A special inspection recorder for safety and emergency response

Abstract

This invention provides a dedicated safety emergency inspection recorder, including a stabilizer, a recorder body, a speed sensor, a local processor, and a cloud server. The recorder body is fixed to a fixed position on the user's body via the stabilizer. Both the speed sensor and the recorder body are connected to the local processor, which is remotely connected to the cloud server. The local processor determines the on / off state of video stream storage based on the speed status collected by the speed sensor and the degree of change in the video stream captured by the recorder body over time. This technical solution uses directional movements as a video stream storage activation signal, effectively controlling the local storage size and reducing useless information, thus ensuring a high proportion of effective video stream storage.

Description

Technical Field

[0001] This invention relates to the field of image communication technology, and in particular to a special inspection recorder for safety emergency response. Background Technology

[0002] An inspection recorder is an instrument used to record the work of inspection personnel. It can record the time when inspection personnel arrive at a certain location and automatically upload it to the database of management software to achieve scientific management of inspection work.

[0003] Inspection recorders are widely used in many fields, such as: refueling stations, where staff use them to better conduct inspections; and inspections of general safety hazards, which often require on-site recording based on personnel data collection, also need to be based on inspection systems.

[0004] However, some common inspection recorders still have certain shortcomings in use. To achieve portability, these recorders are often small in size, resulting in limited storage space. Meanwhile, with the increasing sophistication of video recording technology, video files are often quite large. In actual use, however, the video containing truly useful information typically accounts for only about 20% of the total video stream. Using pre-defined recognition methods, based on the characteristics of machine learning, often requires clearly defining the situations to be identified in the training set. However, given the unforeseen circumstances that may arise during inspections, where inspectors often need to make on-site judgments, this method can lead to missed inspection information.

[0005] Therefore, it is necessary to provide a dedicated safety emergency inspection recorder to solve the above-mentioned technical problems. Summary of the Invention

[0006] This invention addresses the problem that in existing technologies, in actual use, videos containing valid information often account for only 20% of the total video stream. While pre-defined recognition methods, based on machine learning, often require specific information to be identified in the training set, this approach can lead to missed information due to unforeseen circumstances during inspections and the need for inspectors to make on-site judgments. The invention provides a dedicated safety emergency inspection recorder that uses user actions as the basis for video acquisition and judgment, thus solving the aforementioned problems.

[0007] This invention provides a dedicated safety emergency inspection recorder, including a stabilizer, a recorder body, a speed sensor, a local processor, and a cloud server. The recorder body is fixed to a fixed position on the user's body by the stabilizer. The speed sensor and the recorder body are both connected to the local processor. The local processor and the cloud server are remotely connected. The local processor determines the on / off state of the recorder body's video stream storage based on the speed status collected by the speed sensor and the degree of change of the video stream captured by the recorder body over time.

[0008] By filtering and processing videos locally, the pressure on cloud video stream processing is reduced, thus avoiding system crashes caused by stress.

[0009] The safety emergency inspection recorder of the present invention, in a preferred embodiment, includes a local processor comprising a local memory, a local buffer, a video stream processor, a mode switch, and a transmission module. The local memory is used to temporarily store recorded videos locally and transmit them to a cloud server in a quantitative manner through the transmission module. The local buffer is used to pre-store video stream segments to be transmitted to the local memory and to transfer them to the local memory upon confirmation by the video stream processor. The mode switch is used by the user to switch between a normal storage mode and an automatic storage mode. In the normal storage mode, the local memory stores all video streams recorded when the recorder is powered on. In the automatic storage mode, the recorded video stream is stored when the speed sensor detects a slowdown and the video stream's continuous frame change rate is less than a first threshold, until the video stream storage is turned off when the above conditions are no longer met. The video stream processor is used to identify the video stream and speed status in the automatic storage mode.

[0010] The safety emergency patrol recorder of the present invention, in a preferred embodiment, includes a video stream processor comprising a speed analysis module, a timer, a selection module, an image analysis module, and an output module. The timer is a counter that accurately times the data based on the frequency and duty cycle of the input pulse signal. The speed analysis module records speed information and obtains the user's normal movement speed based on historical information. It then determines whether the user is in a deceleration state based on the normal movement speed and transmits the real-time deceleration state signal to the selection module. The selection module continuously receives video streams and, upon receiving a deceleration state signal, transmits the video stream at the corresponding position to the image analysis module. The image analysis module performs image change judgment on the received video streams selected by the selection module and transmits the portion of the video stream below a first threshold to the output module. The output module adds a timestamp to the received video streams based on the pulse signal of the timer and outputs the value to the local memory.

[0011] The safety emergency inspection recorder of this invention, as a preferred embodiment, uses a video stream processor to select video as follows:

[0012] S1. The speed analysis module takes the speed with the longest duration after noise reduction from the historical speed of the speed sensor as the normal moving speed.

[0013] S2. The speed analysis module continuously receives speed information from the speed sensor and uses the signal that is continuously lower than the normal moving speed received by the speed analysis module as a deceleration status signal, and sends the deceleration status signal and the corresponding start time stamp to the selection module.

[0014] S3. The selection module transmits the video stream segments temporarily stored in the local buffer to the local storage module and simultaneously transmits them to the image analysis module.

[0015] S4. The video analysis module extracts frames from the video stream after the starting timestamp.

[0016] S5. Determine whether the proportion of pixels at the same position in each frame of the extracted video stream is lower than the threshold. If yes, proceed to step S6; otherwise, delete the video stream after the starting timestamp in the local storage.

[0017] S6. Determine whether there is a case where the translation speed and deformation ratio of adjacent frames above the threshold are less than or equal to the translation speed and deformation ratio under normal movement speed. If so, continue to send the video stream to the local storage until the slowdown state ends. Otherwise, delete the video stream after the starting timestamp from the local storage.

[0018] This technical solution, based on people's observation habits, provides a supplementary method for identifying inspected objects, building upon machine learning's video target recognition capabilities. Unlike active object recognition technologies, this technology relies on the collector's habits to determine whether valid information might exist in the captured footage, further enhancing the reliability of information recognition in actual video streams when machine learning is still developing. It can serve as a reasonable recording scheme before the inspection system completes machine learning for identifying inspected objects, and it also helps machine learning obtain a larger data library for more accurate training results.

[0019] In the safety emergency inspection recorder described in this invention, the stabilizer is preferably a six-axis stabilizer.

[0020] The beneficial effects of this invention are as follows:

[0021] This technical solution fully considers the user's habits and identifies information based on human experience, using targeted rather than goal-oriented data collection. It can supplement important inspection information missed at the program design level.

[0022] Based on the above principles, this technical solution uses a directional action as a signal to initiate video stream storage, effectively controlling the local storage size and reducing useless information. This ensures a high percentage of effective video stream storage. Attached Figure Description

[0023] Figure 1 This is a schematic diagram of a special inspection recorder for safety and emergency response.

[0024] Figure 2 A schematic diagram of the local processor architecture for a dedicated safety emergency inspection recorder;

[0025] Figure 3 A schematic diagram of a video stream processor architecture for a dedicated safety emergency inspection recorder;

[0026] Figure 4 A flowchart for selecting video feeds for a dedicated safety emergency inspection recorder.

[0027] Figure label:

[0028] 1. Stabilizer; 2. Recorder body; 3. Speed ​​sensor; 4. Local processor; 41. Local memory; 42. Local buffer; 43. Video stream processor; 431. Speed ​​analysis module; 432. Timer; 433. Selection module; 434. Image analysis module; 435. Output module; 44. Mode switch; 45. Transmission module; 5. Cloud server. Detailed Implementation

[0029] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Example 1

[0030] like Figure 1 As shown, a safety emergency patrol recorder includes a stabilizer 1, a recorder body 2, a speed sensor 3, a local processor 4, and a cloud server 5. The recorder body 2 is fixed to a fixed position on the user's body via the stabilizer 1. Both the speed sensor 3 and the recorder body 2 are connected to the local processor 4, which is remotely connected to the cloud server 5. The local processor 4 determines the on / off state of video stream storage of the recorder body 2 based on the speed status collected by the speed sensor 3 and the degree of change in the video stream captured by the recorder body 2 over time. In practical use, to avoid noise caused by speed sensor jitter, the speed sensor is mounted on the stabilizer in this embodiment.

[0031] like Figure 2As shown, the local processor 4 includes a local memory 41, a local buffer 42, a video stream processor 43, a mode switch 44, and a transmission module 45. The local memory 41 is used to temporarily store the recorded video locally and transmit it to the cloud server 5 in a fixed quantity through the transmission module 45. The local buffer 42 is used to pre-store video stream segments to be transmitted to the local memory 41, and transfer them to the local memory 41 after confirmation by the video stream processor 43. The mode switch 44 is used by the user to switch between the normal storage mode and the automatic storage mode. In the normal storage mode, the local memory 41 stores all video streams recorded when the recorder body 2 is powered on. In the automatic storage mode, the recorded video stream is stored when the speed sensor 3 detects a slowdown and the video stream's continuous frame change rate is less than a first threshold, until the video stream storage is turned off after the above conditions are met. The video stream processor 43 is used to identify the video stream and speed status in the automatic storage mode.

[0032] like Figure 3 As shown, the video stream processor 43 includes a speed analysis module 431, a timer 432, a selection module 433, a picture analysis module 434, and an output module 435. The timer 432 is a counter that accurately times based on the frequency and duty cycle of the input pulse signal. The speed analysis module 431 is used to record speed information and obtain the user's normal movement speed based on historical information. Based on the normal movement speed, it determines whether the user is in a deceleration state and transmits the real-time deceleration state signal to the selection module 433. The selection module 433 is used to continuously receive video streams and, after receiving the deceleration state signal, transmits the video stream at the corresponding position to the picture analysis module 434. The picture analysis module 434 is used to judge the picture changes of the video streams selected by the selection module 433 and transmits the part of the video stream below the first threshold to the output module 435. The output module 435 adds a timestamp to the received video stream according to the pulse signal of the timer 432 and outputs the value to the local memory 41.

[0033] like Figure 4 As shown, the specific method by which the video stream processor 43 selects the video is as follows:

[0034] S1. The speed analysis module 431 takes the speed with the longest duration after noise reduction from the historical speed of the speed sensor 3 as the normal moving speed.

[0035] S2. The speed analysis module 431 continuously receives the speed information from the speed sensor 3, and takes the signal that the speed analysis module 431 receives is continuously lower than the normal moving speed as a deceleration state signal, and sends the deceleration state signal and the corresponding starting time stamp to the selection module 433.

[0036] S3. Select module 433 transmits the video stream segments temporarily stored in local buffer 42 to local storage module and simultaneously transmits them to screen analysis module 434.

[0037] S4, the image analysis module 434 performs frame extraction on the video stream after the starting timestamp;

[0038] S5. Determine whether the proportion of pixels at the same position in each frame of the video stream after frame extraction is lower than the threshold. If yes, proceed to step S6; otherwise, delete the video stream after the starting timestamp in the local storage 41.

[0039] S6. Determine whether there is a frame with a translation speed and deformation ratio that is less than or equal to the translation speed and deformation ratio under normal movement speed. If yes, continue to send the video stream to the local storage 41 until the slowdown state ends. Otherwise, delete the video stream after the starting timestamp from the local storage 41.

[0040] Optionally, stabilizer 1 is a six-axis stabilizer 1.

[0041] This system uses a stabilizer to physically compensate the camera, preventing strong shaking during recording and avoiding large fluctuations in speed sensor parameters that could affect the judgment of actual parameters.

[0042] When users are focused, they tend to slow down their observation to prevent unexpected mistakes due to insufficient observational ability. Simultaneously, for important scenes, observers often maintain their attention in a fixed position. This means that scenes containing significant information often appear directional from the user's perspective. When slowing down and fixing the observation point are met, it's often possible to conclude that the observed object possesses certain unique characteristics.

[0043] During the inspection process, two modes are often used to control local storage.

[0044] One approach is user-initiated switching. The objectivity of this mode depends on the operator, and selective recording often reduces the video's objectivity. The second approach uses machine learning to pre-define the targets to be recorded. However, this is limited by the types of problems identified in the machine learning training set, making it difficult to reflect unexpected situations that might require recording.

[0045] This embodiment adopts the third method, which records the inspection based on the operator's instinctive habitual behavior, while avoiding actual intervention by the operator in the selection of records, thus ensuring the objectivity of the inspection video.

[0046] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A special inspection recorder for safety emergency response, characterized in that: The system includes a stabilizer (1), a recorder body (2), a speed sensor (3), a local processor (4), and a cloud server (5). The recorder body (2) is fixed to a fixed position on the user's body via the stabilizer (1). The speed sensor (3) and the recorder body (2) are both connected to the local processor (4). The local processor (4) and the cloud server (5) are remotely connected. The local processor (4) determines the on / off state of the video stream storage of the recorder body (2) based on the speed status collected by the speed sensor (3) and the degree of change of the video stream captured by the recorder body (2) over time. The local processor (4) includes a local memory (41), a local buffer (42), a video stream processor (43), a mode switch (44), and a transmission module (45). The local memory (41) is used to temporarily store the recorded video locally and transmit it to the cloud server (5) in a quantitative manner through the transmission module (45). The local buffer (42) is used to pre-store the video stream segments to be transmitted to the local memory (41) and transfer them to the local memory (41) when confirmed by the video stream processor (43). The mode switch (44) is used by the user to switch between the normal storage mode and the automatic storage mode. In the normal storage mode, the local memory (41) stores all the video streams recorded by the recorder body (2) when it is powered on. In the automatic storage mode, the recorded video stream is stored until the video stream storage is turned off when the speed sensor (3) detects a stable state after the speed reduction and the continuous frame change rate of the video stream is less than a first threshold. The video stream processor (43) is used to identify the video stream and speed status in the automatic storage mode. The video stream processor (43) includes a speed analysis module (431), a timer (432), a selection module (433), a picture analysis module (434), and an output module (435). The timer (432) is a counter that accurately times the input pulse signal based on the frequency and duty cycle. The speed analysis module (431) is used to record speed information and obtain the user's normal movement speed based on historical information. Based on the normal movement speed, it determines whether the user is in a deceleration state and transmits the real-time deceleration state signal to the selection module (433). The selection module (433) is used to continuously receive video streams and transmit the video stream at the corresponding position to the picture analysis module (434) after receiving the deceleration state signal. The picture analysis module (434) is used to judge the picture change of the received video stream selected by the selection module (433) and transmit the part of the video stream below the first threshold to the output module (435). The output module (435) adds a timestamp to the received video stream according to the pulse signal of the timer (432) and outputs the value to the local memory (41).

2. The safety emergency inspection recorder according to claim 1, characterized in that: The specific method for the video stream processor (43) to select video is as follows: S1. The speed analysis module (431) takes the speed with the longest duration after noise reduction from the historical speed of the speed sensor (3) as the normal moving speed. S2. The speed analysis module (431) continuously receives the speed information from the speed sensor (3), and takes the signal that the speed analysis module (431) receives is continuously lower than the normal moving speed as a deceleration state signal, and sends the deceleration state signal and the corresponding starting time stamp to the selection module (433). S3. The selection module (433) transmits the video stream segments temporarily stored in the local buffer (42) and temporarily stores them in the local memory (41), and simultaneously transmits them to the screen analysis module (434). S4. The image analysis module (434) performs frame extraction on the video stream after the starting timestamp; S5. Determine whether the proportion of pixels at the same position in each frame of the video stream after frame extraction is lower than the threshold. If yes, proceed to step S6; otherwise, delete the video stream after the starting timestamp in the local memory (41). S6. Determine whether there is a translation speed and deformation ratio of adjacent frames above the judgment threshold that is less than or equal to the translation speed and deformation ratio under normal movement speed. If yes, continue to send the video stream to the local memory (41) until the deceleration state ends. Otherwise, delete the video stream after the starting timestamp from the local memory (41).

3. The safety emergency inspection recorder according to claim 1, characterized in that: The stabilizer (1) is a six-axis stabilizer.

Images