount of manpower is still needed in the construction or production process, which leads to the need to pay close attention to the safety problems of the construction site. Once an injury or death event occurs, it will have a major negative impact on the enterprise and society. Therefore, construction units and production and manufacturing units have invested a large amount of manpower and resources to reduce and avoid the occurrence of safety accidents. Despite this, the cost of intelligent inspection robots without human participation is still too high, and not all construction sites of projects have the ability to equip intelligent inspection robots. Therefore, the effective method currently widely used in construction sites is to have construction personnel use intelligent equipment to implement safety inspection, regularly check the entire construction site, and discover possible safety hazards as soon as possible. At the same time, the safety hazards found in the inspection process are reported in the inspection information system, and the responsible person is notified to rectify in a timely manner.

[0003] However, in the field of building engineering and some labor-intensive production and manufacturing fields, the operation of the inspection information system is complex, the learning cost is high, and this may lead to invalid work of new inspectors in the process of safety inspection, resulting in waste of resources and failure to achieve the effect of inspection. Moreover, since the inspection information system may only record some key data but not complete inspection records, this may lead to incomplete inspection result information, thereby affecting the accuracy of the inspection result. Furthermore, in order to obtain some benefits or avoid punishment, it is also possible to tamper with the inspection records stored in the inspection information system, resulting in insufficient supervision effectiveness. SUMMARY

[0004] The technical problem to be solved by the present application is that the current construction site safety risk inspection has the technical problems of complex construction personnel inspection operation and poor supervision effectiveness of safety inspection. The present application provides a construction site safety risk non-sensing inspection method and system, which aims to solve the above technical problems.

[0005] The present application adopts the following technical scheme: a construction site safety risk non-sensing inspection method, comprising the following steps:

[0006] reading historical inspection image data of a preset inspection route, and performing risk labeling on the historical inspection image data;

[0007] establishing a risk identification template according to the historical inspection image data with risk labeling;

[0008] receiving real-time inspection image data of the preset inspection route;

[0009] The real-time inspection image data is compared with the risk identification template to obtain a risk identification result.

[0010] The risk identification result is sent to a responsible person, and the real-time inspection image data is converted into a storage hash value, and the risk identification result and the storage hash value are stored.

[0011] By establishing a risk identification template and comparing the risk identification template with new image data obtained in the inspection process to obtain a risk identification result, the construction personnel are assisted in identifying safety risks in the construction site, the accuracy and efficiency of safety inspection are improved, the inspection operation process of the construction personnel is greatly reduced, and the complexity of the inspection operation is reduced. At the same time, by converting the real-time inspection image data into a storage hash value, storing the risk identification result and the storage hash value, the inspection result is prevented from being tampered with by others, and the effectiveness of the supervision of the safety inspection of the construction site is improved.

[0012] Preferably, the preset inspection route is obtained by the following method:

[0013] Receiving static image data of a construction site monitoring area;

[0014] Performing inspection area division operation and inspection point labeling operation according to the static image data;

[0015] Connecting the labeled inspection points in combination with the divided inspection areas to obtain a preset inspection route.

[0016] Preferably, the method for establishing the risk identification template comprises:

[0017] Classifying historical inspection image data containing risk labels, listing different categories of risk items, and recording historical inspection image data and inspection area position information corresponding to all categories of risk items;

[0018] For different categories of risk items, the pixel mean of the image data of each category of risk item is calculated as an abnormal template.

[0019] All categories of abnormal templates constitute a risk identification template.

[0020] Preferably, the method for comparing the real-time inspection image data with the risk identification template to obtain a risk identification result comprises:

[0021] The real-time inspection image data is sequentially corresponding to different categories of risk items;

[0022] The pixel mean of the real-time inspection image data of each category is calculated.

[0023] The quotient of the average pixel value of the real-time inspection image data of each category and the abnormal template of the corresponding category in the risk identification template is recorded as the abnormal similarity.

[0024] If the abnormal similarity is greater than or equal to the first threshold, the risk identification result is that the area where the real-time inspection image data of the current category is located is at risk; if the abnormal similarity is less than the first threshold, the risk identification result is that the area where the real-time inspection image data of the current category is located is safe.

[0025] Preferably, the real-time inspection image data includes inspection positioning coordinates, and the method for sequentially mapping the real-time inspection image data to different categories of risk items includes:

[0026] Convert the location information of the inspection areas corresponding to all categories of risk items into geographic coordinates;

[0027] The inspection location coordinates are matched with the geographical coordinates of the inspection area corresponding to different categories of risk items to obtain the risk item category corresponding to the real-time inspection image data.

[0028] Preferably, the method for converting the real-time inspection image data into a storage hash value includes:

[0029] Inspection terminal captures n i After inspecting the real-time images, extract n. i The hash value of the real-time inspection image frame, denoted as Hi;

[0030] In the n i A marker frame is inserted after the real-time inspection image is captured; the marker frame is used to record the extracted Hi.

[0031] Send the last m bits of Hi to the designated server, where m is an integer constant. The server will then respond with n based on the last m bits of Hi. i+1 To the inspection terminal;

[0032] Repeat the above steps until the inspection terminal has completed all the shooting, then extract the hash values ​​of all frames and record them as the evidence hash values.

[0033] As a preferred approach, n is fed back based on the last m bits of Hi. i+1 The method is as follows:

[0034] n i =min[max(floor(10)] k ·sin(x)), N1), N2]

[0035] Where, n i The initial value n1 is a preset value, x is the value of the last m digits of Hi, and k, N1 and N2 are constant values.

[0036] Preferably, the method for sending the risk identification result to the responsible person and converting the real-time inspection image data into a storage hash value, and storing the risk identification result and the storage hash value, further includes:

[0037] After obtaining the risk identification results, the real-time inspection image data of the preset inspection route, all categories of risk items, and the risk identification results corresponding to each risk item are used to form an inspection result report. The inspection result report is sent to the person in charge, and the real-time inspection image data is converted into a storage hash value. The inspection result report and the storage hash value are then uploaded to a designated server for storage.

[0038] A non-intrusive safety risk inspection system for construction sites includes:

[0039] The inspection terminal is used to read historical inspection image data of the preset inspection route and receive real-time inspection image data of the preset inspection route.

[0040] The data analysis module is used to perform risk labeling on the historical inspection image data and to establish a risk identification template based on the risk-labeled historical inspection image data.

[0041] The risk identification module is used to compare the real-time inspection image data with the risk identification template to obtain the risk identification result;

[0042] The data storage module is used to convert the real-time inspection image data into storage hash values;

[0043] The server is used to send the risk identification results to the responsible person and to store the risk identification results and the evidence hash value.

[0044] The beneficial technical effects of this invention include: It employs a non-intrusive inspection method and system for construction site safety risks. By establishing a risk identification template and comparing it with new image data obtained during the inspection process, the risk identification result is obtained. This assists construction personnel in identifying safety risks at the construction site, improving the accuracy and efficiency of safety inspections. Furthermore, it significantly reduces the inspection process for construction personnel, lowering the complexity of inspection operations. By converting real-time inspection image data into a storage hash value, and storing both the risk identification result and the storage hash value, it prevents unauthorized tampering of the inspection results, thus improving the effectiveness of on-site safety inspection supervision. By using the pixel average of the image data as a comparison feature, the risk identification result is obtained. Obtaining risk identification results can significantly reduce the system's computational load and improve the efficiency of image comparison and recognition, thereby improving the efficiency of safety inspections at construction sites. The use of geographic coordinate matching technology to classify and identify risk items in real-time inspection image data replaces the existing use of neural network models or machine learning methods for classification and identification, greatly reducing the system's computational complexity and thus improving the efficiency of obtaining risk identification results to some extent. By automatically generating inspection result reports from the obtained risk identification results, the manual filling out of safety inspection records by construction personnel is replaced, avoiding the influence of subjective human factors, improving the accuracy of inspection result report filling, and reducing the complexity of inspection operations for construction personnel.

[0045] Other features and advantages of the present invention will be disclosed in detail in the following detailed description and accompanying drawings. Attached Figure Description

[0046] The invention will be further described below with reference to the accompanying drawings:

[0047] Figure 1 This is a flowchart of a method for non-intrusive inspection of safety risks at construction sites, according to an embodiment of the present invention.

[0048] Figure 2 A flowchart illustrating the method for obtaining a preset inspection route in an embodiment of the invention.

[0049] Figure 3 A flowchart illustrating the method for establishing a risk identification template in an embodiment of the present invention.

[0050] Figure 4 This is a flowchart of a method for obtaining risk identification results according to an embodiment of the present invention.

[0051] Figure 5 This is a flowchart illustrating a method for converting real-time inspection image data into evidence hash values ​​according to an embodiment of the present invention.

[0052] Figure 6 This is a schematic diagram of a construction site safety risk non-intrusive inspection system according to an embodiment of the present invention.

[0053] The components include: 1. Inspection terminal, 2. Data analysis module, 3. Risk identification module, 4. Data storage module, and 5. Server. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be explained and described below with reference to the accompanying drawings. However, the following embodiments are only preferred embodiments of the present invention and not all of them. Other embodiments obtained by those skilled in the art based on the embodiments in the implementation methods without creative effort are all within the protection scope of the present invention.

[0055] In the following description, terms such as “inner,” “outer,” “upper,” “lower,” “left,” and “right” are used only to indicate orientation or positional relationship for the convenience of describing the embodiments and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention.

[0056] Before describing the technical solution of this embodiment in detail, the background of the application of this embodiment will be introduced first.

[0057] Although the construction industry has become a fundamental industry and an important pillar of economic development, it has also brought about many socio-economic problems, the most prominent of which is construction safety. It is well known that construction site workers suffer from poor safety awareness, high turnover, and low overall quality; the working environment is complex, labor intensity is high, and there are many high-risk projects; safety management concepts are lacking, and there is a shortage of safety management personnel, especially with outdated safety management technologies, methods, and tools. Due to these characteristics and unfavorable factors, the construction industry has become one of the three most dangerous industries in my country.

[0058] Currently, in the construction engineering sector and some labor-intensive manufacturing industries, the coverage rate of mechanization and automation remains relatively low. Significant manpower is still required during construction or production, making it even more crucial to prioritize safety at construction sites. Injuries or fatalities can result in substantial losses for construction workers and companies, and can also have negative consequences. Therefore, it is necessary to conduct in-depth research into the mechanisms of construction site safety accidents, carefully investigate the causes of potential hazards, and apply modern science and technology to implement targeted management methods and technical preventative measures to reduce the probability of construction site safety accidents.

[0059] While artificial intelligence (AI) technology is driving the construction industry towards automation, digitalization, and intelligence, the complexities of construction sites and the lack of a systematic understanding of potential hazards in various construction scenarios make it difficult to utilize new technologies for hazard identification. Furthermore, the cost of fully automated, human-operated intelligent inspection robots remains too high, and not all construction sites can afford to equip them. Therefore, the currently prevalent and effective method is for construction workers to conduct safety inspections using intelligent equipment, regularly checking the entire construction site to identify potential safety hazards early. These hazards are then reported in the inspection information system, and the responsible parties are notified to rectify them promptly.

[0060] However, in the construction engineering field and some labor-intensive manufacturing sectors, the high complexity and learning cost of inspection information systems can lead to ineffective work by new inspectors during safety inspections, resulting in wasted resources and a failure to achieve the desired inspection effect. Furthermore, since inspection information systems may only record certain key data rather than complete inspection records, the inspection results may be incomplete, affecting their accuracy. Even worse, to gain certain benefits or avoid punishment, inspectors may tamper with the inspection records stored in the system, leading to insufficient regulatory effectiveness. Therefore, it is necessary to research a method and system for on-site safety risk inspections that can reduce the complexity of inspection operations for construction personnel, improve the efficiency of safety inspections, and enhance the effectiveness of supervision.

[0061] Therefore, this application provides a method for non-intrusive inspection of safety risks at construction sites. Please refer to the attached document. Figure 1 This includes the following steps:

[0062] Step A01) Read the historical inspection image data of the preset inspection route and mark the risks in the historical inspection image data.

[0063] The operation of risk labeling for historical inspection image data is similar to the operation of risk labeling for data in the prior art, and this application embodiment does not limit it. Optionally, the implementation method of risk labeling for historical inspection image data can be that experienced inspection personnel manually conduct risk assessment and labeling of historical inspection image data, or it can be implemented by using technologies such as machine learning and artificial intelligence.

[0064] Step A02) Establish a risk identification template based on historical inspection image data with risk labels.

[0065] Step A03) Receive real-time inspection image data of the preset inspection route.

[0066] Optionally, the methods for obtaining real-time inspection image data of the preset inspection route can be as follows: (1) Using drones for aerial photography can obtain relatively comprehensive on-site image data, including the inspection route and surrounding environment. This method is simple to operate, but requires professional personnel to operate and is costly; (2) Construction personnel use handheld inspection terminals to conduct inspections, and use the image acquisition module of the inspection terminal to obtain image data of the inspection route at the construction site and perform motion detection. The image acquisition module of the inspection terminal includes a camera. (3) Installing cameras at the construction site can monitor the inspection route in real time and record the image data of the inspection route. This method can automatically record the image data of the inspection route, but the issues of data confidentiality and security need to be considered. (4) Using GPS and map technology to record the location information and image data of the inspection route in real time, and can also perform data analysis and visualization. This method requires the use of professional GPS equipment and map software and is costly.

[0067] Preferably, in this embodiment, the real-time inspection image data of the preset inspection route is acquired by the image acquisition module of the inspection terminal. The image acquisition module of the inspection terminal includes a camera. During the inspection process, when the construction personnel use the handheld inspection terminal, the image acquisition module of the inspection terminal can acquire the image data of the inspection route at the construction site in real time and perform motion detection, providing data for subsequent comparison with the risk identification template.

[0068] Step A04) Compare the real-time inspection image data with the risk identification template to obtain the risk identification result.

[0069] Step A05) Send the risk identification results to the responsible person and convert the real-time inspection image data into a storage hash value, and store the risk identification results and the storage hash value.

[0070] This embodiment establishes a risk identification template and obtains risk identification results by comparing the risk identification template with new image data obtained during the inspection process. This assists construction personnel in identifying safety risks at the construction site, improving the accuracy and efficiency of safety inspections. It also significantly reduces the inspection process for construction personnel and lowers the complexity of inspection operations. Furthermore, by converting real-time inspection image data into evidence hash values ​​and storing the risk identification results and evidence hash values, it prevents others from tampering with the inspection results and improves the effectiveness of safety inspection supervision at the construction site.

[0071] On the other hand, in this embodiment, please refer to the appendix. Figure 2 The preset inspection route is obtained by the following methods:

[0072] Step B01) Receive static image data of the construction site monitoring area.

[0073] Step B02) Perform the inspection area division and inspection point marking operations based on the static image data.

[0074] Furthermore, the inspection area division operation and inspection point marking operation specifically refer to specifying the inspection area and inspection point on the static image data of the construction site monitoring area obtained through the camera device, and obtaining the coordinate dataset of the corresponding inspection area and inspection point.

[0075] Step B03) Connect the marked inspection points with the divided inspection areas to obtain the preset inspection route.

[0076] The inspection area refers to the area where inspection personnel must carry out inspection work during the inspection process, and the inspection point refers to the equipment or device that inspection personnel must carry out inspection work during the inspection process.

[0077] On the other hand, in this embodiment, please refer to the appendix. Figure 3 Methods for establishing risk identification templates include:

[0078] Step C01) Classify the historical inspection image data containing risk labels, list the risk items of different categories, and record the historical inspection image data and inspection area location information corresponding to all categories of risk items.

[0079] For example, the categories of risk items include: falls from heights, electrical accidents, slippage of construction materials, tower crane collapse, improper enclosure of construction areas, dust pollution, and building structural problems.

[0080] Step C02) For different categories of risk items, calculate the pixel mean of the image data for each category of risk item as an anomaly template.

[0081] Specifically, the method for calculating the pixel mean of image data for each category of risk items is as follows:

[0082] (1) Read image data; use the corresponding functions in the image processing library or programming language to read image data, usually the pixel matrix of the image;

[0083] (2) Calculate the pixel mean: For RGB images, the pixel mean needs to be calculated for each channel (red, green, blue). The formula for calculating the pixel mean is: pixel mean = sum of pixel values ​​ / total number of pixels, where the sum of pixel values ​​refers to the result of adding up all pixel values, and the total number of pixels refers to the number of pixels in the image;

[0084] (3) Calculate the mean of the three RGB channels; sum up the pixel mean of each channel, and the resulting mean of the three RGB channels is the pixel mean of the image data.

[0085] Step C03) All categories of anomaly templates constitute a risk identification template.

[0086] On the other hand, in this embodiment, please refer to the appendix. Figure 4 Methods for obtaining risk identification results include:

[0087] Step D01) The real-time inspection image data is sequentially mapped to different categories of risk items.

[0088] Step D02) Calculate the pixel mean of the real-time inspection image data for each category.

[0089] The method for calculating the pixel mean of real-time inspection image data for each category is the same as the method for calculating the pixel mean of image data for each category of risk items described above, and will not be repeated here.

[0090] Step D03) Calculate the quotient of the pixel mean of the new image data for each category and the mean of the abnormal template of the corresponding category in the risk identification template, and record it as the abnormal similarity.

[0091] Step D04) If the abnormal similarity is greater than or equal to the first threshold, the risk identification result is that the area where the new image data of the current category is located is risky; if the abnormal similarity is less than the first threshold, the risk identification result is that the area where the new image data of the current category is located is safe.

[0092] The method for establishing a risk identification template and obtaining risk identification results in this embodiment of the application is based on the pixel average of image data as a comparison feature, which can significantly reduce the computational load of the system, improve the efficiency of image comparison and recognition, and thus improve the efficiency of safety inspection at the construction site.

[0093] On the other hand, in this embodiment, the real-time inspection image data includes inspection positioning coordinates, and the method for sequentially mapping the real-time inspection image data to different categories of risk items includes:

[0094] Convert the location information of the inspection areas corresponding to all categories of risk items into geographic coordinates;

[0095] By matching the inspection location coordinates with the geographical coordinates of the inspection area corresponding to different categories of risk items, the risk item categories corresponding to the real-time inspection image data are obtained.

[0096] The specific implementation method for matching the inspection positioning coordinates with the geographical coordinates of the inspection area corresponding to different categories of risk items is as follows: use a distance formula to calculate the distance between the inspection positioning coordinates and the geographical coordinates of the inspection area corresponding to each category of risk item, such as Euclidean distance, Manhattan distance, etc., and then by comparing the distances, it can be determined which category of risk item the inspection positioning coordinates belong to and the corresponding inspection area.

[0097] This embodiment uses a geographic coordinate matching technique to classify and identify risk items in real-time inspection image data, replacing the existing technology that uses neural network models or machine learning methods for classification and identification. This significantly reduces the computational complexity of the system and thus improves the efficiency of obtaining risk identification results to a certain extent.

[0098] On the other hand, in this embodiment, please refer to the appendix. Figure 5 Methods for converting real-time inspection image data into evidence hash values ​​include:

[0099] Step E01) Inspection terminal takes pictures n i After inspecting the real-time images, extract n. i The hash value of the real-time inspection image frame, denoted as H. i .

[0100] Step E02) in n i A marker frame is inserted after the real-time inspection of the image. The marker frame is used to record the extracted H. i .

[0101] Step E03) H i The last m bits are sent to the specified server, where m is an integer constant. The server determines the value based on H. i The last m bits of feedback n i+1 To the inspection terminal.

[0102] In this context, a hash value is a fixed-length number, such as the SHA256 hash algorithm which uses a hash value of 256 bits.

[0103] Step E04) Repeat the above steps until the inspection terminal has completed all the shooting, then extract the hash value of all frames and record it as the evidence hash value.

[0104] The total frames include all image frames and all marker frames.

[0105] For example, the implementation process of converting real-time inspection image data into evidence hash values ​​is as follows:

[0106] After the inspection terminal captures 10 frames of real-time inspection images, the hash value of the 10 frames is extracted and denoted as H. i ;

[0107] A marker frame is inserted after 10 real-time inspection images. The marker frame is used to record the hash value H extracted from the 10 real-time inspection images. i ;

[0108] H i The last 3 digits of the value are sent to the specified server, and the server determines the value based on the H value. iThe last 3 digits of the value are used to calculate the number of real-time inspection image frames n that the inspection terminal will capture and extract the hash value for next time. i+1 and n i+1 Feedback is sent to the inspection terminal;

[0109] Repeat the above steps until the inspection terminal has completed all the images taken, that is, after the inspection personnel have walked the inspection route with the inspection terminal and taken all the real-time inspection images. Extract the hash value again from all the real-time inspection image frames and marker frames, and record the hash value extracted at this time as the evidence hash value.

[0110] Unlike encryption algorithms, hash algorithms are irreversible one-way functions. When using highly secure hash algorithms such as MD5 and SHA, it is almost impossible for two different data to produce the same hash value. Therefore, once the data is modified, it can be detected.

[0111] Safety inspections, as one of the most frequently involved management activities in safety management, are specifically divided into three parts: routine inspections by grassroots project departments, inspections by the group headquarters, and government inspections. Government inspections are government actions and unrelated to construction companies, so they are not considered in this application. Regarding inspections by the group headquarters, due to the headquarters' lack of in-depth understanding of the projects, it is difficult to complete the inspection work efficiently, and risks are easily overlooked. Especially for inspections of the group headquarters, since the frequency of headquarters inspections is low, grassroots project departments may report routine inspection results through perfunctory presentations, thereby reducing the effectiveness of the group headquarters' supervision. Furthermore, since the inspection information system may only record some key data, rather than complete inspection records, this leads to incomplete inspection information, affecting the accuracy of the inspection results. Even worse, to gain certain benefits or avoid punishment, there is a possibility that the inspection records stored in the inspection information system may be tampered with, resulting in insufficient supervisory effectiveness. Therefore, in this embodiment of the application, real-time inspection image data is stored by extracting evidence hash values. For example, the extracted evidence hash values ​​are uploaded to the headquarters server to prevent grassroots project departments from tampering with the inspection results and to improve the effectiveness of on-site safety inspection supervision.

[0112] On the other hand, in this embodiment, according to H i The last m bits of feedback n i+1 The method is as follows:

[0113] n i =min[max(floor(10)] k ·sin(x)), N1), N2]

[0114] Where x is H i The last m positions are: k is the constant exponent coefficient, N1 is the minimum constant threshold, and N2 is the maximum constant threshold.

[0115] For example, according to H i The last m bits of feedback n i+1 The implementation process is as follows: Assume that the value of k is 2, the value of N1 is 10, the value of N2 is 200, and H... i The last three digits of x are 136. Substituting x = 136 into the function above, we get...

[0116] n i =min[max(floor(10)] 2 ·sin(136)),10),200]=69,

[0117] The next real-time inspection image frame n to be captured and hash value extracted by the inspection terminal. i+1 It has 69 frames.

[0118] On the other hand, in this embodiment, the method of sending the risk identification results to the responsible person and converting the real-time inspection image data into a storage hash value, and storing the risk identification results and the storage hash value, further includes:

[0119] After obtaining the risk identification results, the real-time inspection image data of the preset inspection route, all categories of risk items, and the risk identification results corresponding to each risk item are used to form an inspection result report. The inspection result report is sent to the person in charge, and the real-time inspection image data is converted into a storage hash value. The inspection result report and the storage hash value are then uploaded to the designated server for storage.

[0120] Because current inspection information systems still require manual intervention for filling out safety inspection records, subjective human factors can lead to information distortion, and it is difficult to quickly identify and assess environmental and equipment hazards at construction sites, resulting in some serious hazards not being detected in a timely manner. Therefore, this application's embodiment automatically generates an inspection result report based on real-time inspection image data of a preset inspection route obtained by the aforementioned method, all categories of risk items, and the risk identification results corresponding to each risk item. This replaces the manual filling out of safety inspection records by construction personnel, avoids the influence of subjective human factors, improves the accuracy of inspection result report filling, and reduces the complexity of inspection operations for construction personnel.

[0121] On the other hand, this application also provides a non-intrusive inspection system for safety risks at construction sites. Please refer to the appendix. Figure 6 ,include:

[0122] Inspection terminal 1 is used to read historical inspection image data of the preset inspection route and receive real-time inspection image data of the preset inspection route.

[0123] Data analysis module 2 is used to label historical inspection image data with risks and to establish risk identification templates based on the historical inspection image data with risk labels.

[0124] Risk identification module 3 is used to compare real-time inspection image data with risk identification templates to obtain risk identification results;

[0125] Data storage module 4 is used to convert real-time inspection image data into storage hash values;

[0126] Server 5 is used to send the risk identification results to the responsible person and to store the risk identification results and the evidence hash value.

[0127] The inspection terminal 1 includes, but is not limited to, cameras, smartphones with cameras, and dedicated smart handheld terminal devices. Dedicated smart handheld terminal devices include, for example, the EM-T695 ruggedized handheld terminal. The server 5 includes, but is not limited to, a headquarters server.

[0128] On the other hand, in this embodiment, the data analysis module 2 is used to perform the following steps:

[0129] Classify the historical inspection image data containing risk labels, list different categories of risk items, and record the historical inspection image data and inspection area location information corresponding to all categories of risk items.

[0130] For different categories of risk items, the average pixel value of the image data for each category of risk item is calculated as an anomaly template;

[0131] All categories of anomaly templates constitute a risk identification template.

[0132] On the other hand, in this embodiment, the risk identification module 3 is used to perform the following steps:

[0133] The real-time inspection image data is sequentially mapped to different categories of risk items;

[0134] Calculate the pixel mean of the real-time inspection image data for each category;

[0135] The quotient of the average pixel value of the real-time inspection image data of each category and the abnormal template of the corresponding category in the risk identification template is recorded as the abnormal similarity.

[0136] If the abnormal similarity is greater than or equal to the first threshold, the risk identification result is that the area where the real-time inspection image data of the current category is located is at risk; if the abnormal similarity is less than the first threshold, the risk identification result is that the area where the real-time inspection image data of the current category is located is safe.

[0137] On the other hand, in this embodiment, the data storage module 4 is used to perform the following steps:

[0138] Inspection terminal captures n i After inspecting the real-time images, extract n. i The hash value of the real-time inspection image frame, denoted as H. i ;

[0139] In n i A marker frame is inserted after the real-time inspection of the image. The marker frame is used to record the extracted H. i ;

[0140] H i The last m bits are sent to the specified server, where m is an integer constant. The server determines the value based on H. i The last m bits of feedback n i+1 To the inspection terminal;

[0141] Repeat the above steps until the inspection terminal has completed all the shooting, then extract the hash values ​​of all frames and record them as the evidence hash values.

[0142] On the other hand, in this embodiment, server 5 is also used to perform the following steps:

[0143] After obtaining the risk identification results, the real-time inspection image data of the preset inspection route, all categories of risk items, and the risk identification results corresponding to each risk item are used to form an inspection result report. The inspection result report is sent to the person in charge, and the real-time inspection image data is converted into a storage hash value. The inspection result report and the storage hash value are then uploaded to the designated server for storage.

[0144] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Those skilled in the art should understand that the present invention includes, but is not limited to, the contents described in the accompanying drawings and the specific embodiments above. Any modifications that do not depart from the functional and structural principles of the present invention will be included within the scope of the claims.

Claims

1. A method for non-intrusive inspection of safety risks at construction sites, characterized in that, Includes the following steps: Read historical inspection image data of the preset inspection route and mark the historical inspection image data for risks; Establish a risk identification template based on historical inspection image data with risk labels; Receive real-time inspection image data of the preset inspection route; The real-time inspection image data is compared with the risk identification template to obtain the risk identification result; The risk identification result is sent to the responsible person, and the real-time inspection image data is converted into a storage hash value. The risk identification result and the storage hash value are then stored. The method for converting the real-time inspection image data into a storage hash value includes: Inspection terminal photography After real-time inspection of the frame, extract The hash value of the real-time inspection image of the frame is denoted as... , initial value This is the default value; In the A marker frame is inserted after the real-time inspection image; the marker frame is used to record the extracted images. ; Will The last m bits are sent to the designated server, where m is an integer constant. The server then... The last m bits of the feedback inspection terminal will be used to capture the next real-time inspection image frame from which the hash value will be extracted. To the inspection terminal; Repeat the above steps until the inspection terminal has completed all the shooting, then extract the hash value of all frames and record it as the evidence storage hash value. Among them, according to The last m bits of the feedback inspection terminal will be used to capture the next real-time inspection image frame from which the hash value will be extracted. The method is as follows: ， Where x is The last m digits of the value, where k, N1, and N2 are constant values.

2. The non-intrusive inspection method for safety risks at construction sites as described in claim 1, characterized in that, The preset inspection route is obtained by the following method: Receive static image data of the construction site monitoring area; The inspection area is divided and the inspection point is marked based on the static image data. By connecting the marked inspection points with the divided inspection areas, a preset inspection route is obtained.

3. The non-intrusive inspection method for safety risks at construction sites as described in claim 1, characterized in that, The method for establishing the risk identification template includes: Classify the historical inspection image data containing risk labels, list different categories of risk items, and record the historical inspection image data and inspection area location information corresponding to all categories of risk items. For different categories of risk items, the average pixel value of the image data for each category of risk item is calculated as an anomaly template; All categories of anomaly templates constitute a risk identification template.

4. The non-intrusive inspection method for safety risks at construction sites as described in claim 3, characterized in that, The method for comparing the real-time inspection image data with the risk identification template to obtain the risk identification result is as follows: The real-time inspection image data is sequentially mapped to different categories of risk items; Calculate the pixel mean of the real-time inspection image data for each category; The quotient of the average pixel value of the real-time inspection image data of each category and the abnormal template of the corresponding category in the risk identification template is recorded as the abnormal similarity. If the abnormal similarity is greater than or equal to the first threshold, the risk identification result is that the area where the real-time inspection image data of the current category is located is at risk; if the abnormal similarity is less than the first threshold, the risk identification result is that the area where the real-time inspection image data of the current category is located is safe.

5. A non-intrusive inspection method for safety risks at construction sites as described in claim 4, characterized in that, The real-time inspection image data includes inspection positioning coordinates, and the method for sequentially mapping the real-time inspection image data to different categories of risk items includes: Convert the location information of the inspection areas corresponding to all categories of risk items into geographic coordinates; The inspection location coordinates are matched with the geographical coordinates of the inspection area corresponding to different categories of risk items to obtain the risk item category corresponding to the real-time inspection image data.

6. The non-intrusive inspection method for safety risks at construction sites as described in claim 4, characterized in that, The method of sending the risk identification result to the responsible person and converting the real-time inspection image data into a storage hash value, and storing the risk identification result and the storage hash value, further includes: After obtaining the risk identification results, the real-time inspection image data of the preset inspection route, all categories of risk items, and the risk identification results corresponding to each risk item are used to form an inspection result report. The inspection result report is sent to the person in charge, and the real-time inspection image data is converted into a storage hash value. The inspection result report and the storage hash value are then uploaded to a designated server for storage.

7. A non-intrusive inspection system for safety risks at construction sites, characterized in that, include: The inspection terminal is used to read historical inspection image data of the preset inspection route and receive real-time inspection image data of the preset inspection route. The data analysis module is used to perform risk labeling on the historical inspection image data and to establish a risk identification template based on the risk-labeled historical inspection image data. The risk identification module is used to compare the real-time inspection image data with the risk identification template to obtain the risk identification result; The data storage module is used to convert the real-time inspection image data into storage hash values; The server is used to send the risk identification results to the responsible person and to store the risk identification results and the evidence hash value. The data storage module is used to perform the following steps: Inspection terminal photography After real-time inspection of the frame, extract The hash value of the real-time inspection image of the frame is denoted as... , initial value This is the default value; In the A marker frame is inserted after the real-time inspection image; the marker frame is used to record the extracted images. ; Will The last m bits are sent to the designated server, where m is an integer constant. The server then... The last m bits of the feedback inspection terminal will be used to capture the next real-time inspection image frame from which the hash value will be extracted. To the inspection terminal; Repeat the above steps until the inspection terminal has completed all the shooting, then extract the hash value of all frames and record it as the evidence storage hash value. Among them, according to The last m bits of the feedback inspection terminal will be used to capture the next real-time inspection image frame from which the hash value will be extracted. The method is as follows: ， Where x is The last m digits of the value, where k, N1, and N2 are constant values.

Images