JPH11203567A - Image processor for monitor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processor for a monitor to reduce unnecessary detection and to surely alarm on a normal object. SOLUTION: This image processor is provided with a background difference extracting part 6 to compare data of the present image of an image of a district to be monitored with background image data and to extract a multilevel variation area equivalent to a mobile object in the image, a featured value calculating part 9 to calculate a prescribed featured value to indicate features of the variation area, a featured value of object to be alarmed storage part 10 to store a featured value parameter as a range of set value of the featured value of the prescribed object to be alarmed, a recognition processing means 11 to recognize that the featured value at the featured value calculating part is suitable for that of the normal object to be alarmed based on the featured value parameter, an alarm processing means 13 to raise an alarm according to recognition and a most frequent object to be alarmed discrimination and processing means 14 to discriminate the most frequent object to be alarmed with high occurrence frequency among the objects to be alarmed from states of occurrence distribution and to correct a center value of the featured value parameter in the featured value of object to be alarmed storage part to be equal to the center value of the featured value of the most frequent object to be alarmed.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a monitoring image processing apparatus for monitoring based on images obtained by a television camera or the like.

[0002]

2. Description of the Related Art FIG. 11 is a diagram showing the configuration of a conventional monitoring image processing apparatus of this type. In the figure, reference numeral 1 denotes a video input unit which receives a video processed by the monitoring image processing apparatus, and 2 denotes an A / A which quantizes a video input from the video input unit 1.
An A / D converter, such as a D converter, 3 is a current image storage unit that receives the output from the A / D converter 2 and stores the current video data, and 4 receives the output from the A / D converter 2. The background image storage unit 5 that stores the "background image" for comparison, which is created by performing the above calculation, is a background update unit that calculates the data of the background image storage unit 4 according to a certain rule and performs "background update". For example, the average of the past five current images is set as a background image.

[0003] Reference numeral 6 denotes a background difference extracting unit that compares the two images of the current image storing unit 3 and the background image storing unit 4 and calculates the difference, that is, "background difference" data. This is a threshold value calculation unit that calculates a threshold value for binarization. In general, an optimum threshold value is obtained based on "background difference" data. Reference numeral 8 denotes a binarizing unit for binarizing "background difference" data with a threshold value obtained by the threshold value calculating unit 7.

[0004] Reference numeral 9 denotes a feature value calculating unit which receives the binary data output from the binarizing unit 8 and calculates a "feature value" of the binary data.
Is an alerting target feature amount storage unit that stores in advance “feature amount parameters” of “regular alerting targets” (for example, persons, vehicles, etc.). 11 is a recognition processing means for comparing “feature value” data output from the feature value calculation unit 9 with “feature value parameter” of the notification target feature value storage unit 10 to determine whether or not “event” has occurred; Reference numeral 12 denotes a notification for notifying the discovery of an "event". Reference numeral 13 denotes an alarm processing unit that receives the notification 12 and issues an alarm (for example, a BEEP sound).

FIG. 12 is a flowchart showing the operation of the recognition processing means 11 of FIG. 11 in more detail.

Hereinafter, a typical example of the conventional operation will be described. In FIG. 11, a video signal externally supplied to the device is received by a video input unit 1 and sent to an A / D conversion unit 2. The A / D converter 2 converts the received video signal into digital data for subsequent processing. It is common to convert to 8-bit multi-value data per pixel.

The current image storage section 3 stores the input current image data and prepares for the subsequent processing. The background image storage unit 4 stores a “background image” created based on the input current image data and prepares for subsequent processing. The background updating unit 5 periodically performs a "background updating" process on the background images stored in the background image storing unit 4 according to a certain rule. This is generally calculated based on the current image that is periodically sampled. For example, when the background image is an average value of the current images of the past five scenes, or when the image is a frame image a predetermined time ago. There are many.

The background difference extracting section 6 compares the outputs of the current image storing section 3 and the background image storing section 4, finds a difference for each pixel, and outputs the difference data as a frame image. The difference data generally represents a portion of the video where the luminance has changed, that is, a “multi-value change area”. Normally, moving objects such as people and vehicles are mainly used, but in rare cases, the brightness of non-reporting targets such as "appearance of shadows caused by the shadows of sunlight" and "reflection of headlights of cars at night". It may output a change.

A threshold calculator 7 calculates a threshold for binarizing the “multi-value change area” sent from the background difference extractor 6. The “multi-value change area” is multi-value data, and is not appropriate for recognition processing as it is. In general, binarization and conversion into binary data are also performed in order to make subsequent determinations at high speed. The binarizing unit 8 binarizes the multi-value change area using the threshold value. As a result, hereafter, the change area becomes a “binary change area. However, depending on the device, the“ multi-value change area ”may be directly recognized and processed without performing this binarization, so this is a reference.

The "change area" will be described with reference to FIG. FIG. 13 is an image of a normal main gate. Now, a person is moving in the image. In this case, the position of the moving person becomes the “change area” as it is. FIG. 14 is an image in which the superimposition in which the “change area” is dyed black is overlapped.

The feature value calculating section 9 receives the input "change area data" and calculates the "feature value" of the "change area". Now, as an example, the “feature amount”, “area”, “length and width”, and “speed” are used. The notification target feature amount storage unit 10 includes:
The feature amount of an object (for example, a person or a vehicle) which is set as a “genuine alarm target” by this device is stored in advance.

The recognition processing means 11 receives the “feature amount” data from the feature amount calculation unit 9, and the “feature amount” is stored in the notification target feature amount storage unit 10 for the “feature amount parameter” of the notification target. Is determined. If it is determined that they are close to each other, an “event” is found, and the notification 12 is output.

In the "recognition process", recognition is performed according to the following procedure shown in FIG. However, this is only an example. In the change area integration processing in step S1, the change areas are observed, and the change areas that are considered to belong to the same moving object are integrated and combined into one change area. In steps S2 to S5, it is determined whether or not the change area integrated in step S1 is an alert target. Here, the "feature value" of the "change area" is compared with the "feature value parameter" of the "normal notification target" stored in the predetermined notification target feature value storage unit 10.
Judge whether or not matches. Here, if a predetermined condition is met, it is regarded as “a regular alarm target”.

Here, the following judgment is made as an example. In step S2, it is determined whether the "change area" exists continuously for a certain period of time. In step S3, it is determined whether or not the “change area” has a predetermined area. In step S4,
It is checked whether the “change area” has a predetermined vertical and horizontal dimension. Then, in a step S5, it is checked whether or not the "change area" has a predetermined moving speed. The “area”, “length and width”, and “moving speed” are generally referred to by calculating an average during a certain time in step S2.

[0015] On the contrary, however, here, temporarily, "the subject of the regular alert"
Even if a completely different object appears (for example, the appearance of a shadow of a building due to the shade of sunlight), if the features match here,
The device determines that the event is an “event”
Leads to. Such false alarms by non-alarm targets,
Hereinafter, it is referred to as “unnecessary detection”. The “event”, whether “unnecessary detection” or normal detection, is transmitted to the alarm processing unit 13 in FIG. 11 by the alarm target discovery in step S6, where “alarm” processing is performed. Usually, it is a process of sending a specific signal to a PC (personal computer) or the like at an upper level, or sounding a buzzer of the apparatus.

If the conditions are not met, it is determined in step S7 that there is no alarming target, and the "event" notification 12
Is not output. Therefore, in this case, the alarm processing means 13 does not report anything, so that no alarm is issued. Note that the recognition processing means 11 and the alert processing means 13 are generally processed by software.

[0017]

Since the conventional monitoring image processing apparatus is configured as described above, the "change area"
Is a non-reporting target (for example, the appearance of a shadow of a building due to the shade of sunlight), a report is issued (unnecessary detection) if the conditions in steps S1 to S5 in FIG. 12 are met. Especially when conducting monitoring work outdoors, in addition to the appearance of shadows of buildings due to the shading of sunlight, reflection of headlights of cars at night and shaking trees (hereinafter referred to as unnecessary objects) are unnecessary In many cases, detection occurs.
The conventional monitoring image processing apparatus has the above problems.

The present invention has been made to solve the above-described problem, and updates feature parameters and algorithms of various parts of the apparatus according to a constant method in accordance with such a change region of a normal alert target. It is an object of the present invention to obtain a monitoring image processing device that facilitates exclusion of unnecessary detection by an unnecessary detection target object and, at the same time, surely issues a notification to a regular notification target such as a “person”.

[0019]

SUMMARY OF THE INVENTION In view of the above objects, the present invention provides a current image storage unit for temporarily recording the current image data of an image of a monitoring target area, and records background image data of the image. Background image storage unit, and background difference extraction for comparing the current image data of the current image storage unit with the background image data of the background image storage unit and extracting a multi-value change area corresponding to a moving object or a change object in the image as a difference. Unit, a feature value calculating unit that calculates a predetermined feature value indicating the feature of the change area, and a report target feature that stores a feature parameter which is a preset value range of the feature value of the report target. Quantity storage unit, and recognition processing means for recognizing that the feature amount in the feature amount calculation unit corresponds to a regular one to be issued based on the feature amount parameter of the notification target feature amount storage unit, According to the recognition of the recognition processing means, An alarm processing means for issuing an alarm when an alarm is to be issued; and a high-frequency alarm object having a high frequency of occurrence among the alarm targets is distinguished from the others based on a state of occurrence distribution. The most frequent report target identification processing means for correcting the central value of the feature value parameter in the central value to the central value of the feature value of the most frequent report target is provided.

Further, in the present invention, the most frequently reported object identification processing means further selects, from the distribution of occurrence, a feature amount of the most frequently reported object that has a large variation, and stores the most frequently reported object feature amount. The monitoring image processing apparatus is characterized in that correction is performed so as to exclude the characteristic amount parameter in the unit from setting.

Further, the present invention estimates a video input section for inputting a video signal serving as an image of the monitoring target area, and an area on the image in which invalid change areas frequently occur from data of the feature amount of the most frequently reported object. And a mask range calculation processing means for causing the video input unit to mask the area of the video.

According to the present invention, there is provided a monitoring system further comprising a recognition range calculation processing means for estimating a region on a video which frequently occurs and which frequently occurs, and instructing the recognition processing unit to set only the region as a monitoring range. In the image processing device.

Further, the present invention performs the integration of the changed areas before the recognition processing means recognizes whether or not the information corresponds to the legitimate notification target, and performs the area integration with respect to the area after the area integration. The monitoring unit further includes an area integration operation processing unit that infers the necessity of area integration from the ratio of the area of the maximum area change area among the change areas, and prohibits the integration of the change areas by the recognition processing unit. In the image processing device.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below according to each embodiment. Embodiment 1 FIG. FIG. 1 is a diagram showing a configuration of a monitoring image processing apparatus according to an embodiment of the present invention. In the figure,
The same or equivalent parts as those of the conventional one are indicated by the same reference numerals,
Description is omitted. Reference numeral 14 denotes a first most-frequently-reported object identification processing unit that identifies the “most-frequently-reported object” from others. Reference numeral 15 denotes a trigger input for instructing the first most frequently reported object identification processing means 14 to perform correction. The feature parameter of the report target feature storage unit 10 receives a notification from the most frequently reported subject identification processing unit 14 and is updated as appropriate. In general, the first most frequently reported object identification processing means 14 is also processed by software.

FIG. 2 shows the first most frequent notification target identification processing means 1.
4 is a flowchart showing the operation of FIG. 4 in detail. The operation of the recognition processing means 11 is basically the same as that shown in the flowchart of FIG.

Next, the operation will be described. The operations of the video input unit 1 to the feature amount calculation unit 9 and the recognition processing unit 11 to the notification processing unit 13 in FIG. Upon receiving the notification 12 indicating the occurrence of the event, the most frequently reported object identification processing means 14 observes the feature amount at that time and identifies the "feature amount" of the "most frequently reported object".

Hereinafter, the behavior example will be described in detail. However, what is described here is an example of identifying the “feature amount” of the “most frequently reported object”. In FIG. 12, the specific contents of the “feature amount” are “existing time” and “area” of the “change area”.
Although “vertical and horizontal dimensions”, “velocity”, and the like are described, here, for simplicity, only the “area” in the “feature parameter” will be described. However, the method is the same for the other “feature parameter” and is identified in the same manner.

Next, the operation of the most frequently reported object identification processing means 14 will be described with reference to FIG. Upon receiving the notification 12 of the event discovery, the most frequently reported target identification processing means 14 refers to (stores) the area value at the time of reporting in step S1. It is plotted in a histogram in step S2. In this histogram, the horizontal axis is the area value, and the vertical axis is the number. If there is no instruction to perform the correction, the operation ends here. Only plotted on histogram.

If there is a trigger input 15 for performing the correction, the process branches at step S3 to perform an operation for calculating the area of the "most frequently reported object". Specifically, the procedure is performed according to the procedure in step S4. With respect to a certain area value X, for example, the area set in step S3 of FIG. 12, the total number of the areas is calculated within the range of the width n (n is a constant determined separately) of the horizontal axis of the histogram with the center as the center. The area value X at which is maximized is determined. The area value X determines the "area of the" most frequently reported object ""
It is. Furthermore, when the entire area of the histogram is statistically processed with the “area of“ most frequently reported ”” regarded as the average value, the standard deviation σ
Is found.

The result is directly sent to the feature quantity storage unit 10 at step S5. Here, the “feature parameter” of the “normal alarm target” is originally stored, but in this embodiment, it is corrected based on the above result. The area of the “feature amount parameter” of the “normal notification target” is usually set by “center value” and “width”. For example, “area = 10000 pixels ± 100
The setting is such as “0 pixel”. The correction is this "center value"
Is replaced with the “area of“ most frequently reported ”” obtained above, and when the “width” is also changed, the “width” is replaced with, for example, “3σ of“ most frequently reported ”” or the like. Done. Thus, the updating of the “feature parameter” is completed.

Although an example of the area has been described above, the same applies to other “feature amount parameters”. A histogram is similarly created for “existing time”, “vertical and horizontal dimensions”, “speed”, and the like of the “change area”, and “existing time of“ most frequently reported objects ”,“ vertically and horizontally dimension of “most frequently reported objects” ”,“ Find the "speed of the most frequently reported object", etc., and replace the "center value" and "width".

As described above, according to this embodiment,
Every time there is an instruction to perform the correction, the value of the “feature parameter” is updated, and the updated result is the “most frequently reported object” at that time. As a result, the “feature parameter” is changed to “ The most frequently reported object "" is tracked and updated. As a result, the “feature parameter” is modified so as to gradually respond to only the “most frequently reported object”.

For example, when the “area” is updated, for example, when the “normal alarm target” is “human”, an “unnecessary detection target” having a large area such as “car” is erroneously determined to be an event. It becomes difficult. Also, for example, by updating the “existence time”, for example, when the normal notification target is “human”, the “unnecessary detection target” having only the “change area” having no continuity such as “shaking of the vegetation” is obtained. On the other hand, it is difficult to judge the event by mistake.

For example, when the "aspect ratio" is updated, for example, when the "normal notification target" is "human", a horizontally long "unnecessary detection target" such as "dog" is erroneously determined to be an event. It is hard to be done. Also, for example, by updating the “speed”, it is difficult for an “unnecessary detection target” having a different speed, such as “bicycle”, to be erroneously determined to be an event when the “regular notification target” is “human”. Become. In addition, an alarm is reliably issued to a person who is a “genuine alarm target”.

As described above, there is provided an apparatus capable of reducing "unnecessary detection" and at the same time reliably issuing "regular notification target".

Embodiment 2 FIG. 3 is a diagram showing a configuration of a monitoring image processing apparatus according to another embodiment of the present invention. In the figure, the same or corresponding parts as those in the conventional and the above-described embodiments are denoted by the same reference numerals, and description thereof will be omitted. Reference numeral 14a denotes a second most-frequently-reported object identification processing means for determining conditions according to the reportable object. Reference numeral 15 denotes a trigger input for instructing the second most frequently-issued target identification processing means 14a to perform correction. The feature amount parameter of the notification target feature amount storage unit 10 receives a notification from the second most frequently reported target identification processing unit 14a and is updated as appropriate. Generally, the second most frequently-issued object identification processing means 14a is also processed by software.

FIG. 4 shows the second most frequent notification target identification processing means 1.
4 is a flowchart showing the operation of FIG. 4a in detail. The operation of the recognition processing means 11 is basically the same as that shown in the flowchart of FIG.

Next, the operation will be described. The operations of the video input unit 1 to the feature amount calculation unit 9 and the recognition processing unit 11 to the notification processing unit 13 in FIG. The second most frequently reported subject identification processing means 14 which has received the notification 12 notifying that an event has occurred.
a observes the “feature amount” at that time, and “the most frequently reported object”
The validity of the individual contents of the “feature parameter” is identified.

Hereinafter, an example of the behavior will be described in detail. However, what is described here is an example of identifying validity. The specific contents of the “feature parameter” include “existing time”, “area”, “length and width”, and “speed” of the “change area” in FIG. 12, but the description is simplified here. Therefore, only the “area” in the content of the “feature parameter” will be described. However, the method is the same for the other “feature parameter” and is identified in the same manner.

Next, the operation of the most frequently reported object identification processing means 14a will be described with reference to FIG. If the most frequent notification target identification processing means 14a receives the notification 12 of the event discovery first,
In step S1, the area value at the time of alarm is referenced (stored). The values are plotted on a histogram in step S2. In this histogram, the horizontal axis is the area value, and the vertical axis is the number. If there is no trigger input 15 for performing the correction, the operation is completed. Only plotted on histogram.

If there is a trigger input 15 for performing the correction, the process branches at step S3 to perform an operation of calculating the area of "the most frequently reported object". Specifically, steps S4 to S
6. Perform the procedure described in 6. With respect to a certain area value X, the total number of the area values is calculated within the range of the width n (n is a constant determined separately) of the horizontal axis of the histogram centered on the area value X, and the area value X that maximizes the number is calculated. The area value X is defined as “the area of“ the most frequently reported object ””. Furthermore, when the “area of“ most frequently reported ”” is regarded as an average value and the entire histogram is statistically processed, a standard deviation σ is obtained.

Next, here, the validity / invalidity of the “area” will be examined based on “the average value and σ of the“ most frequently reported objects ””.
Judgment as invalid is based on the following.・ "Average value of" most frequently reported "" was not clearly found. The maximum value is calculated at a plurality of locations, and there is no protruding portion that is entirely uniform.・ “Σ of“ most frequently reported ”” has exceeded m (m is a constant determined separately). The result is sent as it is to the notification target feature amount storage unit 10.

The notification target feature amount storage unit 10 originally stores
The “feature amount parameter” of the “normal notification target” is stored, but in this embodiment, it is corrected based on the above result. If it is valid, it operates as in the first embodiment.

If invalid, the operation is as follows. The area of the “feature amount parameter” of the “normal notification target” is usually set by “center value” and “width”. For example, “area = 10000 pixels ± 1000 pixels”
The setting is as follows. The correction is made by replacing the "width" with a theoretical maximum value. As described above, the area of the “feature parameter” falls within the “width” of any value of the “report target”, and has no practical meaning.

Thus, the area of the "feature parameter" is excluded from the recognition processing. That is, the process of step S3 in FIG. 12 is not performed thereafter. Thus, the updating of the “feature parameter” is completed.

Although an example of the area has been described above, the same applies to other “feature amount parameters”. A histogram is similarly created for “existing time”, “vertical and horizontal dimensions”, “speed”, etc. of the “change area”, and the validity / invalidity is examined, and “width”
Replace the work.

As described above, according to this embodiment,
Every time there is an instruction to perform the correction, the “feature parameter” such as “area”, “existing time”, “vertical and horizontal dimensions”, “speed”, etc. are excluded from the recognition processing and restored, and the updated result is the “feature parameter” at that time. This is in accordance with the validity / invalidity of the parameter "". As a result, the "feature parameter" is gradually narrowed down to only meaningful "items", and is modified so as to react only to the regular alert target.

For example, by excluding "area", an intruder whose remarkable change in "area" of the "change area" reciprocating from the vicinity of the camera to the far side is erroneously determined not to be an event target. Oversight is difficult.
Also, for example, by excluding “existing time”, the “existing time” of the “change area” which comes and goes irregularly within the field of view of the camera is changed.
It is difficult for an intruder whose remarkable change has occurred to be overlooked, which is erroneously determined not to be an event target.

Also, for example, by excluding “aspect ratio”, a plurality of persons come and go in a group and the arrangement changes every moment, so that an intruder who has a remarkable change in “aspect ratio” in the change area is erroneously detected. It is difficult to overlook when it is determined that the object is not the target of the event. Furthermore, for example, by excluding "speed", an intruder who comes and goes irregularly within the field of view of the camera and has a significant change in "speed" in the "change area" is erroneously determined not to be an event target. Is difficult.

As described above, there is provided an apparatus capable of reducing oversight and reliably issuing an alarm even for a "regular alarming target" which moves irregularly.

Embodiment 3 FIG. FIG. 5 is a diagram showing a configuration of a monitoring image processing apparatus according to still another embodiment of the present invention. In the figure, the same or corresponding parts as those in the conventional and the above-described embodiments are denoted by the same reference numerals, and description thereof will be omitted. 14b is the third that identifies the “most frequently reported object” from others
This is the most frequent notification target identification processing means. Reference numeral 15 denotes the most frequent notification target identification processing means 14b and the mask range calculation processing means 1
6 is a trigger input for instructing execution of correction.

Reference numeral 16 denotes a mask range calculation processing unit that receives an instruction from the most frequently reported target identification processing unit 14b and calculates a mask range. Further, the video input unit 1 receives the notification from the mask range calculation processing means 16 and updates the mask range of the image as appropriate. In general, the most frequently reported object identification processing means 14b and the mask range calculation processing means 16 are also processed by software.

FIG. 6 is a flowchart showing the operation of the mask range calculation processing means 16 in detail. Further, the operations of the recognition processing means 11 and the most frequent notification target identification processing means 14b are basically the same as those shown in the flowcharts of FIGS. 12 and 2, respectively.

Next, a typical example of the operation will be described.
Regarding the video input unit 1 to the trigger input 15 in FIG.
The description is omitted because it is basically the same as the above embodiment. However, the following points are different. Upon receiving the notification 12 indicating the occurrence of the "event", the most frequently-reported object identification processing means 14b observes the "coordinates" at that time in addition to the operation of the first embodiment, and the Is calculated, the "unrecognized unnecessary range" exists.

Hereinafter, an example of the behavior will be described in detail. However, what is described here is an example of calculation. The most frequently occurring object identification processing means 14b first sends the event discovery notification 12
Accordingly, the updating operation of the “feature parameter” as described in the first embodiment is performed. This operation is performed in the first embodiment.
It is omitted because it is the same as.

At the same time, the most frequently reported object identification processing means 1
In step 4b, each time an "event" occurs, a report 16a indicating "whether or not the current" event "is a" most frequently reported object "is output to the mask range calculation processing means 16. The mask range calculation processing means 16 receives the report 16a and performs the operation shown in FIG.

In FIG. 6, first, in step S1, the "coordinates" at the time of the alarm are referenced (stored). Next, in steps S2 to S4, a “recognition unnecessary range” is determined based on the “coordinates”.

An example of a specific method will be described below. First, a distribution map is created with “X coordinate” on the horizontal axis and “Y coordinate” on the vertical axis. However, based on the report 16a, two sheets are created, one for the "most frequently reported" and the other for the other.・ If the distribution of the “most frequently reported” and the distribution of other reported objects are concentrated in the non-overlapping area, and their occurrence varies with time, The distributed range is referred to as a “recognition unnecessary range”.

The "recognition unnecessary range" is characterized by "depending on the place" and "change in a relatively short period". For example, a “change region” of a branch or leaf when a tree is shaken by the wind corresponds to this. This occurs only at the location of the trees (depending on the location) and becomes noticeable during periods of wind (relatively short-term changes).

In step S3, it is checked whether correction has been performed. If there is no instruction, the operation is completed. Only the “recognition unnecessary range” is determined (this need not be performed, and the coordinates may be simply plotted on the distribution map).
If there is an instruction, the process branches at step S3 and branches to step S4.
To perform the operation of calculating the "" mask range "" optimal for the "" recognition unnecessary range "". The "mask range" is to mask an image to make it invisible, thereby preventing unnecessary image capture.

The calculation of "" mask range "" is specifically performed as follows.
As described below. The following is an example.
If the “recognition-unnecessary range” can be specified, the center thereof is obtained, and a rectangular area having the same midpoint that is large enough to hide about 60% of the “recognition-unrequired range” is considered.
This is the “mask range”. "" Mask range ""
Is used in the video input unit 1 to directly mask black solid on a video image. As a result, the video output from this is as if a black cloth was attached to the front of the camera. In step S5, this "mask range" is returned to the upper level.

This data is passed to the video input 1 as shown in FIG. The video input unit 1 performs an operation of masking an input image according to data. As a result, about 60% of the ““ recognition unnecessary range ”” is covered with the mask, and the changed area of the covered portion is not visible. As a result, the “change area” no longer occurs at the corresponding location.

As described above, in this embodiment, the "mask range" is updated every time there is an instruction to execute the correction.
In addition, the updated result covers the frequently occurring area of the “unnecessary detection target” at that time, and as a result, the video input unit 1 tracks and hides the “unnecessary detection target”.

As a result, the “mask range” of the video input unit 1 is optimized according to the “unwanted detection target”, for example, such as the “change area” of a branch or leaf when a tree is shaken by wind. "Unnecessary detection target", which is characterized by "depending on location" and "change in a relatively short period", is less likely to be erroneously determined to be an event. As described above, the present invention provides an apparatus capable of reducing "unnecessary detection" and accurately issuing "regular notification target".

Embodiment 4 FIG. 7 is a diagram showing a configuration of a monitoring image processing apparatus according to still another embodiment of the present invention. In the figure, the same or corresponding parts as those in the conventional and the above-described embodiments are denoted by the same reference numerals, and description thereof will be omitted. 14b is the third that identifies the “most frequently reported object” from others
This is the most frequent notification target identification processing means. Reference numeral 15 denotes the most frequent notification target identification processing means 14b and the recognition range calculation processing means 17
Is a trigger input for instructing the execution of correction.

Reference numeral 17 denotes a recognition range calculation processing unit that receives an instruction from the most frequently reported target identification processing unit 14b and calculates a recognition range. The recognition processing means 11 receives the notification from the recognition range calculation processing means 17 and updates the recognition range of the image as appropriate. In general, the most frequently reported object identification processing means 14b and the recognition range calculation processing means 17 are also processed by software.

FIG. 8 is a flowchart showing the operation of the recognition range calculation processing means 17 in detail. Further, the operations of the recognition processing means 11 and the most frequent notification target identification processing means 14b are basically the same as those shown in the flowcharts of FIGS. 12 and 2, respectively.

Next, a typical example of the operation will be described.
Regarding the video input unit 1 to the trigger input 15 in FIG.
The description is omitted because it is basically the same as the above embodiment. However, the following points are different. Upon receiving the notification 12 indicating the occurrence of the "event", the most frequently reported object identification processing means 14b observes the "coordinates" at that time in addition to the work of the first embodiment, and the "mostly reported object" exists unevenly. "Recognition range" is calculated. Hereinafter, the behavior example will be described in detail.

However, what is described here is an example of calculation. The most frequently reported target identification processing means 14b first receives the event discovery notification 12 and updates the "feature parameter" as described in the first embodiment. Since this operation is the same as that of the first embodiment, the description is omitted. At the same time, the most frequently reported object identification processing means 14b outputs a report 17a of "whether or not the current" event "is the most frequently reported object" to the recognition range calculation processing means 17 every time an "event" occurs. I do.

The recognition range calculation processing means 17
The operation shown in FIG. In FIG. 8, first, in step S1, the "coordinates" at the time of the alarm are referenced (stored). Next, in step S2, "" recognition range "" is determined based on "coordinates".

An example of the specific method will be described below. First, a distribution map is created with “X coordinate” on the horizontal axis and “Y coordinate” on the vertical axis. However, based on the report 17a, two sheets are created, one of which is the "most alarmed object" and the other of which is the "alarmed object". -If the distribution of the "most-reported object" and the distribution of the other-reported objects are concentrated in non-overlapping ranges, and their occurrence varies with time, the "most-reported object"
Is defined as ““ recognition range ””. The “recognition range” is a range in which the “most frequently reported object” frequently occurs, and specifically, a range in which a road or a vacant lot suitable for an intruder exists.

In the third embodiment, on the other hand, "unnecessary detection" is prevented by using a technique called "mask range" in other ranges. However, in a situation where the distribution of the “most frequently reported object” and the distributions of other reported objects are close to each other or overlap, by setting ““ mask range ””, the range where the “most frequently reported object” frequently occurs May have adverse effects.
This is because the distribution range of the “most frequent object” is also partially masked, and the alarm target is partially covered. In this case, oversight occurs.

The “recognition range” considered here is an area that is the distribution range of the “most frequently reported object” and does not belong to the distribution of the “unwanted detection object”. "" Recognition range ""
By considering only the change areas in
Even if the target partially overlaps the range set as "" mask range "", the alarm can be issued without overlooking. That is, it is possible to prevent a change area existing over both the “recognition range” and the non-recognition range from being ignored beforehand.

In step S3, the presence or absence of a trigger input 15, which is an instruction for performing correction, is checked. If there is no instruction, the operation is completed. Only judge "" recognition range ""
(You can just plot the coordinates on the distribution map without doing this.) If there is an instruction, step S3
And the process goes through steps S4 and S5, and returns the obtained "" recognition range "" to the upper level.

As shown in FIG. 7, this data is passed to the recognition processing means 11. The recognition range calculation processing means 17 updates the "" recognition range "" of the recognition processing means 11 and reflects it in the next recognition.

As described above, according to the present embodiment, the “recognition range” is updated each time there is an instruction to perform correction,
In addition, the update result is that of the “most frequently reported object” at that time, and as a result, the recognition processing unit 11 tracks “the most frequently reported object”.

Thus, the processing of the recognition processing means 11 is optimized in accordance with the “most frequently reported object”. For example, in a situation where a tree is shaking in the wind, the tree passes in front of the tree while partially overlapping the tree. Such an intruder can be found without being overlooked. In addition, it is possible to prevent an “unnecessary detection target” such as a “change area” of the tree branch from being erroneously determined as an event.

Even in the case where the areas are overlapped in this way, the “regular notification target” is kept while “unnecessary detection” is suppressed.
The present invention provides a device capable of accurately issuing an alarm.

Embodiment 5 FIG. FIG. 9 is a diagram showing a configuration of a monitoring image processing apparatus according to still another embodiment of the present invention. In the figure, the same or corresponding parts as those in the conventional and the above-described embodiments are denoted by the same reference numerals, and description thereof will be omitted. Reference numeral 14c denotes a fourth most frequently reported object identification processing means for distinguishing the most frequently reported object from others. Reference numeral 15 denotes a trigger input for instructing the most frequent notification target identification processing means 14c and the area integration calculation processing means 18 to perform correction.

Numeral 18 is an area integration operation processing means which receives an instruction from the most frequently reported object identification processing means 14c and judges the effectiveness of area integration. The recognition processing unit 11 receives the notification from the region integration calculation processing unit 18 and appropriately updates the necessity of the region integration of the image. Most frequently reported object identification processing means 14c
In general, the area integration operation processing means 18 is also processed by software.

FIG. 10 is a flowchart showing in detail the operation of the area integration operation processing means 18. Also, the operations of the recognition processing means 11 and the most frequent notification target identification processing means 14c are basically the same as those shown in the flowcharts of FIGS. 12 and 2, respectively.

Next, a typical example of the operation will be described.
Regarding the video input unit 1 to the trigger input 15 in FIG.
The description is omitted because it is basically the same as the above embodiment. However, the following points are different. Upon receiving the notification 12 notifying of the occurrence of the “event”, the most frequently reported target identification processing unit 14c observes (stores) the “area” at that time in addition to the work of the first embodiment, and determines “effectiveness of area integration”. calculate. Hereinafter, the behavior example will be described in detail. However, what is described here is an example of calculation.

First, the most frequently reported object identification processing means 14c receives the notification 12 of the "event" discovery and updates the "feature parameter" as described in the first embodiment. This operation is the same as in the first embodiment, and will not be described. At the same time, the most frequently reported object identification processing means 14c outputs a report 18a of "whether or not the current" event "is the most frequently reported object" to the area integration processing means 18 every time an "event" occurs. I do.

In the case of the “most frequently reported object”, the area integration operation processing means 18 receives the report 18a and
Do the work. The following description is all about "the most frequently reported object". In FIG. 10, first, in step S1, the "area" at the time of the alarm is referred to. In step S2, the “area ratio before and after“ region integration ”” is determined based on “area”.

An example of the specific method will be described below. A change region having the largest area among the “change regions” obtained by “region integration” is defined as a core change region.・ "Area of nuclear change" for "Area after" region integration ""
Is defined as “area ratio before and after“ region integration ””. Create a histogram with the horizontal axis representing the "area ratio before and after" region integration "" and the vertical axis representing the number of times. If the histogram indicates that the “area ratio before and after“ region integration ”” is biased to, for example, 70% or more, it is determined that region integration is unnecessary. If the histogram indicates that the “area ratio before and after“ region integration ”” is biased to, for example, 50% or less, it is determined that region integration is necessary.・ Other than that, it is judged as it is.

The “region integration” is a “change region” which is divided due to the deterioration of image quality or the like despite being the same object.
This is the task of integrating by analogy with relative distances and the like to bring it closer to the original shape. The area after “region integration” usually indicates the total area after integrating the divided “change regions”. A large ratio of the “nuclear“ change region ”” to the “area after“ region integration ”” means that the “nuclear“ change region ”” occupies the change region by itself.

In general, when the "change area" is not properly cut out due to deterioration of image quality or the like, the "change area" of the "report target" is cut into a plurality of pieces. The fact that the ratio of the “nuclear“ change region ”” is large indicates that this phenomenon has not occurred, and that the image quality can be determined to be good.

The “area ratio before and after“ region integration ”” is “70%
In the case where the deviation is biased toward the above, it can be generally determined that this is a "situation in which a clear" change region "with little division of the" change region "is selected". In such a case, the “area integration” processing is not required. Conversely, there is a problem in that an erroneous “area integration” is performed and a target object is misidentified.
Therefore, not performing “region integration” more than necessary can contribute to stable recognition processing.

In step S3, the presence or absence of a trigger input 15, which is an instruction for performing correction, is checked. If there is no instruction, the operation is completed. It is only determined "whether or not" region integration "processing can be performed" (there is no need to perform this, and only plotting on a histogram may be performed). If there is an instruction, the process branches at step S3, passes through steps S4 and S5, and returns the result of "necessity of" area integration "processing" that is optimal for the "most frequently reported object" to the upper level. As shown in Fig. 1-5-1,

This data is passed to the recognition processing means 11. In the recognition processing means 11, "necessity of" area integration "processing"
Will be updated and reflected in the next recognition. That is, when an instruction of “No” is given to the area integration processing, the processing of step S1 in FIG. 12 is not performed.

As described above, according to this embodiment,
Every time there is an instruction to perform correction, "necessity of" area integration "processing"
Is updated, and the update result is suitable for the “most frequently reported object” at that time. As a result, the ““ region integration ”” of the recognition processing unit 11 is changed to “the most frequently reported object”.
It is optimized according to the instability, such as the "variable area" of the branches and leaves, such as when the tree is shaken by the wind. It is difficult for the detection target "to be erroneously determined to be an event.

As described above, the present invention provides an apparatus capable of reducing "unnecessary detection" and accurately issuing "regular notification target".

[0093]

As described above, according to the present invention, a current image storage unit for temporarily recording data of a current image of an image of a monitoring target area, a background image storage unit for recording background image data of the image, A background difference extraction unit that compares the current image data of the current image storage unit with the background image data of the background image storage unit, and extracts a moving object in the image and a multivalued change region corresponding to a change object as a difference; A feature value calculation unit for calculating a predetermined feature value indicating the feature of the feature, a report target feature value storage unit for storing a feature parameter which is a preset value range of the feature value of the report target, Recognition processing means for recognizing that the feature quantity in the feature quantity calculation section corresponds to a regular alerting target based on the feature quantity parameter of the report target feature quantity storage section, and recognition of the recognition processing means. When subject to alarm An alarm processing means for generating an alarm; and identifying the most frequent alarm target having a high frequency of occurrence among the alarm targets from the occurrence distribution state, and identifying the feature parameter in the alarm target feature amount storage unit. Since the monitoring image processing apparatus is provided with a most-frequently-reported object identification processing means for correcting the center value to the central value of the feature value of the most-frequently-reported object, the feature in the report-targeted feature amount storage unit By updating the volume parameter according to the feature value of the most frequently reported object,
It is possible to provide a device that can reduce unnecessary detection and, at the same time, accurately issue a notification to a regular notification target.

Further, in the present invention, the most frequently reported object identification processing means further selects, from the state of occurrence distribution, those having a large variation among the feature amounts of the most frequently reported objects,
Since the correction is performed so that the feature parameter in the notification target feature storage unit is excluded from the setting, a feature parameter having a large variation among the feature parameter in the notification target feature storage is temporarily removed from the recognition process. , A device in which unnecessary detection is further reduced can be provided.

Further, according to the present invention, a video input unit for inputting a video signal serving as an image of the monitoring target area, and an area on the image where invalid change areas frequently occur based on feature amount data of the most frequently reported object. A mask range calculation processing means for causing the video input unit to mask the area of the image, so that an unnecessary recognition range is masked to further reduce unnecessary detection. Can be provided.

Further, the present invention further comprises a recognition range calculation processing means for estimating an area on a video which frequently occurs and which frequently occurs, and instructing the recognition processing means to set only the area as a monitoring range. Further, it is possible to provide a device that can reduce unnecessary detection and, at the same time, accurately issue a notification to a regular notification target.

Further, in the present invention, before the recognition processing means recognizes whether or not the information corresponds to the normal notification target, the change area is integrated, and the area integration with respect to the area after the area integration is performed. It further includes a region integration operation processing unit that infers the necessity of region integration from the ratio of the area of the maximum area change region of each of the change regions, and prohibits the integration of the change regions in the recognition processing unit. By performing only appropriate area integration, it is possible to provide a device that issues a more accurate alert to a regular alert target.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a monitoring image processing apparatus according to an embodiment of the present invention.

FIG. 2 is a flowchart showing in detail the operation of the most frequent notification target identification processing means of FIG. 1;

FIG. 3 is a diagram showing a configuration of a monitoring image processing apparatus according to another embodiment of the present invention.

FIG. 4 is a flowchart showing in detail the operation of the most frequent notification target identification processing means of FIG. 3;

FIG. 5 is a diagram showing a configuration of a monitoring image processing apparatus according to still another embodiment of the present invention.

FIG. 6 is a flowchart showing the operation of the mask range calculation processing means of FIG. 5 in detail;

FIG. 7 is a diagram showing a configuration of a monitoring image processing apparatus according to still another embodiment of the present invention.

FIG. 8 is a flowchart showing in detail the operation of the recognition range calculation processing means of FIG. 7;

FIG. 9 is a diagram showing a configuration of a monitoring image processing apparatus according to still another embodiment of the present invention.

FIG. 10 is a flowchart showing in detail the operation of the area integration operation processing means of FIG. 9;

FIG. 11 is a diagram showing a configuration of a conventional monitoring image processing apparatus of this type.

FIG. 12 is a flowchart showing the operation of the recognition processing means of FIG. 11 in further detail;

FIG. 13 is a diagram illustrating a change area in the operation of the monitoring image processing apparatus.

FIG. 14 is a diagram illustrating a change area in the operation of the monitoring image processing apparatus.

[Explanation of symbols]

Reference Signs List 1 video input section, 2 A / D conversion section, 3 current image storage section, 4 background image storage section, 5 background update section, 6 background difference extraction section, 7 threshold value calculation section, 8 binarization section, 9 Amount calculation unit, 10 notification target feature amount storage unit, 11 recognition processing unit, 13 notification processing unit, 14 first maximum notification target identification processing unit, 14a second maximum notification target identification processing unit, 14b 3, the most frequent notification target identification processing means, 14
c fourth most frequently reported object identification processing means, 16 mask range calculation processing means, 17 recognition range calculation processing means, 18
This is an area integration calculation processing means.

Claims (5)

[Claims] 1. A current image storage unit for temporarily recording data of a current image of an image of a monitoring target area, a background image storage unit for recording background image data of the image, and a current image of the current image storage unit A background difference extraction unit that compares data with the background image data of the background image storage unit and extracts a multivalued change region corresponding to a moving object or a change object in the image as a difference; and a predetermined feature amount indicating a feature of the change region. , A report target feature storage unit that stores a feature parameter that is a preset value range of the feature of the report target, and a report target feature storage. Recognition processing means for recognizing that the feature quantity in the feature quantity calculation section corresponds to a regular one to be issued based on the feature quantity parameter; Alarm processing means for reporting And, the most frequently reported object among the reported objects is identified as the other based on the appearance distribution, and the central value of the feature parameter in the reported target feature amount storage unit is determined as the most frequently reported object. A surveillance image processing apparatus, comprising: a most-notification-target identification processing means for performing correction to align with the center value of the characteristic amount of 2. The most frequently reported object identification processing means further selects, from the state of occurrence distribution, the feature amount of the most frequently reported object from the state of occurrence distribution. The monitoring image processing apparatus according to claim 1, wherein the correction is performed such that the feature amount parameter is excluded from the setting. 3. A video input unit for inputting a video signal serving as an image of the monitoring target area, and estimating an area on an image in which invalid change areas frequently occur from data of a feature amount of the most frequently reported object. The monitoring image processing apparatus according to claim 1, further comprising: mask range calculation processing means for causing the input unit to mask the area of the image. 4. A recognition range calculation processing means for estimating an area on a video which frequently occurs and which frequently occurs, and instructing the recognition processing means to set only the area as a monitoring range. The monitoring image processing device according to claim 1. 5. The method according to claim 1, wherein the change processing unit integrates the change regions before the recognition processing unit recognizes whether or not the change corresponds to the legitimate notification target. It further comprises a region integration operation processing means for inferring the necessity of region integration from the ratio of the area of the maximum area change area of the area and prohibiting the integration of the change area by the recognition processing means. The monitoring image processing device according to claim 1.