JPWO2019008635A1 - Video surveillance device adjusting method and video surveillance device - Google Patents

Abstract

Provided is a method for adjusting an image monitoring device, which appropriately detects an image and / or an error occurring in a digital image signal according to content. Further, the present invention provides an image monitoring device that appropriately detects an image and / or an error in a digital image signal through learning. The video monitoring device has a video / audio monitoring unit having a built-in memory for storing a threshold and a parameter optimization learning unit, and the video monitoring device uses all of a plurality of parameters that change according to a video signal. However, when a threshold value determined according to the monitoring item is exceeded, it is determined that an error has occurred. At least two of the plurality of parameters are related to each other and Depending on the video and / or audio content, any set of thresholds stored in memory is selected. The parameter optimization learning unit learns by analyzing the error reporting result, and the threshold is updated based on the learning result.

Description

  The present invention relates to a method of adjusting a video monitoring device and a video monitoring device capable of mechanically detecting an error of a video or audio included in a digital video signal.

  In the old analog television broadcasting era, the quality of the video and audio was detected by an observer who actually watched (monitored) the video or audio. On the other hand, in the modern era when digital television broadcasting is started, the processing of video signals becomes easier and the environment for mechanically monitoring video and audio is being prepared, and in some cases machine monitoring has already been performed. On the other hand, there are many evaluations that machine monitoring is still inferior to human monitoring in terms of accuracy of error detection, and there is the actual situation that the monitoring by the conventional monitor is also performed in parallel.

  However, in addition to the terrestrial wave business that transmits digital video, it is expected that video content distribution services such as Internet content distribution business and IPTV (Internet Protocol Tele Vision) business will expand in the future. It is believed that it will be practically difficult for an observer to monitor all the contents because the number of contents to be downloaded will increase dramatically. Therefore, improving the accuracy of error detection in machine monitoring to the level of humans can be said to be an urgent issue in all video distribution services.

  Patent Document 1 discloses a voice inspection method for detecting a video error caused by noise caused by various causes in a digital video signal and a voice error caused by noise caused by various causes in a digital audio signal. Has been done.

International Publication No. 2015/059782

  According to Patent Document 1, a continuous digital audio signal is divided into 5 msec or less and sampled, a high frequency component is extracted from the sampled signal, and a value based on the extracted high frequency component is compared with a threshold value to obtain a voice. The error that occurs can be detected mechanically.

  However, in the technique of Patent Document 1, error detection is performed by comparing the characteristic value (parameter) of the audio signal with a threshold value, but there is a problem of how to set the threshold value. That is, when the threshold value is set to be strict, while error detection is frequently performed, when many of the detected errors are insignificant to the viewer, detection may be wasted. On the other hand, if the threshold value is set loosely, the frequency of error detection decreases, but there is a risk that the viewer may miss an error that cannot be ignored. This is one of the reasons why machine monitoring is inferior to human monitoring. Especially when there are multiple parameters and they are related to each other, changing one parameter affects other parameters. Therefore, it is necessary to change two or more parameters at the same time. Is very difficult.

  One of the objects of the present invention is to provide a method of adjusting an image monitoring apparatus that appropriately detects an image and / or an error occurring in a digital image signal according to content.

  Another object of the present invention is to provide an image monitoring apparatus that appropriately detects an image and / or an error in a digital image signal through learning.

According to a first aspect of the present invention, there is provided an image monitoring apparatus adjusting method, which inputs a video signal corresponding to video and / or audio to be monitored and detects that an error has occurred in a monitoring item. In the adjustment method,
The video monitoring device is configured to determine that the error has occurred when all of the plurality of parameters that change according to the video signal exceed a threshold determined according to the monitoring item. , At least two of the plurality of parameters are interrelated,
Each of the thresholds of the plurality of parameters is determined and a plurality of threshold groups are created and stored in a memory,
One of the sets of threshold values stored in the memory is selected according to the video and / or audio content to be monitored.

  As a result of earnest research, the present inventor has found that the appropriate values of the threshold value group are the contents (“movie”, “drama”, “variety”, “sports”, “documentary” of the video and / or audio content to be monitored. , “Entertainment”, “Anime”, etc.). The present invention was derived based on this finding. That is, if the thresholds of the plurality of parameters are respectively determined and a plurality of groups of the thresholds are created and stored in the memory, the thresholds are stored in the memory according to the video and / or audio contents to be monitored. By selecting any one of the threshold groups, as compared with the case of using a single threshold group, it is possible to set a more appropriate threshold group for monitoring, so that it is possible to effectively detect an error in the video monitoring device. It can be prevented.

  Further, the video monitoring device performs learning in parallel with the video and / or audio monitoring operation, updates at least one threshold value of the threshold value group based on the learning result, and stores the updated threshold value group in a memory. It is preferable to store in.

Furthermore, the video surveillance device is designed to issue an alarm when an error is detected,
In the learning, a video signal corresponding to the video and / or audio to be monitored is input, and at a constant time interval, the first event in which the alarm is issued for the error for which the alarm should be issued. Counting the number of cases and the number of second events in which the alarm is issued for an error that should not be issued, respectively.
Further, the number of third events in which the alarm is not issued for an error for which an alarm is to be issued by inputting an inspection video signal in which the content and location of the error are known to the video monitoring device. Count
An evaluation value is calculated by weighting the obtained number of the first event, the number of the second event, and the number of the third event, and the threshold value group is calculated based on the obtained evaluation value. Is preferable to update.

  In the learning, it is preferable to count the number of the first event, the number of the second event, and the number of the third event while individually changing the thresholds of the threshold group.

The video monitoring apparatus of the second aspect of the present invention is a video monitoring apparatus for inputting a video signal corresponding to video and / or audio to be monitored and detecting that an error has occurred in a monitoring item,
A threshold value storage unit that stores a threshold value of a plurality of parameters that change according to the video signal, in which at least two of the plurality of parameters are associated with each other;
All of the plurality of parameters, when it exceeds the threshold determined according to the monitoring item, it is determined that the error has occurred, a determination unit for issuing an alarm,
A learning unit that performs learning in parallel with the video and / or audio monitoring operation,
An updating unit that updates the threshold value based on a learning result in the learning unit,
When the determination unit detects the occurrence of the error, a video storage unit for storing the video signal before and after the occurrence of the error, and
The learning unit learns by analyzing a result of reporting the error.

  According to the present invention, by analyzing the report result of the error and classifying it into, for example, “correct alarm”, “unnecessary alarm, and“ slip through ”, the learning unit performs learning, and the threshold value is more appropriate. Since it can be updated to any value, the accuracy of error detection is improved.

  Furthermore, when the video signal corresponding to the video and / or the audio to be monitored is input, the learning unit issues the alarm at a constant time interval for an error to be issued. An image for inspection in which the number of first events and the number of second events in which the alarm has been issued for an error for which an alarm should not be issued are counted, and the content and location of the error are known. When a signal is input, the number of the third event in which the alarm is not issued is counted for the error for which the alarm is issued, and the number of the obtained first event and the obtained number of the second event are counted. It is preferable that an evaluation value is calculated by weighting and calculating the number of cases and the number of the third event, and the threshold values stored in the threshold value storage unit are ranked based on the obtained evaluation value.

  Further, it is preferable that the learning unit counts the number of the first event, the number of the second event, and the number of the third event while individually changing the thresholds of the monitoring items.

但し、
Ｗｔ（ｍ）は前記第１事象の重み付け
Ｗｆ（ｍ）は前記第２事象の重み付け
Ｗｈ（ｍ）は前記第３事象の重み付けFurther, the learning unit is configured such that, for the m-th video signal of the M video signals, the alarm is issued for an error for which an alarm should be issued between times t1 and t2. Let T (m) be the number of events, F (m) be the number of second events for which the alarm was issued for an error for which an alarm should not be issued, and for an error for which an alarm should be issued. When the number of third events in which the alarm is not issued is H (m), it is preferable to obtain the evaluation value A (m) according to the following evaluation function.

However,
Wt (m) is the weighting of the first event Wf (m) is the weighting of the second event Wh (m) is the weighting of the third event

  Furthermore, it is preferable that the video storage section stores the input video signal by cutting it out at a predetermined timing and a predetermined length.

  Further, it is preferable to have a display unit that displays the learning result of the learning unit and reproduces video and / or audio based on the video signal before and after the occurrence of the error.

  Furthermore, it is preferable that the updating unit lists new threshold candidates that replace the current threshold in a prioritized manner.

  In the present specification, the “voice” includes not only human and animal voices but also sounds. The "video signal" includes the case of either a video signal or an audio signal. “Content” is the content of a video. Specifically, the major categories of contents include "movies", "drama", "variety", "sports", "documentary", "entertainment", and "animation". On the other hand, as a subcategory of contents, for example, "sports" includes "baseball", "soccer", "volleyball", "judo", "sumo", "athletics" and the like. “Learning” refers to associating a monitoring result in which a video and / or audio flowing in real time is watched by a monitor with a result of monitoring a video signal corresponding to the same video and / or audio with a video monitoring device, and setting it as a threshold value. Including feedback. “Interrelated” means that adjusting one parameter affects the other parameter. "Exceeding the threshold" includes both cases of exceeding the threshold and falling below the threshold.

  According to the present invention, it is possible to provide a method of adjusting an image monitoring apparatus that appropriately detects an image and / or an error occurring in a digital image signal according to content. Further, according to the present invention, it is possible to provide an image monitoring device that appropriately detects an image and / or an error through learning in a digital image signal.

3 is a block diagram showing the video monitoring device 100. FIG. FIG. 7 is a diagram showing an example of inspection items in the video / audio monitoring unit 103 displayed on the display unit 107. FIG. 7 is a diagram showing an example of alarm information displayed on a display unit 107. FIG. 7 is a diagram showing an example of parameters for each monitoring item displayed on the display unit 107. FIG. 10 is a diagram showing an example in which correct answer alarms, unnecessary alarms, and slip-throughs that occur when error detection is performed using the same threshold value are summarized for each monitoring item and content. It is a figure which shows the list | wrist of the example ((a) before update, (b) after update) of the several groups of the threshold value when such update is performed multiple times in the inspection item freeze regarding a certain content.

  An image monitoring apparatus and an adjusting method thereof according to the present embodiment will be described with reference to the drawings. FIG. 1 is a block diagram showing a video monitoring device 100. In FIG. 1, a video monitoring apparatus 100 inputs a video signal for inspection and a main input unit 101 for inputting a video signal corresponding to video and / or audio to be monitored (hereinafter referred to as an inspection target video signal). A sub-input unit 102, a video / audio monitoring unit (determination unit) 103 having a built-in memory for storing a threshold value, a video / audio clip storage unit (video storage unit) 104, an alarm output unit 105, and a built-in memory (threshold value). Parameter optimization learning unit (learning unit / updating unit) 106 including a storage unit 106a, a display unit 107 such as a display through which the information displayed by the monitor MN can be visually recognized, and the monitor MN can input information. And a monitor input unit 108 such as a keyboard.

  The monitoring operation of the video monitoring device 100 will be described. FIG. 2 is a diagram showing an example of inspection items in the video / audio monitoring unit 103 displayed on the display unit 107. The video signals to be monitored are video signals of all formats such as SDI signal, file, IP format, and HDMI (registered trademark). While watching the monitoring items displayed on the display unit 107, the monitor MN changes the box corresponding to each monitoring item via the monitor input unit 108, and “inspect” or “off” for each monitoring item. You can choose either of The inspection item for which the monitor selects “inspection” will be monitored by the video monitoring apparatus 100, but the inspection item for which “off” has been selected is not monitored.

  Video monitoring items include "freeze", "blackout", "blocking freeze", "blocking blackout", "block noise", "flashing red", "blinking brightness", "scene change", and "video reversal". , "Line noise", "Cut point abnormality", "Time code discontinuity". On the other hand, the monitoring items regarding voice include “mute”, “blocking mute”, “push sound (Audio Pop Noise)”, “sound skip”, “voice noise”, “loudness”, and “true peak”. The inspection items are not limited to these.

  As a premise of the monitoring operation, it is assumed that the video / audio monitoring unit 103 is set with a plurality of thresholds (threshold group, which will be described later in detail) for each monitoring item. Referring to FIG. 1, when the video monitoring device 100 inputs an inspection target video signal from the main input unit 101, an image frame, pixel data, and audio sampling data are extracted from the input inspection target video signal to obtain a video image. The parameters are transmitted to the voice monitoring unit 103, and parameters corresponding to various sensitivity adjustments and durations applicable to the monitoring algorithm are calculated here.

  According to the monitoring algorithm, the video / audio monitoring unit 103 compares the obtained plurality of parameters with the threshold value corresponding to the monitoring item, and if all the parameters exceed the threshold value, it detects an error corresponding to the monitoring item. The determination is made and an alarm signal associated with the monitoring item is output to the alarm output unit 105. The alarm output unit 105 inputs the alarm information including the error occurrence time, the content of the detected error, and the severity thereof to the display unit 107 according to the input alarm signal, so that the display unit 107 displays the alarm information. It is displayed and can be visually recognized by the monitor MN. Further, the video / audio monitoring unit 103 cuts out video signals before and after the parameter exceeds the threshold value and stores them in the video / audio clip storage unit 104.

  FIG. 3 is a diagram showing an example of alarm information displayed on the display unit 107. In FIG. 3, detected errors and their contents are displayed in time series in the lower part of the screen, and “category” indicating that the error is video or audio in the upper part of the screen, and the level of the error is normal or serious. The “class” shown, the “inspection item” showing the type of error, and the “number of occurrences” of the error are displayed together. When the monitor MN selects one of the errors via the monitor input unit 108, corresponding video signals before and after the error are read from the video / audio clip storage unit 104 and displayed on the display unit 107. Is entered. As a result, the video and / or audio before and after the error is output from the display unit 107, so that the monitor MN can actually visually recognize or hear the content of the error.

  By the way, when the monitor MN compares the content of the detection error displayed on the display unit 107 with the actual video and / or audio, the alarm is issued for the error for which the alarm should be issued (correct answer). An alarm is issued, an alarm is issued for an error that should not be issued (called an unnecessary alarm), and an alarm is not issued for an error that should be issued (pass through). Can be recognized respectively. Originally, it is ideal that all of the alarms issued by the video monitoring device 100 when an error is detected are correct alarms, but in reality, unnecessary alarms and omissions occur. This is because the error detection standard by the machine monitoring of the video / audio monitoring unit 103 does not exactly match the error detection standard by the monitoring by the monitor MN. Therefore, unless the error detection by machine monitoring approaches the error detection by human monitoring, it becomes difficult to monitor the video monitoring apparatus 100 alone.

  Therefore, in order to bring the error detection standard by the machine monitoring of the video / audio monitoring unit 103 closer to the error detection standard by the monitoring of the monitor MN, the threshold value of the parameter set for each monitoring item is changed. This is called threshold tuning. Here, if there are m monitoring items and the number of parameters is I (m), the number of thresholds to be tuned is 1 · I (1) + 2 · I (2) + ... + m · It becomes I (m), and it can be seen that the number increases as the number of monitoring items increases.

  FIG. 4 is a diagram showing an example of parameters displayed on the display unit 107 for each monitoring item. The observer MN can input an arbitrary numerical value for each parameter in the box corresponding to the parameter via the observer input unit 108 while watching the parameter displayed on the display unit 107. There is.

  As an example, four parameters of “sensitivity threshold (activity)”, “sensitivity threshold (noise)”, “time threshold (start)”, and “time threshold (end)” are set as parameters corresponding to the inspection item “freeze”. There is one. The "graph scale" represents the scale of the graph for display and is not a parameter here. That is, in order to detect the video freeze as an error, the thresholds (threshold groups) of the four parameters must be set appropriately. However, the “sensitivity threshold (activity)” and the “sensitivity threshold (noise)” are the upper limit value and the lower limit value of the value distributed for each small block included in one video frame as parameters, and are related to each other. It can be said that it fits. Such parameters are disclosed in detail in WO2015-059782. The "time threshold value (start)" and the "time threshold value (end)" indicate the length of the period in which it is determined that the video is phrased, and are related to each other. Therefore, when either one of the thresholds is adjusted, the freeze cannot be properly detected unless the other threshold is also changed.

  If an appropriate threshold value is not input for all the inspection items, unnecessary alarms and omissions may occur when an error is detected. However, if the number of inspection items is large as shown in FIG. It is difficult to determine a proper threshold for. Therefore, it is desirable to determine a default threshold value (initial value) in advance. Such a default threshold value can be stored in a built-in memory of the video / audio monitoring unit 103 and used. However, according to the examination result of the present inventor, when the same threshold value is used for all contents, unnecessary alarms or skips are reduced in one content, but unnecessary alarms or skips occur in another content. I understood.

  FIG. 5 is a diagram showing an example in which correct answer alarms, unnecessary alarms, and slip-throughs that occur when error detection is performed using the same threshold value are grouped by monitoring item and content. From the example of FIG. 5, it can be seen that the correct alarm, the unnecessary alarm, and the slip-through occurrence frequency differ for each content. Based on this examination result, the present inventor has found that the default threshold value may be determined according to the content. Such a default threshold can be obtained from simulations and accumulated experience.

  On the other hand, if a default threshold value is determined according to the content, it is possible to reduce unnecessary alarms and omissions to some extent, but it is not always optimal. Even if the contents are the same, the frequency of unnecessary alarms or passing through may change depending on the reception state and the like. Therefore, in parallel with the monitoring operation, it is preferable to make the video monitoring apparatus 100 learn about the currently input video signal to determine whether or not the default threshold value is appropriate, and further update if not appropriate. I can say. As a result, the number of correct alarms can be increased and the frequency of passing through unnecessary alarms can be reduced. The updated threshold value can be newly set as the default threshold value of the content.

(Learning mode)
The learning function of the video monitoring device 100 will be described below. In the following learning example, it is possible to determine which is better than the default threshold when there is a threshold candidate to be changed. In the video monitoring device 100 of FIG. 1, the monitor MN sets a predetermined learning period from the monitor input unit 108 while the video and / or audio is being monitored. Then, the video monitoring device 100 can perform learning during this learning period.

  Specifically, the monitoring operation is performed using the default threshold value for one monitoring item in M video signals of the same type of content. First, during the learning period (time t1 to t2, for example, time unit such as hour, day, week, month, etc.), the error detected by the video monitoring device 100 is displayed on the display unit 107 as shown in FIG. . Since the error displayed as shown in FIG. 3 may include an inappropriate error, it is necessary to check the adequacy of the error for learning.

  Here, the observer MN reads one of the video signals before and after the error stored in the video / audio clip accumulating unit 104 by specifying one of the errors displayed in FIG. Video and / or audio can be viewed on the display unit 107. As a result, the observer MN can count the number of correct alarms and the number of unnecessary alarms that occurred between times t1 and t2. The video signal before and after each error stored in the video / audio clip accumulating unit 104 has a length of about 3 seconds, so that it does not take much time to view and the check is completed in a short time. The burden is small.

  However, the number of slip-throughs cannot be counted in the above check. Therefore, a video signal for inspection, in which the content of the error and the location (time) of the error are known, which has been prepared in advance, is input to the video / audio monitoring unit 103 via the sub-input unit 102, and the same threshold value as described above is input. The video / audio monitoring unit 103 detects an error using. As a result, when the alarm information similar to that shown in FIG. 3 is obtained, the observer MN obtains the number of correct alarms and the number of slip-throughs by collating the content of the known error with the occurrence location. be able to. The number of slip-throughs is extracted and used here. Since the inspection video signal has a length corresponding to a viewing time of about tens of minutes for each content, the inspection does not take time and effort. The number of correct alarms, the number of unnecessary alarms, and the number of passing through obtained for each of the M video signals as described above are learned by the monitor MN from the monitor input unit 108 for parameter optimization. Input to the unit 106.

但し、
Ｗｔ（ｍ）は正解アラームの重み付け
Ｗｆ（ｍ）は不要アラームの重み付け
Ｗｈ（ｍ）はすり抜けの重み付けHere, the number of correct alarms (first event) in the m-th video signal is T (m), the number of unnecessary alarms (second event) is F (m), and the number of slip-throughs (third event) is Letting H (m), the parameter optimization learning unit 106 obtains the evaluation value A (m) according to the following evaluation function.

However,
Wt (m) is the correct answer weighting Wf (m) is the unnecessary alarm weighting Wh (m) is the slip-through weighting

  In error detection, it is generally preferable that the number of correct alarms is large, the number of unnecessary alarms is small, and it is preferable that slip-through is eliminated as much as possible. Therefore, it is preferable to set Wt (m)> 0> Wf (m)> Wh (m) as the weight stored in the memory 106a. For example, Wt (m) = + 3 and Wf (m) = − 1. And Wh (m) = − 5. However, weighting may be fixed for the same type of content. According to this evaluation function, the higher the obtained evaluation value A (m), the more accurate the error detection, and the lower the value, the more inaccurate the error detection.

  Then, the monitor MN replaces the default threshold value with a threshold value desired to be changed through the monitor input unit 108, and executes the same monitoring operation as above using the video signal of the same content. To obtain the evaluation value. If the evaluation value of the monitoring operation using the default threshold is equal to or higher than the evaluation value of the monitoring operation using the threshold desired to be changed, the parameter optimization learning unit 106 continues to use the default threshold for the content. As a matter of fact, the stored default threshold value is not changed. On the other hand, if the evaluation value of the monitoring operation using the default threshold value is lower than the evaluation value of the monitoring operation using the threshold value that is desired to be changed, the parameter optimization learning unit 106 determines that the default threshold value for the content. It is determined that the change is necessary, and the change is replaced with a desired threshold value, that is, learning is performed. Thereby, the accuracy of error detection can be further improved.

  FIG. 6 is a diagram showing a list of an example ((a) before update, (b) after update) of a plurality of groups of thresholds when such update is performed a plurality of times in the inspection item: freeze regarding a certain content. . A plurality of threshold groups shown in FIG. 6 are listed and stored in the built-in memory 106a, and displayed on the display unit 107 as needed so that the supervisor MN can confirm them.

  According to the pre-update list of FIG. 6A, the default threshold value group is currently listed as the highest priority level 1, but the evaluation value A (m) is “85”. On the other hand, the evaluation value A (m) of the threshold group newly evaluated as the first candidate is “92”, and the evaluation value A (m) of the threshold group newly evaluated as the second candidate is “81”. Suppose there is.

  In this case, the evaluation value A (m)-“92” of the first candidate threshold value group is higher than the evaluation value A (m) “85” of the default threshold value group up to now, so the parameter optimization learning unit 106. However, as shown in FIG. 6B, the priority is updated to 1 to replace it with a new default threshold value group, which is transmitted to the video / audio monitoring unit 103 for monitoring. As a result, the default threshold group up to now is moved down to the priority level 2.

  On the other hand, the evaluation value A (m) “81” of the second candidate threshold value group is lower than the evaluation value A (m) “85” of the replaced default threshold value group, and thus the parameter optimization learning unit 106. However, the priority is set to 3 by updating, so that it is listed up in a lower order. The list is stored in the memory 106a of the parameter optimization learning unit 106 and used. As a result, a threshold value group having a high evaluation value A (m) for the content can be set as a default threshold value group. However, if the contents are different, the priority order may be different even if the threshold values use the same value. It should be noted that such updating of the threshold value group may be automatically performed by the parameter optimization learning unit 106, or may be performed after waiting for permission of the supervisor MN. Alternatively, regardless of the value of the evaluation value A (m), the threshold specified by the monitor MN via the monitor input unit 108 can be set as the default threshold.

  Note that the first candidate, the second candidate, ... Are decided, but the monitor MN inputs an arbitrary value from the monitor input unit 108 to the parameter optimization learning unit 106. The evaluation value may be calculated after obtaining the number of alarms, the number of unnecessary alarms, and the number of slip-throughs. Alternatively, the parameter optimization learning unit 106 individually changes the threshold value by + 5% or −5% with respect to the default threshold value group, and each time, the number of correct answer alarms, the number of unnecessary alarms, and the skipping are skipped. Alternatively, the evaluation value may be calculated after the number of is calculated. Evaluation values can be similarly obtained for other monitoring items.

  The inspection video signal can be input between the input of the monitoring target video signal, or the inspection can be performed in the background in parallel with the input of the monitoring target video signal. This allows error detection to be performed using the same threshold. Further, the video / audio storage clip unit 104 cuts the monitoring target video signal input from the main input unit 101 at a predetermined timing over a predetermined length, and associates it with the monitoring result performed by the monitor MN. Alternatively, it may be stored as a video signal for inspection in a memory (not shown). The stored inspection video signal is input from the sub-input unit 102 at a necessary timing and used for updating the threshold value as described above.

  INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide a method of adjusting an image monitoring apparatus that appropriately detects an image and / or an error generated in a digital image signal according to content, and appropriately adjusts the image and / or the error in the digital image signal through learning. It is possible to provide a video monitoring device for detecting the.

100 video monitoring device 101 main input unit 102 sub-input unit 103 video / audio monitoring unit 104 video / audio clip storage unit 105 alarm output unit 106 parameter optimization learning unit 106a built-in memory 107 display unit 108 monitor input unit MN monitor

Claims (11)

In a method of adjusting a video monitoring device, which inputs a video signal corresponding to video and / or audio to be monitored and detects that an error has occurred in a monitoring item,
The video monitoring device is configured to determine that the error has occurred when all of the plurality of parameters that change according to the video signal exceed a threshold determined according to the monitoring item. , At least two of the plurality of parameters are interrelated,
Each of the thresholds of the plurality of parameters is determined and a plurality of threshold groups are created and stored in a memory,
An adjusting method for a video surveillance device, characterized in that any one of the sets of threshold values stored in the memory is selected in accordance with a video and / or audio content to be monitored.   The video monitoring device performs learning in parallel with the video and / or audio monitoring operation, updates at least one threshold value of the threshold value group based on the learning result, and stores the updated threshold value group in a memory. The method for adjusting a video surveillance device according to claim 1, further comprising: The video monitoring device is designed to issue an alarm when an error is detected,
In the learning, a video signal corresponding to the video and / or audio to be monitored is input, and at a constant time interval, the first event in which the alarm is issued for the error for which the alarm should be issued. Counting the number of cases and the number of second events in which the alarm is issued for an error that should not be issued, respectively.
Further, the number of third events in which the alarm is not issued for an error for which an alarm is to be issued by inputting an inspection video signal in which the content and location of the error are known to the video monitoring device. Count
An evaluation value is calculated by weighting the obtained number of the first event, the number of the second event, and the number of the third event, and the threshold value group is calculated based on the obtained evaluation value. The method for adjusting a video surveillance device according to claim 2, wherein   In the learning, while counting the thresholds of the threshold group individually, the number of the first event, the number of the second event, and the number of the third event are counted. A method for adjusting the described video surveillance device. In a video monitoring device that inputs a video signal corresponding to video and / or audio to be monitored and detects that an error has occurred in a monitoring item,
A threshold value storage unit that stores a threshold value of a plurality of parameters that change according to the video signal, in which at least two of the plurality of parameters are associated with each other;
When all of the plurality of parameters exceed the threshold value determined according to the monitoring item, it is determined that the error has occurred, a determination unit for issuing an alarm,
A learning unit that performs learning in parallel with the video and / or audio monitoring operation,
An updating unit that updates the threshold value based on a learning result in the learning unit,
When the determination unit detects the occurrence of the error, a video storage unit for storing the video signal before and after the occurrence of the error, and
An image monitoring apparatus, wherein the learning unit learns by analyzing a report result of the error.   The learning unit is configured such that, when a video signal corresponding to a video and / or audio to be monitored is input, the first alarm is issued for an error for which an alarm should be issued at constant time intervals. The number of events and the number of second events in which the alarm is issued for an error for which an alarm should not be issued are counted, and a video signal for inspection in which the content of the error and the location of the error are known is counted. When input, the number of the third events in which the alarm was not issued for the error for which the alarm was issued is counted, and the obtained number of the first events and the number of the second events are obtained. , The number of the third event is weighted to calculate an evaluation value, and the threshold values stored in the threshold value storage unit are ranked based on the obtained evaluation value. Video surveillance described in 5 Location.   7. The learning unit counts the number of the first event, the number of the second event, and the number of the third event while individually changing the thresholds of the monitoring items. The video surveillance device described in. The learning unit detects, for the m-th video signal of the M video signals, between the times t1 and t2 of the first event in which the alarm is issued for an error for which the alarm is issued. Let T (m) be the number of events, F (m) be the number of second events in which the alarm was issued for an error for which an alarm should not be issued, and the alarm for the error for which an alarm should be issued. The video surveillance device according to claim 6 or 7, wherein an evaluation value A (m) is obtained in accordance with the following evaluation function, where H (m) is the number of the third events in which is not issued. .

However,
Wt (m) is the weighting of the first event Wf (m) is the weighting of the second event Wh (m) is the weighting of the third event   9. The video monitoring device according to claim 6, wherein the updating unit lists the threshold value for each monitoring item according to the priority order.   10. The video monitoring apparatus according to claim 5, wherein the video storage unit stores the input video signal by cutting it out at a predetermined timing and a predetermined length.   11. The display unit according to claim 5, further comprising a display unit for displaying a learning result of the learning unit and reproducing video and / or audio based on a video signal before and after the occurrence of the error. Video surveillance device.

Images