JP2012208878A - Image monitoring device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image monitoring device for improving convenience of a regular user by permitting installation of a predetermined article while maintaining performance of intrusion detection.SOLUTION: A change area extraction part 231 compares a monitoring image obtained by successively imaging a monitoring space by an imaging part 21 with a background image 221 stored in a storage part 22 to extract a changed area, and an object area detection part 232 detects an object area by making the outer shape of the changed area approximate to a predetermined graphic. An object area tracing part 233 associates object areas whose characteristics are similar with each other among object areas detected on sequential time, and calculates a moving distance from the detected position for each of the associated object areas. A shielding abnormality determination part 235 compares the moving distance of the object area with a pre-set static determination value, calculates the rate of the changed area occupying the object area to compare the rate with a pre-set shielding allowable value, and when the moving distance is less than the static determination value, and the rate is equal to or more than the shielding allowable value, determines that there exists a shielding object which hinders the detection of an intruder.

Description

  The present invention relates to an image monitoring apparatus that detects an intruder abnormally using an image obtained by imaging a monitoring space and detects the presence of an obstruction that prevents the intruder from being detected abnormally.

  Conventionally, in an image sensor that detects an intruder by processing an image captured by an imaging unit, the presence of a shielding object that blocks the field of view of the imaging unit is detected abnormally in order to continuously detect intrusion.

  In the image sensor described in Patent Literature 1, the degree of similarity between the current image and the reference image is determined for each determination block into which the image is divided, and the visual field is blocked if the number of determination blocks with a low similarity is equal to or greater than a predetermined number It is determined.

  Further, in this image sensor, a counter is provided for each determination block, the counter of the determination block having a low similarity is increased, and the counter of the determination block having a high similarity is decreased, so that the counter value exceeds the threshold value. If a certain number of determination blocks is greater than or equal to a predetermined number, it is determined as abnormal. As a result, it was determined that an obstacle that obstructed the field of view was placed without erroneous determination due to a temporary environmental change.

JP 2000-194866 A

  However, according to the prior art, a resident who has a certain size in the premises, but has a gap and does not interfere with monitoring, it is judged as an abnormality that obstructs the field of view even if an article such as a bicycle or a stepladder is placed. , Could reduce the resident's convenience.

  In other words, when these items are placed, the degree of similarity is reduced in a predetermined number or more of judgment blocks because they are large enough to block intruders, and the counter increases because they are stationary. Subsequently, the existence is determined to be abnormal.

  The present invention has been made in view of the above problems, and realizes an image monitoring apparatus capable of improving the convenience of a regular user by allowing installation of a predetermined article while maintaining intrusion detection performance. With the goal.

  In order to solve such a problem, the present invention provides an image monitoring apparatus that detects an intruder using an image obtained by imaging a monitoring space, an imaging unit that sequentially images the monitoring space, and a background image captured at least at a predetermined time. A change area extraction section that extracts a change area by comparing a monitoring image captured after a predetermined time with a background image, and detects an object area by approximating the outline of the change area with a predetermined figure An object region detecting unit that associates object regions with similar characteristics among object regions detected at preceding and following times, and a moving distance calculating unit that calculates a moving distance from a detection position for each associated object region, Compare the moving distance of the area with a preset stillness judgment value, calculate the ratio of the changed area in the object area, and compare it with the preset shielding tolerance value. And ratio to provide an image monitoring apparatus characterized by comprising a, and determining shielding abnormality determining unit that there is an interrupting object interfering with the detection of the intruder when it is shielded acceptable value or more.

  In such an image monitoring apparatus, when there are a plurality of change areas detected in proximity to each other within a preset allowable range, the object area detection unit detects the object area with respect to the group of the plurality of change areas. It is preferable.

  In addition, the shielding abnormality determination unit determines that the object region whose moving distance is less than the stationary determination value and the ratio is less than the shielding allowance value is a low shielding object region that does not hinder the detection of the intruder, and the change region extraction unit is a low shielding unit. It is preferable to replace the background image in the object area with the monitoring image.

  In addition, the shielding abnormality determination unit stores the change region extracted in the low shielding object region in the storage unit, and uses the combined change region obtained by combining the change region extracted after the storage and the stored change region. Is preferably calculated.

  Further, the shielding abnormality determination unit stores the low shielding object region in the storage unit, calculates the ratio of the combined change region in the overlapping region of the object region detected after the storage and the stored low shielding object region, and shields it. It is preferable to determine the object.

  Since the installation of bicycles or stepladders with many gaps can be allowed while maintaining the performance of intrusion detection, an image monitoring apparatus with improved convenience for authorized users can be realized.

1 is an overall view of an outer periphery monitoring system. It is the block diagram which showed the structure of the image sensor. It is a figure explaining a mode that an object area | region is detected from the difference image obtained by carrying out background difference processing of the monitoring image with which the intruder who is passing behind the stepladder was imaged. It is an example in the case where the division of the change area does not occur, and is a diagram illustrating the change area and the object area with respect to an intruder image captured without being shielded. It is an example in the case where division of the change area does not occur, and is a diagram showing a change area and an object area for a screen image. FIG. 6 is a diagram illustrating a change area and an object area with respect to a stepladder image in an example in which a change area is not split. It is a figure explaining a mode in the case of an article with many gaps, when a plurality of the articles overlap on an image, it fills up gaps and prevents detection of an intruder. It is a flowchart explaining an image monitoring process. It is a flowchart explaining an abnormality determination process.

  Hereinafter, as an example of a preferred embodiment of the image monitoring apparatus of the present invention, an outer periphery monitoring system 1 that notifies the security center when an intruder and a shield are detected by monitoring the outer periphery of a building will be described.

<Configuration of outer periphery monitoring system 1>
FIG. 1 is an overall view of the outer periphery monitoring system 1. The outer periphery monitoring system 1 includes an image sensor 2, an authentication device 3, a mobile terminal 4, a controller 5, and a center device 6.

  The image sensor 2 and the authentication device 3 are connected to a controller 5 via a communication line, and the controller 5 is connected to a center device 6 installed at a remote place such as a security center via a wide-area communication network such as a telephone line or an internet line. Is done.

  The image sensor 2 is installed outdoors and monitors the outer periphery of the building which is a monitoring space. The image sensor 2 outputs an intrusion abnormality signal when it detects an intruder into the monitoring space, and outputs a shielding abnormality signal when it detects a shield that obstructs the detection of the intruder. These abnormal signals output from the image sensor 2 are transmitted to the center device 6 via the controller 5.

  The authentication device 3 authenticates regular users such as residents and employees in the monitoring space by communicating with the mobile terminal 4 in the monitoring space. The mobile terminal 4 possessed by the authorized user is registered in the authentication device 3 in advance, and when the authentication device 3 detects the registered mobile terminal 4, it outputs an authentication signal to the image sensor 2 via the controller 5. When the image sensor 2 detects a person, the image sensor 2 determines whether the user is an authorized user or an intruder by referring to the presence or absence of an authentication signal.

  FIG. 2 is a block diagram showing the configuration of the image sensor 2. The image sensor 2 includes an imaging unit 21, a storage unit 22, and a communication unit 24 connected to a control unit 23.

  The imaging unit 21 is a so-called surveillance camera. The imaging unit 21 sequentially captures the monitoring space at a predetermined time interval to generate a monitoring image, and sequentially outputs the captured monitoring image to the control unit 23. The imaging interval is 1/5 second, for example. Hereinafter, the unit of time recorded at this imaging interval is referred to as time.

  The storage unit 22 is a storage device such as a ROM (Read Only Memory) or a RAM (Random Access Memory). The storage unit 22 stores various programs and various data, and inputs / outputs such information to / from the control unit 23. The various data includes a background image 221, object information 222, acceptable shielding object information 223, and camera parameters 224.

The control unit 23 includes a CPU (Central Processing Unit) and a DSP (Digital Signal).
Processor, a control unit such as MCU (Micro Control Unit), and the like, by reading a program from the storage unit 22 and executing it, a change region extraction unit 231, an object region detection unit 232, an object region tracking unit 233, an intrusion It functions as an abnormality determination unit 234 and a shielding abnormality determination unit 235.

  The communication unit 24 includes a communication circuit between the controller 5 and driver software. The control unit 23 is connected to one of the communication units 24 and the controller 5 is connected to the other. An abnormal signal or the like from the control unit 23 is transmitted to the controller 5 via the communication unit 24, and an authentication signal or the like from the controller 5 is input to the control unit 23 via the communication unit 24.

  Hereinafter, the background image 221, the object information 222, the acceptable shielding object information 223, and the camera parameter 224 stored in the storage unit 22 will be described.

  The background image 221 is an image in which only the background of the monitoring space is captured. Prior to the abnormality detection process, the background image 221 is generated from an unattended monitoring image, stored in the storage unit 22, and thereafter appropriately updated. The background image 221 does not include an intruder, a regular user, or a newly brought-in shield, and by comparing the background image 221 with the monitoring image, a portion where the monitoring image has changed due to the appearance of these objects is displayed. It can be detected as a change area.

  The object information 222 is data in which the detection position and the feature amount of the object area (object area) detected from the monitoring image at each time are associated with each object. The object area is an area that approximates the outer shape of the change area. The feature amount of the object area includes the size of the object area, the shape of the object area, the luminance histogram of the monitoring image in the object area, and the like. In addition, object regions generated by the same object among the object regions detected at each time are associated by tracking processing, and each element of the object information 222 is managed in association with an identification code (object ID) for each object. The detection position is used for calculating and tracking the movement distance of the object region, and the feature amount is used for tracking.

  The acceptable shielding object information 223 is information on a shielding object (low shielding object) that is accepted without being abnormal because the degree of shielding is low. The acceptable shielding object information 223 includes an object area (low shielding object area) of the low shielding object and a change area extracted in the low shielding object area.

  The camera parameters 224 include external parameters such as the installation position and imaging direction of the imaging unit 21, internal parameters such as the focal length, field angle, lens distortion, and other lens characteristics of the imaging unit 21, and the number of pixels of the imaging device. These parameters obtained by actual measurement are preset and stored in the storage unit 22. By applying the camera parameter 224 to a camera model such as a known pinhole camera model, the pixel position in the monitoring image is represented by coordinates on the imaging surface of the imaging unit 21 (imaging surface coordinates) and coordinates in the real space (real coordinates). It is possible to convert coordinates between the two. Hereinafter, conversion from imaging surface coordinates to real coordinates is referred to as reverse perspective transformation, and conversion from actual coordinates to imaging surface coordinates is referred to as perspective transformation.

  Hereinafter, the change region extraction unit 231, the object region detection unit 232, the object region tracking unit 233, the intrusion abnormality determination unit 234, and the shielding abnormality determination unit 235 that the control unit 23 has as functions thereof will be described.

  The change area extraction unit 231 detects the change area in the monitoring image by comparing the monitoring image input from the imaging unit 21 with the background image 221, and detects the change area in the object area detection unit 232 and the shielding abnormality determination unit 235. Output to.

  Specifically, the change area extraction unit 231 extracts a change area by background difference processing. That is, the change area extraction unit 231 calculates the absolute value of the difference between the pixel values between the corresponding pixels of the monitoring image and the background image 221, and selects adjacent pixels among pixels whose absolute value of the difference is equal to or greater than the difference threshold. Change areas are detected together. The difference threshold is set in advance based on a preliminary experiment.

  In another embodiment, the change area extraction unit 231 extracts a change area by background correlation processing. That is, the change area extraction unit 231 calculates a correlation value of pixel values instead of the difference, and detects a pixel whose correlation value is equal to or less than a preset correlation threshold as a change area.

  In addition, the change area extraction unit 231 adapts the background image 221 to environmental changes in the monitoring space by appropriately updating the background image 221 using the monitoring image in order to continue to detect the changing area accurately.

  Specifically, the change region extraction unit 231 replaces the background image 221 in the low shielding object region accepted by the shielding abnormality determination unit 235 described later with a monitoring image. By this update, the low shielding object is taken into the background image 221 and the changed area due to the low shielding object is not extracted. Therefore, when a new object appears in the accepted low shielding object area, it is possible to prevent the object area of the object from being unduly enlarged including the changed area due to the accepted low shielding object. Can be performed. Incorporating the accepted low shielding object into the background image 221 also means the end of tracking of the low shielding object.

  Further, the change area extraction unit 231 may update (replace, average, etc.) the monitor image excluding the object area being tracked with the background image 221 in order to cope with the brightness fluctuation of the monitor space.

  The object area detection unit 232 detects the object area by allowing the integration of the change areas included in the object frame set to the size and shape of the intruder and approximating the outline of the change area with a predetermined figure. The detected object area is output to the object area tracking unit 233, the intrusion abnormality determining unit 234, and the shielding abnormality determining unit 235. The object frame can be a rectangle imitating the maximum size of an intruder (for example, 2 m long × 1 m wide), and the object region can be a circumscribed rectangle of the change region. In another embodiment, an ellipse may be used instead of a rectangle.

  When the intruder passes behind the shield with a gap, a plurality of change regions that are close to each other are extracted at positions corresponding to the intruder image. FIG. 3 shows a state in which an object region is detected from a difference image 33 obtained by performing background difference processing on a monitoring image 32 in which an intruder 31 passing behind a stepladder 30 is captured. At the time of processing, the image of the stepladder 30 is already captured in the background image 221, and a plurality of change areas 34 (hatched portions) are extracted from the portion corresponding to the image of the intruder 31.

  The object area detection unit 232 calculates the area of the change area including the object frame 35 (one-dot chain line) at each position of the difference image 33 and calculates the area of the object frame where the area shows a peak. The same label is given to the included change area. The change areas to which the same label is assigned are recognized as a group of change areas generated by the same object. In the example illustrated in FIG. 3, the same label is assigned to each of the plurality of change regions 34.

  Here, since the apparent size of the intruder on the image varies depending on the distance from the imaging unit 21, the object region detection unit 232 sets the object frame with an apparent size corresponding to each position on the image. . That is, the object region detection unit 232 performs perspective conversion of an object frame preset in real space size to an apparent size in the monitoring image, and sets the object frame at each position. It is assumed that the lowest end of each change area is the intruder's foot position (ground plane), and the object area detection unit 232 sets the object frame only at the position where the lowest end of each change area is the lowest end of the object frame. Good.

  In addition, the object area detection unit 232 assigns individual labels to each change area that is not the target of the integration process.

  In this way, the object area detection unit 232 that assigns a label to each change area obtains, as the object area, the smallest rectangle that includes the change area for each label. For the plurality of change regions 34 illustrated in FIG. 3, a rectangular object region 36 circumscribing these groups is detected.

  On the other hand, FIGS. 4 to 6 are examples in the case where the change region is not split. FIG. 4 shows an image of the intruder 41 captured without being blocked, a change area 44 extracted corresponding to the image of the intruder 41, and an object area 46 of the intruder 41 detected based on the change area 44. ing. FIG. 5 shows an image of the partition 50, a change area 54 extracted corresponding to the image of the partition 50, and an object area 56 of the partition 50 detected based on the change area 54. FIG. 6 shows an image of the stepladder 60, a change area 64 extracted corresponding to the image of the stepladder 60, and an object area 66 of the stepladder 60 detected based on the change area 64.

  The object area tracking unit 233 is a movement distance calculation unit that calculates the movement distance of the object area by tracking processing. The object region tracking unit 233 associates object regions having similar characteristics among the object regions detected from the monitoring images at the preceding and following times, and calculates the movement distance of the object region from the detected position of the associated object region. Then, the calculated movement distance is output to the intrusion abnormality determination unit 234 and the shielding abnormality determination unit 235. The object area tracking unit 233 stores the object ID, the history of the detected position of the object area, and the feature amount of the object area in the object information 222 for the tracking process. When a plurality of object areas exist simultaneously, the above process is performed for each object area.

  The intrusion abnormality determining unit 234 outputs an intrusion abnormality signal based on an object region whose moving distance is equal to or greater than a preset stationary determination value. That is, the intrusion abnormality determination unit 234 compares the moving distance of each object region at each time with the stationary determination value, confirms the presence or absence of an authentication signal at the time, and has an object region whose moving distance is equal to or larger than the stationary determination value, and When the authentication signal is not confirmed, an intrusion abnormality signal is output as an intruder exists in the monitoring space. On the other hand, when there is no object region whose moving distance exceeds the stationary determination value and when the authentication signal is confirmed, the intrusion abnormality determining unit 234 does not output the intrusion abnormality signal.

  The intrusion abnormality determination unit 234 further calculates the disturbance attribute value and the human attribute value of the object region using the size, movement, and texture similarity of the monitoring image and the background image 221 in the object region. It can be added to the output condition of the intrusion abnormality signal that the human attribute value is higher than the attribute value.

  The shielding abnormality determination unit 235 compares the moving distance with a preset stillness determination value, calculates the ratio of the change area in the object region (shielding degree), and compares the shielding degree with a preset shielding tolerance value. Then, when there is an object region whose moving distance is less than the stationary determination value and whose shielding degree is greater than or equal to the shielding tolerance value, a shielding abnormality signal is output. On the other hand, when there is no object region whose moving distance is less than the stationary determination value and when there is no object region whose shielding degree is equal to or greater than the shielding acceptance value, the shielding abnormality determination unit 235 does not output a shielding abnormality signal.

  Specifically, the shielding abnormality determination unit 235 calculates the shielding degree by dividing the number of change regions in the object region by the total number of pixels in the object region. The shielding tolerance value can be set to the number of pixels equivalent to 30%, for example.

  In this way, by outputting a shielding abnormality signal based on the moving distance and the shielding degree, when there is a shielding object that actually prevents intruder detection, there is a shielding object that can output a shielding abnormality signal and does not interfere with intruder detection. Control that does not output a shielding abnormality signal even if it exists is possible. Therefore, it is possible to improve the convenience of authorized users without impairing intruder detection performance.

  For example, in the example of the partition 50 shown in FIG. 5, the moving distance of the object region 56 is approximately 0, and the change region 54 occupies most of the object region 56, so that the degree of shielding higher than the shielding allowable value is calculated. . For this reason, the screen 50 outputs a shielding abnormality signal on the assumption that the partition 50 is a shielding object that prevents intruder detection. In fact, if the partition 50 is left unattended, the tracking of the intruder passing behind it is interrupted and the travel distance cannot be calculated correctly, or the object area of the intruder is lost and the feature amount of the object area is reduced. Intruder detection performance deteriorates due to failure to extract and judge correctly. It is possible to take measures such as eliminating the partition 50 by outputting the shielding abnormality signal and notifying the existence of the partition 50.

  In the example of the stepladder 60 shown in FIG. 6, the moving distance of the object region 66 is substantially zero, but the degree of shielding lower than the shielding tolerance is calculated because the ratio of the change region 64 to the object region 66 is small. For this reason, the stepladder 60 is a low shielding object that does not hinder the detection of an intruder, and no shielding abnormality signal is output. Therefore, it is possible to perform intrusion detection by allowing an authorized user to place an article with a large gap such as a stepladder or a bicycle in the monitoring space.

  On the other hand, in the example of the intruder 31 and the intruder 41 shown in FIGS. 3 and 4, the moving distance of the object area 36 and the object area 46 is equal to or greater than the stationary determination value, and no shielding abnormality signal is output. An intrusion abnormality determination unit 234 outputs an intrusion abnormality signal to the intruder 31 and the intruder 41.

  Here, even if an article has many gaps, if a plurality of the articles overlap on the image, the gaps may be filled to prevent detection of an intruder. FIG. 7 shows the situation at this time. Therefore, the shielding abnormality determination unit 235 uses the change area 71 included in the object area (low shielding object area) 72 whose moving distance is less than the stationary determination value but the shielding degree is less than the shielding tolerance value as the acceptable shielding object information 223. The degree of occlusion is calculated using a composite change area 75 that is stored and superposed with the newly extracted change area 73 and the stored change area 71. By doing so, even if the shielding object does not interfere with intrusion detection by itself, a shielding abnormality signal can be output without overlooking this when a plurality overlaps and obstructs intrusion detection.

  At this time, the shielding abnormality determination unit 235 stores the low shielding object region 72 in the acceptable shielding object information 223, and stores the newly detected object region 74 in the overlapping region 76 of the stored low shielding object region 72. The ratio of the combined change area 75 is calculated as the shielding degree. As a result, since the degree of shielding can be calculated only in the region where the gap is actually filled, the oversight of the shielding abnormality can be further reduced.

<Operation of the outer periphery monitoring system 1>
The operation of the outer periphery monitoring system 1 will be described focusing on the operation of the image sensor 2.

  When an administrator who has confirmed that the monitoring space is unmanned activates the image sensor 2, the imaging unit 21 starts imaging, and the change area extraction unit 231 of the control unit 23 performs only a preset initialization time. By averaging the monitoring images from the imaging unit 21, an initial background image 221 is created and stored in the storage unit 22.

  When the initialization time elapses, the image monitoring process shown in the flowchart of FIG. 8 starts. In the image monitoring process, each time a monitoring image is input from the imaging unit 21 to the control unit 23, the processes in steps S10 to S18 are repeated. Hereinafter, the time when a new monitoring image is input is referred to as the current time.

  First, the control unit 23 acquires a monitoring image at the current time (S10), and the change area extraction unit 231 of the control unit 23 compares the monitoring image at the current time with the background image 221 to extract a change area (S11). That is, the change area extraction unit 231 performs difference processing between the monitoring image and the background image 221 to generate a difference image in which the pixel value in the change area is set to 1 and the pixel value outside the change area is set to 0.

  Next, the object region detection unit 232 of the control unit 23 detects the object region based on the change region extracted at the current time (S12).

  First, the object area detection unit 232 labels the change area based on the object frame. That is, the object region detection unit 232 obtains the lowest coordinate (maximum Y coordinate value) of each change region from the difference image, performs inverse perspective transformation to the real space coordinate system, and converts each lowest coordinate obtained by inverse perspective transformation to each part of the base. Is set perpendicular to the line of sight from the imaging unit 21, and the set object frame is perspective-transformed and sequentially set in the coordinate system of the monitoring image to derive an apparent object frame. The number of pixels in the change area that falls within each object frame is calculated to determine the setting of the object frame that has the peak number of pixels. Then, the object area detection unit 232 assigns the same label to the change area that fits in the object frame for each object frame that forms a peak.

  Next, the object area detection unit 232 labels the remaining change area. In other words, the object area detection unit 232 gives different labels to each change area that has not been given a label.

  Subsequently, the object region detection unit 232 obtains the maximum value and the minimum value of the X coordinate and the Y coordinate of the pixel constituting the change region to which the label is assigned for each label, and sets the X coordinate minimum value and the Y coordinate minimum value. And a rectangle having a pair of X coordinate maximum value and Y coordinate maximum value as two diagonal coordinates is detected as an object region. The object area detection unit 232 calculates the midpoint coordinates (corresponding to the foot position) of the bottom of the object area as the detection position of the object area.

  After the detection processing, the object region detection unit 232 outputs the detection result of the object region to the object region tracking unit 233 of the control unit 23. When there is no change area, the object area detection unit 232 outputs to the object area tracking unit 233 that there is no object area.

  The object region tracking unit 233 associates similar object regions between the object region detected at the current time and the object region detected one hour ago (S13). For this purpose, the object region tracking unit 233 stores the object region detection position and the feature amount of each object detected up to one hour ago in the storage unit 22 as object information 222. The object area tracking unit 233 obtains the size (number of pixels) of each object area detected at the current time, calculates the difference from the size of each object area stored, and calculates each object detected at the current time. Find the shape of the area (aspect ratio) and calculate the difference from the shape of each object area stored. Find the brightness histogram of each object area detected at the current time and store the brightness of each object area. Calculate the difference from the histogram, extrapolate the detection position of the current time to the history of detection positions of each object area stored, and the detection position of each object area detected at the current time For each combination of object regions, the similarity is calculated by applying each difference to a preset evaluation function, and combinations exceeding a preset similarity threshold are associated with each other.

  After the association, the object region tracking unit 233 updates the object information 222 with the association result (S14). In other words, the object region tracking unit 233 adds the same object ID as the object information 222 obtained by associating with the detected position and feature amount of the object region at the current time and appends the object information to the object information 222. Further, the object area tracking unit 233 assumes that the object area at the current time for which the association has not been obtained is a newly appearing object, and assigns a new object ID to the detected position and feature amount of the object area to the object information 222. Add. Further, the object region tracking unit 233 determines that the object information 222 for which the association has not been obtained is a lost object, and deletes the object information 222 from the storage unit 22.

  Subsequently, the object region tracking unit 233 refers to the object information 222 and calculates the movement distance from the detection position of each object (S15). In other words, the object region tracking unit 233 calculates the movement distance by accumulating the distance between the actual positions at the time before and after converting the detection position of each object into the actual position by inverse perspective transformation.

  When the tracking process and the movement distance calculation process progress in this way, an abnormality determination is performed by the intrusion abnormality determination unit 234 and the shielding abnormality determination unit 235 of the control unit 23 (S16). The abnormality determination process will be described with reference to FIG.

  In the abnormality determination, the control unit 23 sequentially sets each object stored in the object information 222 as a target object, and executes the loop process of S20 to S34 for the object being tracked.

  In the object loop processing, first, the intrusion abnormality determination unit 234 compares the moving distance of the object of interest with the stationary determination threshold (S21), and skips the following steps S22 to S25 if the moving distance is less than the stationary determination threshold (S21). NO → S26).

  On the other hand, if the movement distance is equal to or greater than the stillness determination threshold (YES in S21), the intrusion abnormality determination unit 234 proceeds to step S22, assuming that the object of interest may be an intruder. The intrusion abnormality determining unit 234 calculates the human attribute value and the disturbance attribute value of the target object and compares the values (S22, S23). If the human attribute value is equal to or greater than the disturbance attribute value, the target object is a person. Is confirmed whether or not an authentication signal is received (YES in S23 → S24). If the authentication signal has not been received, the intrusion abnormality determining unit 234 generates intrusion abnormality information on the assumption that the object of interest is an intruder, and temporarily stores it in the storage unit 22 (NO in S24 → S25).

  On the other hand, if the human attribute value is less than the disturbance attribute value, the intrusion abnormality determining unit 234 skips steps S24 and S25 (NO in step S23 → S26) because the object of interest is a disturbance of a small animal or the like. If the authentication signal is received, the intrusion abnormality determination unit 234 skips step S25 (YES in S24 → S26) because the object of interest is an authorized user. Disturbances and legitimate user object areas will be tracked until they disappear.

  Subsequently, the object of interest is determined by the shielding abnormality determination unit 235.

  First, in steps S26 and S27, the shielding abnormality determination unit 235 confirms that the object of interest is a stationary object. That is, the shielding abnormality determination unit 235 compares the moving distance of the object of interest with the stationary determination threshold (S26), and skips the subsequent steps S27 to S33 if the moving distance is equal to or greater than the stationary determination threshold (NO in S26 → S34). ). If the moving distance is less than the stationary determination threshold value (YES in S26), shielding abnormality determination unit 235 further compares the tracking time with the stationary determination time (S27). The reason for confirming the stillness determination time is to prevent erroneous recognition of a moving object that is paused. The stillness determination time is set in advance to a time of about 10 seconds, for example. The number of detected positions of the target object stored in the object information 222 can be used as the tracking time. If tracking time is less than stationary determination time, shielding abnormality determination unit 235 skips subsequent steps S28 to S33 (NO in S27 to S34).

  When it is confirmed that the target object is a stationary object, the shielding abnormality determination unit 235 calculates the degree of shielding of the target object.

  That is, the shielding abnormality determination unit 235 checks whether or not a low shielding object region overlapping with the object region of the object of interest is stored in the acceptable shielding object information 223 of the storage unit 22 (S28). A logical sum image of the difference image generated in S11 and the change area of the acceptable occlusion object information 223 is generated as a combined change area (YES in S28 → S29), and the low occlusion stored as the object area of the object of interest. In the overlapping area with the object area, the number of pixels having a pixel value of 1 in the logical sum image is counted, and the count result is divided by the number of pixels in the overlapping area to calculate the degree of occlusion of the object of interest (S30). On the other hand, if the overlapping low occluded object region is not stored (NO in S28), the occlusion abnormality determination unit 235 uses a pixel value of 1 in the difference image within the object region of the object of interest without using the composite change region. The number of pixels is counted, and the count result is divided by the number of pixels in the object region of the target object to calculate the degree of occlusion of the target object (S30). If the overlapping area is small enough not to prevent intrusion detection in step S30, the degree of shielding can be calculated without using the overlapping area. In this case, the shielding abnormality determination unit 235 determines that the pixel value of the logical sum image in the object region of the object of interest is 1 if the number of pixels in the overlapping region is smaller than a preset overlapping lower limit value (set to the size of a flowerpot). Count the number of pixels and divide the counting result by the number of pixels in the object area of the object of interest to calculate the degree of occlusion of the object of interest. If the number of pixels in the overlap area is equal to or greater than the overlap lower limit, And the degree of occlusion is calculated from the composite change area.

  Subsequently, the shielding abnormality determination unit 235 compares the calculated shielding degree with the shielding tolerance value (S31), and if the shielding degree is equal to or greater than the shielding tolerance value, the shielding abnormality information is determined that the target object is a shielding object that prevents intrusion detection. Generated and temporarily stored in the storage unit 22 (NO in S31 → S32).

  On the other hand, if the degree of occlusion is less than the occlusion tolerance value, it is assumed that the object of interest is a low occlusion object that does not prevent intrusion detection, and the occlusion abnormality determination unit 235 skips step S32 and the object area of the object of interest (low occlusion object area). And the information of a change area is newly memorize | stored in the acceptable shielding object information 223 (it is YES-> S33 in S31).

  When the shielding abnormality determination unit 235 finishes the process of step S32 or S33, the control unit 23 confirms whether the determination of all objects has been completed (S34). If determination of all objects is not completed (NO in S34), control unit 23 returns to step S20 to start determination of the next object, and determination of all objects is completed (YES in S34). ), The controller 23 advances the process to step S17 in FIG.

  Returning to FIG. 8, the intrusion abnormality determining unit 234 and the shielding abnormality determining unit 235 confirm the presence / absence of temporarily stored abnormality information (S17). If there is intrusion abnormality information, the intrusion abnormality determining unit 234 sends the information to the communication unit 24. If there is shielding abnormality information, shielding abnormality determination unit 235 outputs the information to communication unit 24 (YES in S17 → S18). If there is no abnormality information (NO in S17), step S18 is skipped.

  Subsequently, the change area extraction unit 231 updates the background image 221 using the monitoring image (S19). That is, the change area extraction unit 231 weights and adds the image obtained by cutting out the area in which the object area was not detected at the current time from the monitoring image and the background image 221 at a ratio of α: (1-α). The image 221 is updated, and if there is an area determined to be a low shielding object area at the current time, the area is cut out from the monitoring image, and the corresponding portion in the background image 221 is replaced with the cut out image to replace the background image 221. Update. However, α is a preset value in the range of 0 <α <1.

When the above process is completed, the process returns to step S10, and the process for the monitoring image at the next time is performed.

DESCRIPTION OF SYMBOLS 1 ... Perimeter monitoring system 2 ... Image sensor 3 ... Authentication apparatus 4 ... Portable terminal 5 ... Controller 6 ... Center apparatus 21 ... Imaging part 22 ... Memory | storage part 221. ..Background image 222... Object information 223... Acceptable shielding object information 224... Camera parameter 23... Control section 231... Change area extraction section 232. Object region tracking unit 234 ... intrusion abnormality determination unit 235 ... shielding abnormality determination unit 24 ... communication unit

Claims (5)

In an image monitoring device that detects an intruder using an image of a monitoring space,
An imaging unit that sequentially images the monitoring space;
A storage unit for storing a background image;
A change area extraction unit that extracts a change area by comparing the captured monitoring image with the background image;
An object region detection unit for detecting an object region by approximating the outer shape of the change region with a predetermined figure;
A movement distance calculation unit that associates object areas having similar characteristics to each other among the object areas detected at the time before and after, and calculates a movement distance from a detection position of the associated object area;
The moving distance of the object area is compared with a preset stationary determination value, and the ratio of the changed area to the object area is calculated and compared with a preset shielding tolerance value. A shielding abnormality determination unit that determines that there is a shielding object that prevents detection of the intruder when the ratio is less than the determination value and the ratio is equal to or greater than the shielding tolerance value;
An image monitoring apparatus comprising: The image monitoring apparatus according to claim 1,
When there are a plurality of the change areas detected in proximity to each other within a preset allowable range, the object area detection unit detects the object area with respect to a group of the plurality of change areas. . The image monitoring apparatus according to claim 1 or 2,
The shielding abnormality determination unit determines that the object area in which the moving distance is less than the stationary determination value and the ratio is less than the shielding tolerance value is a low shielding object area that does not hinder the detection of the intruder,
The change area extraction unit is an image monitoring apparatus that replaces the background image in the low shielding object area with the monitoring image. The image monitoring apparatus according to claim 3.
The shielding abnormality determination unit stores the change region extracted in the low shielding object region in the storage unit, and combines the change region extracted after the storage and the stored change region. An image monitoring apparatus that calculates the ratio using a computer. The image monitoring apparatus according to claim 4,
The shielding abnormality determination unit stores the low shielding object region in the storage unit, and calculates a ratio of the composite change region in an overlap region of the object region and the stored low shielding object region detected after the storage. An image monitoring apparatus for calculating and determining the shielding object.