Multi-terminal cooperative ferromagnetic detection visual management system

Abstract

This invention discloses a multi-terminal collaborative ferromagnetic detection visualization management system, relating to the field of ferromagnetic detection technology. It addresses the problems of ferromagnetic alarm systems failing to accurately attribute alarm events to specific individuals and the inefficiency of on-duty response due to untimely video stream distribution in heterogeneous terminal environments. By collecting ferromagnetic intensity time-series waveforms and identifying the start time and duration of disturbances based on background magnetic field baseline deviation, the system aligns the timing of each candidate object crossing the detection line in the host camera's view with the start time of the disturbance to generate a rhythm matching degree. Combined with the visual contour features of the object, it generates visual evidence assessment and outputs an attribution and merging indication based on the rhythm matching degree. It also detects the streaming response delay of various terminals and implements differentiated priority push for the assigned person's video segment, achieving accurate attribution of ferromagnetic alarm events and multi-terminal collaborative visualization distribution, thus improving the timeliness and accuracy of ferromagnetic security control at the entrance of the MRI department.

Description

Technical Field

[0001] This invention relates to the field of ferromagnetic detection technology, and more specifically, to a multi-terminal collaborative ferromagnetic detection visualization management system. Background Technology

[0002] In the MRI department of an institution, ferromagnetic safety management of the entrance detection area is a crucial aspect of ensuring the safe operation of MRI examinations. Ferromagnetic objects entering the vicinity of the MRI scanning room pose a risk of strong magnetic field accidents. Therefore, ferromagnetic sensors are typically installed at the entrance of the MRI department to continuously monitor passersby. Existing ferromagnetic detection systems continuously collect magnetic field intensity values ​​within the detection area using magnetic field sensitive units such as fluxgate sensors or Hall elements. On-duty personnel then determine the source of the object based on the on-site situation or video footage.

[0003] The existing technology has the following shortcomings:

[0004] Currently, existing technologies rely solely on magnetic field strength thresholds to trigger alarms, lacking a mechanism for temporal alignment and visual corroboration of alarm events with specific individuals. This makes it impossible to accurately link ferromagnetic events to the specific individuals who triggered the alarm, resulting in a significant reduction in alarm response time. Furthermore, the lack of detection and differentiated adaptation for response delays across different terminals prevents some duty terminals from acquiring effective images immediately after an alarm occurs. Therefore, a multi-terminal collaborative ferromagnetic detection visualization management system is proposed.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a multi-terminal collaborative ferromagnetic detection visualization management system. This system addresses the problems mentioned in the background art by employing a person attribution and merging mechanism that combines temporal rhythm alignment with visual contour features of the carried object, and a differentiated hierarchical distribution strategy based on terminal pull response delay.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a multi-terminal collaborative ferromagnetic detection visualization management system, including a disturbance capture module, a rhythm alignment module, an attribution merging module, and a hierarchical distribution module, with signal connections between the modules;

[0008] The disturbance capture module is used to continuously acquire the ferromagnetic intensity time-series waveform, and simultaneously perform sliding statistical updates on the background magnetic field baseline during periods when no one passes through. Based on the background magnetic field baseline, the deviation of the ferromagnetic intensity time-series waveform is identified and the deviation amount is obtained. The deviation amount is used to determine whether a disturbance state has been entered. Once a disturbance state has been entered, the disturbance start time is obtained and the disturbance duration interval is generated. The disturbance start time and the disturbance duration interval are then transmitted to the rhythm alignment module.

[0009] The rhythm alignment module is used to receive the start time of the disturbance and the duration of the disturbance, set up a detection line, detect and track each individual person in the host camera's view, collect the entry trajectory of each person and record the timing rhythm of each person crossing the detection line, construct a candidate set, align the timing rhythm of each person crossing the detection line in the candidate set with the start time of the disturbance one by one to generate alignment deviation, calculate the rhythm matching degree based on the alignment deviation, and pass the candidate set and rhythm matching degree to the attribution and merging module.

[0010] The attribution and merging module is used to receive the candidate set and rhythm matching degree, retrieve the visual contour features of the carried objects in the host camera image area where the candidate set is located, generate visual evidence evaluation based on the visual contour features of the carried objects, combine the rhythm matching degree and visual evidence evaluation to perform attribution and merging of the candidate set, generate attribution and merging instructions, and pass the attribution and merging instructions to the hierarchical distribution module.

[0011] The hierarchical distribution module is used to receive the attribution merging instruction, detect the terminal pull-stream response delay, and perform hierarchical distribution of the attribution merging instruction according to the terminal pull-stream response delay. For different levels, it selects to push the high-priority video stream of the screen segment where the attribution person is located or the key frame sequence of the screen segment where the attribution person is located. When pushing the key frame sequence, the remaining frames are gradually supplemented according to the current available bandwidth.

[0012] In a preferred embodiment, the disturbance capture module continuously acquires the ferromagnetic intensity time-series waveform within the detection area at the entrance of the MRI department. When determining whether a disturbance state has been entered, the module determines that a disturbance state has been entered when the deviation exceeds a preset deviation threshold.

[0013] The moment when the deviation first exceeds the preset deviation threshold is taken as the start time of the disturbance, and the moment when the deviation falls back below the preset deviation threshold is taken as the end time of the disturbance. The duration from the start time of the disturbance to the end time of the disturbance is taken as the disturbance duration interval.

[0014] The rhythm alignment module sets a detection line in the host camera image according to the physical boundary of the detection area at the entrance of the MRI department. The moment when the human body crosses the detection line is taken as the timing rhythm of each person crossing the detection line. Based on the duration of the disturbance, the host camera image delineates all the people who fall within the duration of the disturbance and complete the crossing to form a candidate set.

[0015] In a preferred embodiment, the visual contour features of the carried object in the rhythm alignment module include the outline of the handheld object, the bulging shape of the clothing pocket, and the shape of the portable push device.

[0016] The terminal pull response latency in the hierarchical distribution module includes the terminal pull response latency of Android displays, tablets, and mobile terminals during the pull handshake phase.

[0017] In a preferred embodiment, in the disturbance capture module, the ferromagnetic intensity time-series waveform is a sequence of magnetic induction values ​​continuously output over time at the location of the MRI department entrance detection area.

[0018] The background magnetic field baseline is retrieved, and the period when the deviation of the ferromagnetic intensity time waveform from the existing background magnetic field baseline is continuously lower than the preset deviation threshold is taken as the period when no one passes through.

[0019] During periods when no one passes through, the ferromagnetic intensity time-series waveform is averaged and updated according to a preset sliding window to obtain a new background magnetic field baseline, which is continuously updated in subsequent acquisitions.

[0020] Deviation identification is performed on the ferromagnetic intensity time series waveform based on the background magnetic field baseline. The absolute value of the difference between the current value of the ferromagnetic intensity time series waveform and the value of the background magnetic field baseline at the same time is taken as the deviation amount at the corresponding time.

[0021] In a preferred embodiment, in the rhythm alignment module, each individual person in the host camera's view is detected and tracked, and the continuous change trajectory of the centroid coordinates of each person over time from entering the host camera's view to leaving the host camera's view is recorded as the entry trajectory of each person.

[0022] Compare the timing and rhythm of each character crossing the probe line with the duration of the disturbance one by one:

[0023] When the timing of a character crossing the detection line falls within the duration of the disturbance, the character is included in the candidate set.

[0024] Otherwise, the person will not be included in the candidate list;

[0025] The absolute value of the difference between the timing of the character crossing the probe line and the starting time of the disturbance is taken as the alignment deviation of the character.

[0026] The rhythm matching degree of each candidate is calculated based on the alignment deviation.

[0027] In a preferred embodiment, the object holding outline is the outer edge shape of an object that extends from the hand area of ​​the candidate subject group and is separated from the human body outline; the clothing pocket bulge shape is the outline deformation feature of the pocket position of the candidate subject group's upper or lower garment relative to the flat clothing surface; the portable push device shape is the overall outer edge shape of a wheeled or push rod device that appears in front of or to the side of the candidate subject group and accompanies the subject body.

[0028] In a preferred embodiment, for the area of ​​the host camera screen occupied by each person in the candidate set, three types of visual contour features of the carried object are detected: the outline of the handheld object, the shape of the bulging pocket of the clothing, and the shape of the push device. Each type of feature is output in a binary manner: when the corresponding feature is detected, it is recorded as 1, and when the corresponding feature is not detected, it is recorded as 0.

[0029] The sum of the visual contour feature detection values ​​of the three types of objects carried by each person in the candidate set is used as the visual corroboration assessment of the person.

[0030] In a preferred embodiment, rhythm matching is single and significantly superior, and visual corroboration is used to evaluate non-spatial-time bound characters.

[0031] When the difference between the maximum and the second largest rhythm matching degree does not exceed the preset rhythm differentiation threshold, the largest and only one is identified by visual evidence and assigned to the binding.

[0032] When the candidate with the highest rhythmic match is not unique and the candidate with the greatest visual evidence is not unique, mark it as multiple unresolved objects and retain all the screen segments;

[0033] The significant dominance of rhythm matching degree is determined by comparing the difference between the maximum and the second largest rhythm matching degree of the candidate set with a preset rhythm differentiation threshold. The preset rhythm differentiation threshold is the lower limit of the distribution of the difference in rhythm matching degree between adjacent characters in a scene where a single person holds a ferromagnetic object.

[0034] In a preferred embodiment, in the hierarchical distribution module, before the distribution affiliation and merging instruction, the host sends a pull-stream handshake probe packet to each of the registered Android display, tablet, and mobile phone terminals one by one, records the time when each terminal returns the first handshake response packet, and uses the difference between the time when the pull-stream handshake probe packet is sent as the terminal pull-stream response delay of the corresponding terminal.

[0035] When the terminal's streaming response latency is lower than the preset latency classification threshold, the terminal is determined to be a short latency terminal, and the high-priority video stream of the segment where the person in the attribution and merging instruction is located is pushed first.

[0036] When the terminal's streaming response delay is not lower than the preset delay classification threshold, the terminal is determined to be a long-latency terminal. First, the key frame sequence of the screen segment where the person to be assigned is located in the attribution and merging instruction is pushed, and the remaining frames of the screen segment where the person to be assigned is located are gradually supplemented according to the terminal's current available bandwidth.

[0037] The technical effects and advantages of this invention are as follows:

[0038] This invention identifies the start time and duration of disturbances by deviating from the background magnetic field baseline. It then aligns the timing of candidate objects crossing the detection line with the start time of the disturbance to generate a rhythm matching degree. Combined with visual evidence, it outputs an attribution and merging indication, enabling accurate and automatic attribution of ferromagnetic alarm events to the triggering individuals. This avoids the time loss caused by manual frame-by-frame comparison by on-duty personnel. It prioritizes pushing high-priority video streams to short-latency terminals and pushes key frame sequences to long-latency terminals, supplementing them according to bandwidth. This achieves differentiated and timely distribution of attribution video segments in heterogeneous terminal environments, ensuring that personnel at multiple locations can obtain effective images as soon as an alarm is triggered. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the implementation of the multi-terminal collaborative ferromagnetic detection visualization management system of the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0041] This invention achieves accurate attribution of ferromagnetic alarm events and multi-terminal collaborative visualization by combining temporal rhythm alignment with visual evidence and a hierarchical distribution strategy based on terminal pull response delay, thereby comprehensively improving the response timeliness of ferromagnetic security control at the entrance of the MRI department.

[0042] Example: A multi-terminal collaborative ferromagnetic detection visualization management system, such as... Figure 1 As shown, it includes a disturbance capture module, a rhythm alignment module, a home merging module, and a hierarchical distribution module, with signal connections between each module;

[0043] The functions of each module are as follows:

[0044] The disturbance capture module is used to continuously acquire ferromagnetic intensity time-series waveforms in the detection area at the entrance of the MRI department, and simultaneously perform sliding statistical updates on the background magnetic field baseline during periods when no one passes through. Based on the background magnetic field baseline, the ferromagnetic intensity time-series waveform is deviated. When the deviation exceeds the preset deviation threshold, it is determined to enter a disturbance state. The moment when the deviation first exceeds the preset deviation threshold is taken as the disturbance start time, and the moment when the deviation falls back below the preset deviation threshold is taken as the disturbance end time. The duration from the disturbance start time to the disturbance end time is taken as the disturbance duration interval. The disturbance start time and the disturbance duration interval are transmitted to the rhythm alignment module.

[0045] The rhythm alignment module is used to receive the start time of the disturbance and the duration of the disturbance. A detection line is preset in the host camera image according to the physical boundary of the detection area at the entrance of the MRI department. Each individual in the host camera image is detected and tracked, and the entry trajectory of each person is collected. The moment when each person's physical heart crosses the detection line is recorded as the temporal rhythm of each person crossing the detection line. According to the duration of the disturbance, all persons who fall within the duration of the disturbance and complete the crossing are selected in the host camera image to form a candidate set. The temporal rhythm of each person crossing the detection line in the candidate set is aligned with the start time of the disturbance one by one to generate alignment deviation. The rhythm matching degree is calculated based on the alignment deviation. The candidate set and the rhythm matching degree are passed to the attribution and merging module.

[0046] The attribution and merging module is used to receive the candidate set and rhythm matching degree, retrieve the visual contour features of the carried items in the host camera screen area where the candidate set is located. The visual contour features of the carried items include the outline of the handheld item, the shape of the bulging pocket of clothing, and the shape of the portable push device. Based on the visual contour features of the carried items, a visual evidence assessment is generated. The candidate set is attributed and merged by combining the rhythm matching degree and the visual evidence assessment, and an attribution and merging instruction is generated. The attribution and merging instruction is then passed to the hierarchical distribution module.

[0047] The hierarchical distribution module is used to receive the attribution merging instruction, detect the terminal pull response latency of various terminals such as Android displays, tablets, and mobile phones during the pull handshake phase, and perform hierarchical distribution of the attribution merging instruction according to the terminal pull response latency. For terminals with shorter response latency, the high-priority video stream of the segment where the attribution person is located is pushed first, and for terminals with longer response latency, the key frame sequence of the segment where the attribution person is located is pushed first, and the remaining frames are gradually filled in according to the current available bandwidth.

[0048] In the disturbance capture module, a ferromagnetic sensing device is arranged in the detection area at the entrance of the MRI department to continuously acquire the time-series waveform of ferromagnetic intensity.

[0049] The ferromagnetic intensity time-series waveform is a sequence of magnetic intensity values ​​continuously output by the ferromagnetic sensing device at the detection area at the entrance of the MRI department over time. It is used to characterize the changes in the magnetic field caused by the presence or absence of ferromagnetic materials in the detection area.

[0050] The ferromagnetic sensing device is deployed at the boundary of the detection area at the entrance of the MRI department. It consists of magnetic field sensitive units such as fluxgate sensors or Hall elements and corresponding signal conditioning circuits. It is used to detect the magnetic field strength in the detection area in real time at a preset sampling frequency and output the corresponding value to the host. The value output when the ferromagnetic object enters or leaves the detection area changes with the position of the corresponding ferromagnetic object in the detection area and its ferromagnetic content.

[0051] While acquiring the time-series waveform of ferromagnetic intensity, the background magnetic field baseline is retrieved;

[0052] The background magnetic field baseline is a reference sequence obtained by performing sliding statistics on the time-series waveform of ferromagnetic intensity during periods of no human passage. It is used to characterize the steady-state magnetic field level of the detection area when there is no ferromagnetic interference.

[0053] Specifically, periods in which the deviation of the ferromagnetic intensity time-series waveform from the existing background magnetic field baseline is continuously lower than a preset deviation threshold are defined as periods when no one passes through. During these periods, the ferromagnetic intensity time-series waveform is averaged according to a preset sliding window to obtain a new background magnetic field baseline, which is continuously updated in subsequent acquisitions to adapt to the effects of long-term, slowly changing factors such as environmental temperature drift and equipment start-up and shutdown at the entrance of the MRI department.

[0054] Deviation identification is performed on the ferromagnetic intensity time series waveform based on the background magnetic field baseline. The absolute value of the difference between the current value of the ferromagnetic intensity time series waveform and the value of the background magnetic field baseline at the same time is taken as the deviation at that time.

[0055] ;

[0056] in, for The ferromagnetic intensity time-series waveform values ​​at time t. for The baseline value of the background magnetic field at time 10:00. for The deviation at any given time has dimensions consistent with the time-series waveform of ferromagnetic intensity.

[0057] The greater the deviation, the more significantly the ferromagnetic intensity time-series waveform deviates from the background magnetic field baseline at that moment, reflecting a higher probability of ferromagnetic interference in the detection area.

[0058] Compare the deviation amount with the preset deviation threshold:

[0059] When the deviation exceeds the preset deviation threshold for the first time and remains above it for the preset minimum duration, it is determined that a disturbance state has been entered, and the moment when the deviation first exceeds the preset deviation threshold is taken as the disturbance start time.

[0060] When the deviation falls below the preset deviation threshold during the duration of the disturbance, that moment is considered the end of the disturbance.

[0061] If the deviation amount only momentarily exceeds the preset deviation threshold but does not reach the preset minimum duration, it is not considered a disturbance and no disturbance start time is generated.

[0062] The duration from the start of the disturbance to its end is defined as the disturbance duration interval, reflecting the time span during which the ferromagnetic material lingers in the detection area at the entrance of the MRI department and generates magnetic field disturbances.

[0063] Through the above deviation identification and time recording process, the start time of the disturbance and the duration of the disturbance are obtained and transmitted to the rhythm alignment module for alignment with the timing rhythm of each person crossing the detection line in the picture.

[0064] It should be noted that the preset deviation threshold is set as follows: during periods when no one passes through the detection area at the entrance of the MRI department, the ferromagnetic intensity time-series waveform and the background magnetic field baseline are continuously collected for several hours. The distribution range of the deviation during this period is statistically analyzed, and a certain multiple of the upper limit of the deviation distribution is taken as the preset deviation threshold to ensure that normal environmental noise will not trigger the disturbance judgment. The preset minimum duration is obtained by statistically analyzing the shortest dwell time of normal pedestrians passing through the detection area of ​​the ferromagnetic sensor at the entrance of the MRI department, and its lower limit is taken to filter out the peak deviation caused by instantaneous electromagnetic interference.

[0065] In the rhythm alignment module, the start time of the disturbance and the duration of the disturbance are received from the disturbance capture module, and a detection line is preset in the host camera image according to the physical boundary position of the detection area at the entrance of the MRI department.

[0066] The detection line is a pixel-length straight line segment that is pre-set in the coordinate system of the host camera screen and coincides with the physical boundary of the entrance side of the detection area at the entrance of the MRI department on the ground. It is used to determine whether the person in the host camera screen has completed crossing into the detection area.

[0067] The system detects and tracks each individual person in the host camera's view, recording the continuous change trajectory of the centroid coordinates of each person over time from entering the host camera's view to leaving the view, which serves as the entry trajectory of each person.

[0068] The entry trajectory of each character is used to represent the movement process of the character gradually entering the detection area from the outside of the screen, reflecting spatial motion information such as the direction of entry and the speed of movement.

[0069] The moment when each person's center of mass first crosses the detection line is taken as the temporal rhythm of each person's crossing of the detection line. That is, for each person, the moment when their center of mass moves from the outside of the detection line to the inside of the detection line in the host camera's view is recorded as the temporal rhythm of that person's crossing of the detection line.

[0070] The timing of each character crossing the detection line corresponds to the specific time point at which the character arrives at the physical boundary of the detection area at the entrance of the MRI department.

[0071] Based on the duration of the disturbance, a set of candidate individuals is selected in the main camera's view. Specifically, the timing of each individual crossing the detection line is compared with the duration of the disturbance.

[0072] When the timing of a character crossing the detection line falls within the interval of continuous disturbance, that character is included in the candidate set.

[0073] If the timing of a character crossing the probe line does not fall within the duration of the disturbance, that character will not be included in the candidate set.

[0074] The timing of each candidate character crossing the probe line is aligned with the start time of the disturbance. The absolute value of the difference between the timing of the character crossing the probe line and the start time of the disturbance is taken as the alignment deviation of that character.

[0075] ;

[0076] in, For the candidate list The timing and rhythm of each character traversing the probe line. This represents the start time of the disturbance. For the first Positional deviation of individual characters and All units are seconds. The dimension is seconds.

[0077] The smaller the alignment deviation, the closer the moment the person crosses the detection line is to the start of the disturbance, and the higher the probability that the person is the source of the ferromagnetic alarm.

[0078] Calculate the rhythm matching degree of each candidate based on alignment deviation:

[0079] ;

[0080] in, For the first The alignment deviation of an individual figure, measured in seconds. The duration of the disturbance interval is expressed in seconds. For the first The rhythmic matching degree of a character is calculated by dividing the numerator and denominator, which have the same dimensions, to obtain a dimensionless value, with a range of values ​​of [value missing]. .

[0081] The smaller the alignment deviation, the higher the rhythm matching degree; when the moment when the person crosses the detection line is exactly the moment when the disturbance begins, the alignment deviation is zero, the rhythm matching degree takes the maximum value of 1, reflecting that the person's crossing behavior and the start of the magnetic field disturbance are completely coincident in time.

[0082] By defining the candidate set and calculating the rhythm matching degree, a quantitative alignment relationship is established between the temporal characteristics of magnetic field disturbance and the temporal characteristics of the person's passage. The candidate set and the rhythm matching degree are then passed to the attribution merging module for attribution merging in conjunction with visual evidence.

[0083] In the attribution and merging module, the candidate object set and rhythm matching degree are received from the rhythm alignment module, and the visual contour features of the carried objects in the host camera image area where the candidate object set is located are retrieved.

[0084] The visual contour features of the carried object refer to the visual appearance features related to the ferromagnetic object carried by the person, obtained by contour extraction algorithm within the area occupied by each person in the candidate set in the host camera image. These features include three categories: the outline of the handheld object, the shape of the bulging pocket of clothing, and the shape of the push device carried by the person.

[0085] Among them, the outline of the handheld object is the outer edge shape of the object that extends from the hand area of ​​the candidate and is separated from the outline of the human body; the bulging shape of the clothing pocket is the outline deformation feature of the pocket of the candidate's upper or lower garment relative to the flat surface of the clothing; the shape of the portable push device is the overall outer edge shape of the wheeled or push rod device that appears in front of or to the side of the candidate and accompanies the person.

[0086] For the area of ​​the host camera screen occupied by each person in the candidate set, the visual contour features of the three types of carried objects are detected respectively. Each type of feature is output in binary form: 1 is recorded when the corresponding feature is detected, and 0 is recorded when the corresponding feature is not detected.

[0087] Visual corroboration assessments are generated based on the visual contour feature analysis of the carried items. The sum of the detection values ​​of the three types of visual contour features of the carried items for each person in the candidate set is used as the visual corroboration assessment for that person.

[0088] ;

[0089] in, For the first The detection value of the handheld object contour of each candidate. For the first Detection values ​​for bulging pocket shapes in the clothing of each candidate. For the first The morphological detection values ​​of the personal push device of each candidate are all dimensionless indicators, with values ​​of 0 or 1. For the first Visual corroboration assessment of each candidate, dimensionless, with values ​​of 0, 1, 2, or 3.

[0090] The higher the visual corroboration assessment, the more complete the visual evidence for the candidate object as a source of ferromagnetic alarms. When the visual corroboration assessment is 0, it means that no visual contour features of any type of object were detected in the candidate object, and the visual evidence is empty.

[0091] By combining rhythm matching and visual corroboration assessments, the candidate object set is categorized and merged to generate categorization indicators:

[0092] When there is only one person in the candidate set whose rhythm matching degree is significantly superior and whose visual evidence evaluation is not empty, the alarm event is bound to that person, and the attribution and merging indicator output is the corresponding screen segment of that person.

[0093] When multiple candidates have similar rhythm matching in the candidate group, the candidate with the superior visual evidence assessment is assigned to the binding, and the output of the merging instruction is the screen segment corresponding to the candidate with the highest visual evidence assessment.

[0094] When there is no significant difference between the rhythm matching degree and the visual evidence assessment, the alarm event is marked as multi-object unresolved, and the attribution and merging instruction retains the screen segments of all candidates.

[0095] By generating and classifying visual evidence through the above-mentioned visual evidence assessment, the temporal rhythm matching degree and the spatial visual evidence assessment are combined for the determination of the person attribution in alarm events, and the classification instruction is output and transmitted to the hierarchical distribution module.

[0096] It should be noted that the method for determining significant dominance in rhythm matching is as follows: The rhythm matching scores are sorted in descending order in the candidate set, and the difference between the highest and second-highest rhythm matching scores is taken. If this difference exceeds the preset rhythm differentiation threshold, it is determined that a single rhythm matching score is significantly dominant. The method for determining dominance in visual corroboration evaluation is as follows: The maximum value of the visual corroboration evaluation in the candidate set is taken. If this maximum value appears uniquely in the candidate set, it is determined that the visual corroboration evaluation is dominant. The preset rhythm differentiation threshold is set by recording the difference in rhythm matching scores between adjacent individuals in two scenarios during the pre-test phase at the entrance of the MRI department: one person holding a ferromagnetic object passing through and multiple people passing through simultaneously. The lower limit of the distribution of the difference in adjacent matching scores in the single-person passing through scenario is taken as the preset rhythm differentiation threshold to ensure that multiple people passing through simultaneously will not be misjudged as having a single rhythm matching score significantly dominant.

[0097] In the hierarchical distribution module, the home merging instruction is received from the home merging module, and the terminal pull response latency of multiple terminals is detected.

[0098] Terminal streaming response latency is the time elapsed from the issuance of the handshake request to the return of the first handshake response packet during the handshake phase when various terminals such as Android displays, tablets, and mobile phones initiate video stream retrieval from the host. It is measured in milliseconds and reflects the terminal's response rate to receive the video stream pushed by the host under the current network and load conditions.

[0099] Specifically, before distributing the ownership and merging instruction, the host sends a pull-stream handshake probe packet to each of the registered Android display, tablet, and mobile phone terminals one by one, records the time when each terminal returns the first handshake response packet, and uses the difference between this time and the time when the pull-stream handshake probe packet is sent as the terminal pull-stream response delay of the corresponding terminal.

[0100] The attribution and merging instructions are distributed hierarchically based on the terminal streaming response latency, and the terminal streaming response latency of each type of terminal is compared with a preset latency classification threshold:

[0101] When the terminal's streaming response latency is lower than the preset latency classification threshold, the terminal is determined to be a short latency terminal, and the high-priority video stream of the segment of the screen where the person in the attribution and merging instruction is located is pushed to the terminal first.

[0102] When the terminal's streaming response delay is not lower than the preset delay classification threshold, the terminal is determined to be a long-latency terminal. The key frame sequence of the screen segment where the person to be assigned is located in the attribution and merging instruction is first pushed to the terminal, and the remaining frames of the screen segment where the person to be assigned is located are gradually supplemented according to the terminal's current available bandwidth.

[0103] Among them, the high-priority video stream is the complete continuous frame sequence of the screen segment where the person to be assigned is located in the attribution and merging instruction. It is continuously pushed to the short-latency terminal at the original frame rate of the host, so that the on-duty personnel can obtain continuous images in real time when handling the situation at the entrance of the MRI department. The key frame sequence is a sparse frame sequence extracted from the screen segment where the person to be assigned is located in the attribution and merging instruction at a preset frame extraction interval. It is pushed to the long-latency terminal in advance to ensure that the basic screen information is available. Then, the remaining frames are gradually filled in according to the available bandwidth of the terminal.

[0104] It should be noted that the preset latency classification threshold is set as follows: During the normal working hours of the MRI department, the terminal pull response latency of various terminals such as Android monitors, tablets, and mobile phones is sampled and statistically analyzed. The median value of the distribution of all terminal pull response latency is taken as the preset latency classification threshold to ensure that the division between short-latency terminals and long-latency terminals matches the actual network conditions on site. The preset frame extraction interval is determined based on the total duration of the image segment in the attribution and merging indication and the minimum available bandwidth of the long-latency terminal class. The lower limit of the interval is taken so that the total amount of key frame sequence data can complete the first batch of push within an acceptable latency under the minimum available bandwidth of the long-latency terminal.

[0105] Through the aforementioned terminal pull response delay detection and hierarchical distribution, high-priority video streams are prioritized for terminals with shorter response delays, while key frame sequences are pushed to terminals with longer response delays and distributed according to bandwidth. This ensures that the attribution and merging instructions for ferromagnetic alarm events at the entrance of the MRI department are presented to the on-duty personnel in a timely and accessible manner in a heterogeneous terminal environment.

[0106] Finally, it should be noted that in this paper, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0107] Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0108] In this document, the singular forms “a,” “an,” and “the” may also include the plural forms unless the context clearly indicates otherwise. It should also be understood that terms such as “comprising / including” or “having” specify the presence of the stated features, integrals, steps, operations, components, parts, or combinations thereof, but do not preclude the possibility of the presence or addition of one or more other features, integrals, steps, operations, components, parts, or combinations thereof. Meanwhile, the term “and / or” as used in this specification includes any and all combinations of the associated listed items.

[0109] The various embodiments in this specification are described in a progressive manner. Each embodiment focuses on the differences from other embodiments. The various embodiments can be combined as needed, and the same or similar parts can be referred to each other.

[0110] The above description of the disclosed embodiments will enable those skilled in the art to make or use various modifications to these embodiments. It will be readily apparent to those skilled in the art that the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multi-terminal collaborative ferromagnetic detection visualization management system, characterized in that: It includes a disturbance capture module, a rhythm alignment module, a home merging module, and a hierarchical distribution module, with signal connections between the modules; The disturbance capture module is used to continuously acquire the ferromagnetic intensity time-series waveform, and simultaneously perform sliding statistical updates on the background magnetic field baseline during periods when no one passes through. Based on the background magnetic field baseline, the deviation of the ferromagnetic intensity time-series waveform is identified and the deviation amount is obtained. The deviation amount is used to determine whether a disturbance state has been entered. Once a disturbance state has been entered, the disturbance start time is obtained and the disturbance duration interval is generated. The disturbance start time and the disturbance duration interval are then transmitted to the rhythm alignment module. The rhythm alignment module is used to receive the start time of the disturbance and the duration of the disturbance, set up a detection line, detect and track each individual person in the host camera's view, collect the entry trajectory of each person and record the timing rhythm of each person crossing the detection line, construct a candidate set, align the timing rhythm of each person crossing the detection line in the candidate set with the start time of the disturbance one by one to generate alignment deviation, calculate the rhythm matching degree based on the alignment deviation, and pass the candidate set and rhythm matching degree to the attribution and merging module. The attribution and merging module is used to receive the candidate set and rhythm matching degree, retrieve the visual contour features of the carried objects in the host camera image area where the candidate set is located, generate visual evidence evaluation based on the visual contour features of the carried objects, combine the rhythm matching degree and visual evidence evaluation to perform attribution and merging of the candidate set, generate attribution and merging instructions, and pass the attribution and merging instructions to the hierarchical distribution module. The hierarchical distribution module is used to receive the attribution merging instruction, detect the terminal pull-stream response delay, and perform hierarchical distribution of the attribution merging instruction according to the terminal pull-stream response delay. For different levels, it selects to push the high-priority video stream of the screen segment where the attribution person is located or the key frame sequence of the screen segment where the attribution person is located. When pushing the key frame sequence, the remaining frames are gradually supplemented according to the current available bandwidth.

2. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 1, characterized in that: The disturbance capture module continuously acquires the ferromagnetic intensity time-series waveform in the detection area at the entrance of the MRI department. When determining whether a disturbance state has been entered, the module determines that a disturbance state has been entered when the deviation exceeds the preset deviation threshold. The moment when the deviation first exceeds the preset deviation threshold is taken as the start time of the disturbance, and the moment when the deviation falls back below the preset deviation threshold is taken as the end time of the disturbance. The duration from the start time of the disturbance to the end time of the disturbance is taken as the disturbance duration interval. The rhythm alignment module sets a detection line in the host camera image according to the physical boundary of the detection area at the entrance of the MRI department. The moment when the human body crosses the detection line is taken as the timing rhythm of each person crossing the detection line. Based on the duration of the disturbance, the host camera image delineates all the people who fall within the duration of the disturbance and complete the crossing to form a candidate set.

3. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 1, characterized in that: The visual contour features of the carried object described in the rhythm alignment module include the outline of the handheld object, the bulging shape of the clothing pocket, and the shape of the portable push device. The terminal pull response latency in the hierarchical distribution module includes the terminal pull response latency of Android displays, tablets, and mobile terminals during the pull handshake phase.

4. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 2, characterized in that: In the disturbance capture module, the ferromagnetic intensity time-series waveform is a sequence of magnetic induction values ​​continuously output over time at the detection area location at the entrance of the MRI department. The background magnetic field baseline is retrieved, and the period when the deviation of the ferromagnetic intensity time waveform from the existing background magnetic field baseline is continuously lower than the preset deviation threshold is taken as the period when no one passes through. During periods when no one passes through, the ferromagnetic intensity time-series waveform is averaged and updated according to a preset sliding window to obtain a new background magnetic field baseline, which is continuously updated in subsequent acquisitions. Deviation identification is performed on the ferromagnetic intensity time series waveform based on the background magnetic field baseline. The absolute value of the difference between the current value of the ferromagnetic intensity time series waveform and the value of the background magnetic field baseline at the same time is taken as the deviation amount at the corresponding time.

5. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 2, characterized in that: In the rhythm alignment module, each individual character in the host camera's view is detected and tracked, and the continuous change trajectory of the centroid coordinates of each character over time is recorded from the moment they enter the host camera's view to the moment they leave the host camera's view, which serves as the entry trajectory of each character. Compare the timing and rhythm of each character crossing the probe line with the duration of the disturbance one by one: When the timing of a character crossing the detection line falls within the duration of the disturbance, the character is included in the candidate set. Otherwise, the person will not be included in the candidate list; The absolute value of the difference between the timing of the character crossing the probe line and the starting time of the disturbance is taken as the alignment deviation of the character. The rhythm matching degree of each candidate is calculated based on the alignment deviation.

6. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 3, characterized in that: The outline of the object being held refers to the outer edge of the object that extends from the hand area of ​​the candidate and is separated from the outline of the human body; the bulging shape of the clothing pocket refers to the outline deformation feature of the pocket of the candidate's upper or lower garment that protrudes relative to the flat surface of the clothing. The portable propulsion device is the overall outer edge of a wheeled or push rod-type device that appears in front of or to the side of the candidate character and accompanies the character.

7. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 1, characterized in that: For the area of ​​the host camera screen occupied by each person in the candidate set, three types of visual contour features of the carried objects are detected: the outline of the handheld object, the shape of the bulging pocket of the clothing, and the shape of the push device. Each type of feature is output in binary form: 1 is recorded when the corresponding feature is detected, and 0 is recorded when the corresponding feature is not detected. The sum of the visual contour feature detection values ​​of the three types of objects carried by each person in the candidate set is used as the visual corroboration assessment of the person.

8. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 7, characterized in that: The rhythm matching degree is significantly superior and the visual evidence is used to evaluate non-space-time bound characters. When the difference between the maximum and the second largest rhythm matching degree does not exceed the preset rhythm differentiation threshold, the largest and only one is identified by visual evidence and assigned to the binding. When the candidate with the highest rhythmic match is not unique and the candidate with the greatest visual evidence is not unique, mark it as multiple unresolved objects and retain all the screen segments; The significant dominance of rhythm matching degree is determined by comparing the difference between the maximum and the second largest rhythm matching degree of the candidate set with a preset rhythm differentiation threshold. The preset rhythm differentiation threshold is the lower limit of the distribution of the difference in rhythm matching degree between adjacent characters in a scene where a single person holds a ferromagnetic object.

9. The multi-terminal collaborative ferromagnetic detection visualization management system according to claim 3, characterized in that: In the hierarchical distribution module, before the distribution ownership and merging instruction, the host sends a pull-stream handshake probe packet to each of the registered Android display, tablet, and mobile phone terminals one by one, records the time when each terminal returns the first handshake response packet, and uses the difference between the time when the pull-stream handshake probe packet is sent as the corresponding terminal pull-stream response delay. When the terminal's streaming response latency is lower than the preset latency classification threshold, the terminal is determined to be a short latency terminal, and the high-priority video stream of the segment where the person in the attribution and merging instruction is located is pushed first. When the terminal's streaming response delay is not lower than the preset delay classification threshold, the terminal is determined to be a long-latency terminal. First, the key frame sequence of the screen segment where the person to be assigned is located in the attribution and merging instruction is pushed, and the remaining frames of the screen segment where the person to be assigned is located are gradually supplemented according to the terminal's current available bandwidth.

Images