JP2001008193A - Image sensor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect an insect or the like adhered to a monitor window by photographing a scene at the outside of the monitor window via a reflecting mirror and applying image processing to the image while being divided into an image area used for conventional photographing and an image area of the monitor window. SOLUTION: A photographed image of an image sensor has a looped back visual area where an image reflected in a mirror 5 is photographed at an upper part of an image display range and a monitor camera visual area of a monitor camera photographing the monitor area. A front face of a monitor window 4 is always displayed on the looped back visual area. Simultaneously the monitor camera visual area displays a video image of a monitor area from the image sensor. When an object is detected in the looped back visual area, whether or not the detected object is moved is discriminated. When the detected object is moved, it is discriminated to be an insect, an insect flag is set and when the object is not moved, it is discriminated to be dirt. When the monitor camera visual area detects a mobile object, whether the mobile object is an insect or a human body is discriminated based on the insect flag.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image sensor for photographing a surveillance area through a surveillance window to determine the presence or absence of an object to be detected, such as an intruder. The present invention relates to an image sensor capable of distinguishing a non-detection target such as an intruder and a detection target such as an intruder located at a position away from a monitoring window.

[0002]

2. Description of the Related Art A conventional image sensor continuously captures images of entrances and exits of a building and other monitoring areas, compares the captured image with a reference image, and intrudes from the size of a changing area. There is an image sensor that determines the presence or absence of a person.

In these conventional image sensors, a sudden change in an image caused by an intruder to be detected is distinguished from a slow change in an image caused by an environmental change such as sunlight entering a monitoring area. For this purpose, a reference image that is sequentially updated with the latest photographed image at regular time intervals is used.

FIG. 7 is a view showing an image photographed when an insect sticks to the monitoring window of the imaging means.

[0005] As shown in FIG. 7, an image pickup means (hereinafter, referred to as a “surveillance camera”) installed to photograph a desired surveillance area.
Insects 201 attached to the monitoring window of the intruder appear on the captured image as a moving object having substantially the same size as a distant intruder. Further, when the insect moves in the direction of the arrow 202, the insect appears in the same state as the movement of the intruder.

[0006]

As described above, in the conventional image sensor, an insect stuck to the monitoring window of the surveillance camera and an intruder located away from the surveillance camera have an image of substantially the same size. Become. Furthermore, when an insect moves, it assumes a state similar to the movement of an intruder. Therefore, an insect stuck to the monitoring window of the monitoring camera may be erroneously determined to be an intruder, and there is a problem that the reliability of the image sensor is impaired due to a false report.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to solve the conventional problems and to distinguish between a non-detection target such as an insect stuck to a monitoring window of a monitoring camera and a detection target such as an intruder. An object of the present invention is to provide a highly reliable image sensor capable of detecting a detection target without causing a false report by the target.

[0008]

Means for Solving the Problems The inventors have conducted intensive studies in order to solve the above-mentioned conventional problems. As a result, the outside of the surveillance window is photographed through a reflector,
It has been found that insects and the like stuck to the monitoring window can be detected by performing image processing by dividing the image region into an image region used for normal monitoring and an image region of the monitoring window.

[0009] The present invention has been made based on the above-mentioned knowledge, and a first aspect of the image sensor according to the present invention is an imaging means for photographing a monitoring area through a monitoring window to obtain a monitoring area image. And a reflecting mirror disposed at a position where a monitoring window image can be obtained by photographing the outer surface of the monitoring window by the imaging unit, a predetermined reference image, the captured monitoring region image, and the Based on the monitoring window image, the change area appearing in the monitoring window image is determined as a non-detection target,
The image sensor includes an image processing unit that identifies and processes a detection target and a non-detection target.

In a second aspect of the image sensor according to the present invention, the image processing means processes the monitoring window image and the monitoring region image individually, and applies the processing to the monitoring window image and the monitoring region image, respectively. An image sensor characterized in that the non-detection target is determined based on a correspondence relationship between appearing positions.

In a third aspect of the image sensor according to the present invention, when the non-detection target moves within a predetermined time, the image processing means determines the non-detection target as an insect, and determines whether the non-detection target is a predetermined time. The image sensor is characterized in that the non-detection target is stained when it is stopped for a period of time, and it is determined that there is a device abnormality including obstruction of the visual field.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, an embodiment of the image sensor according to the present invention will be described in detail.

According to the image sensor of the present invention, it is possible to obtain a monitoring window image by capturing an image of a monitoring region via a monitoring window to obtain a monitoring region image, and capturing an outer surface of the monitoring window by the imaging device. Based on the reflecting mirror disposed at the position where it can be performed, a predetermined reference image, and the captured monitoring region image and monitoring window image, a change region appearing in the monitoring window image is determined as a non-detection target, and The image sensor includes an image processing unit that identifies and processes a non-detection target.

That is, in the image sensor of the present invention,
A reflecting mirror is provided, and the reflecting mirror is arranged at a position where an image of the outer surface of the monitoring window is captured by the imaging means to obtain a monitoring window image. Therefore, the surface of the monitoring window is photographed through the reflector.

Further, in the image sensor according to the present invention,
The image processing means individually processes the monitoring window image captured through the reflecting mirror and the monitoring area image directly captured by the imaging means, and corresponds to a position appearing in each of the monitoring window image and the monitoring area image. A non-detection target is determined based on the relationship. That is, non-detection targets, for example, insects, dirt, etc., appear in both the monitoring window image and the monitoring area image. A detection target, for example, a person appears only in the monitoring area image.

Further, in the image sensor according to the present invention,
The image processing means determines that the non-detection target is an insect when the non-detection target moves within a predetermined time, and stains the non-detection target when the non-detection target is stopped for a predetermined time, thereby obstructing the visual field. It is determined that the device is abnormal. That is, not only the sticky insects but also dirt on the monitoring window is detected.

That is, the movement is detected by the monitoring window image,
If there is a movement, it is determined as an insect, and if there is no movement, it is determined as a stain. Further, the movement is detected based on the monitoring area image, and if there is a movement, it is determined that there is a high possibility of being a detection target.

Here, the reference image will be described. The reference image is an image of the monitoring area in a state where the detection target object is not shown, and is stored in advance. The reference image may be sequentially updated at predetermined time intervals, or may be an image stored at an early stage when the image sensor is installed.

Hereinafter, the image sensor of the present invention will be described in more detail.

FIG. 1 is a side sectional view of one embodiment of the image sensor of the present invention. The image sensor has a monitoring camera 2 as an imaging unit and a substrate 3 inside a housing 1. Substrate 3
Controls the imaging of the surveillance camera 2 and performs processing of the obtained imaging signal and output of an abnormality based on the signal.

A cover for concealing the surveillance camera 2 from the outside is attached to the front of the surveillance camera 2 to form a surveillance window 4. A mirror, ie, a mirror 5, is fixedly arranged. The mirror 5 is arranged in a part of the surveillance camera field of view 6 of the surveillance camera 2.

FIG. 2 is a view for explaining a picked-up image obtained by the image sensor having the structure shown in FIG. FIG. 2 (a)
Indicates a model of the obtained captured image. The captured image of the image sensor of the present invention has, at the upper part of the image display range, a folded visual field area A reflected by the mirror 5 and a surveillance camera visual field area B photographing the monitored area. The surface of the cover (monitoring window) 4 shown in FIG. At the same time, an image of the monitoring area of the image sensor is displayed in the monitoring camera view area B.

FIG. 2B is an image obtained when there is an insect near the monitoring window 4 of the image sensor or when there is dirt attached to the monitoring window 4. As shown in FIG.
3 is reflected in both the folded visual field area A and the surveillance camera visual field area B.

Next, as shown in FIG. 2 (c), the return visual field area A and the surveillance camera visual field area B are represented by A _pq
And matrix elements such as B _pq . The divided matrix elements are used in the operation of the image sensor described below.
An object for judging insects, dirt and a human body is extracted to obtain its position, which is used to display where the extracted object appears in the matrix element shown in FIG.

FIG. 3 is a flowchart for explaining the operation of the image sensor of the present invention.

In S1, a reference image of the folded visual field area A and the surveillance camera visual field area B shown in FIG. 2 is obtained, and a human body value R for human body detection determination and an output indicating that dirt has been detected are determined. Dirt count Y is 0
And

In the insect and dirt determination in S2, when an object is detected in the folded visual field area A, it is determined whether or not the detected object has moved. When the detected object moves, the object is determined to be an insect, and the insect flag is set.
Turn ON. The insect flag is turned off when no object is detected, or when an object is detected but does not move.

If the object is detected but has not moved, it is determined that the object is dirty, and the above-described dirt count Y is incremented by one. In this way, when the time counted by the stain count Y exceeds the predetermined value, it is determined that the detected object is dirty. At this time, the device abnormality of S5 is output via the branch shown in FIG.

When it is not determined that the object is soiled, that is, when it is determined that the object is an insect, or when it is determined that the object is soiled, but when the soil count Y is equal to or less than a predetermined value, or when the object is If not detected,
The processing of S3 is performed via the branch shown in FIG. The insect and dirt determination in S2 will be described later in detail with reference to FIG.

In the human body determination in S3, when a moving object is detected in the field of view B of the surveillance camera, whether the extracted moving object is an insect or a human body is distinguished by an insect flag.

When it is determined that there is a human body in the monitoring area,
Then, the human body detection of S4 is outputted through the branch shown in FIG. On the other hand, if it is determined that it is not a human body, the process returns to S2 and the same processing is repeated. The human body determination in S3 will be described later in detail with reference to FIG.

In S4, the image sensor detects a human body,
Alternatively, after the device abnormality is output in S5, the insect flag is turned off in S6, the dirt count Y is cleared to 0, and the process proceeds to S2, where the image sensor continues monitoring.

FIG. 4 is a flowchart for explaining the logic for judging insects and dirt.

In S31, it is determined whether or not the current image of the folded visual field area A includes an object that is not included in the reference image by taking a difference from the reference image. When the object not included in the reference image is not included in the current image, that is, when there is no change, the process proceeds to S38 and the insect flag is set.
After the OFF and dirt flags are cleared to 0, the process proceeds to the next process via the output terminal. Conversely, in S31,
If an object that is not in the reference image has been extracted in the current image, the process proceeds to S32.

The processing after S32 is processing relating to calculation of the position and movement of the object when the object not present in the reference image is extracted in the current image.

In step S32, as shown in FIG. 2C, the object extracted from the current image divides the folded visual field into matrix elements A _pq and determines which matrix element A _{pq the} object is reflected in. And the result is represented by the position C of the object in the current image.
Store in C _i as _i = A _pq .

[0037] determined in whether the position C _i-1 of the object obtained in this way positions C _i of the object in the current image obtained by the previous image is the same S33, the same, i.e.,
When C _i = C _i−1 , it is determined that the object has not moved, that is, it is determined that the object is highly likely to be dirt, the insect flag is turned off, and the dirt count Y is incremented by 1 in S34. Move to.

Conversely, when the positions are different, ie, C
_{When i} ≠ C _i−1 , it is determined that the object is moving, that is, the object is an insect, the insect flag is turned on, and the dirt count Y is cleared to 0, and the process proceeds to S37. S3
The determination result shifted to 7 shifts to the next process via the output terminal indicated by.

The result of the judgment at the step S34, that is,
When it is determined that the object extracted in the movement determination does not move and is highly likely to be dirty, it is determined in S35 whether or not the stain count Y exceeds a predetermined value.

When the dirt count Y is equal to or more than the predetermined value, it is determined that the extracted object is dirt, and the processing shifts to the next processing via the output terminal shown in FIG. However, the dirt count Y
Is less than or equal to the predetermined value, the extracted object is not yet determined to be dirty, and the process proceeds to the next process via the output terminal indicated by.

It should be noted that the stains, once generated, continue unless artificially cleaned or the like. In addition, the dirt takes a long time until the surveillance camera view area B is completely closed.
For this reason, the determination of the dirt is determined to be dirt, for example, when the dirt count is counted up continuously for one week.

In the current image, when a moving object not present in the reference image is extracted in the surveillance camera view area B shown in FIG. 2A, it is determined that a human body exists in the surveillance area. FIG.
Is a flowchart for explaining the human body determination logic.

In step S41, a difference process is performed between the current image and the reference image, and it is determined whether or not an object not present in the reference image is included in the monitoring camera field of view B of the current image. The object in B is extracted. When an object that is not in the reference image is not included in the current image, it is determined that the human body does not exist in the monitoring area, that is, it is normal, and the processing shifts to the output terminal shown in and performs the next processing. Conversely, when an object is extracted, the processing after S42 is performed.

First, the position of the extracted object is set in S42.
And calculate. In S42, the object extracted from the current image
A matrix is needed to divide the field of view B of the surveillance camera shown in FIG.
Element B _pqWhich matrix element B_pqJudge whether it is reflected in
And the result is represented by the position D of the object in the current image._i= B_pq
As D_iTo be stored.

[0045] In S43, the determination whether the position D _i-1 of the object obtained by the position D _i and the previous image of the object in the current image are the same, the same, i.e. D _i = D _i- When it is _1, it is determined that there is no movement of the object, and the processing shifts to the output terminal shown in and the next processing is performed. However, D _i and D
_{When i-1} is different as shown by D _i ≠ D _i-1 , it is determined that the object has moved, and the process proceeds to S44.

Here, the movement of an object is determined by taking advantage of the fact that an intruder always moves within the monitoring area, while environmental fluctuations such as sunlight do not move suddenly. This is to reduce false alarms.

The process of S44 is a process of judging the state of the insect flag output from the "judgment and dirt judgment" described with reference to FIG.

When the insect flag is ON, that is, when there is an insect in the return visual field A, through the branch shown in FIG.
In S45, the extracted moving objects are collated to determine whether they are the same or foreign.

The collation of the moving object is performed by confirming the correspondence between the positions Ci _-1 and _Di-1 of the individually extracted objects in the return visual field area A and the surveillance camera visual field area B.

The correspondence of the positions is performed using a matrix element correspondence table shown in FIG. As a result of the position collation, when the correspondence between the extracted object positions Ci _-1 and _Di-1 does not match, it is determined that the object is a foreign substance, and the processing shifts to the next processing via the output terminal shown in FIG. When the correspondence between C _i-1 and D _i-1 is coincident, that is, when the extracted moving objects are the same, it is determined that the extracted moving object is an insect, and through the branch shown in FIG. , And the following processing is performed.

If it is determined in S45 that the moving object is a foreign body, that is, if the extracted moving object is likely to be a human body, the processing from S46 onward is performed via the branch shown in FIG.

In S44, the status of the insect flag is changed to OF.
F, or when it is determined that the moving objects extracted in S45 are not the same, in S46, various parameters of the object extracted from the difference between the reference image and the current image in the monitoring camera view area B, For example, the human body value R is calculated using the shape, the size, and the like.

In S47, it is determined whether or not the calculated human body value R exceeds a predetermined threshold value of the human body value R for the human body. S47
As a result of the determination in, when the human body value R exceeds the human body threshold value, it is determined that a human body is present in the monitoring area, and the processing shifts to the next processing via the output terminal shown in. R
If is smaller than the predetermined value, it is unlikely that a human body is in the monitoring area, and the process shifts to S48.

The determination of the human body value R is based on the shape,
This is performed based on the result of calculating parameters such as size and continuity. For this determination method, a method obvious to those skilled in the art is used.

If no human body is detected in the monitoring area, the process goes to S48 in S48 for the next determination.
Perform processing to clear the human body value R calculated in 6 to 0,
The processing shifts to the next processing via the output terminal shown in FIG.

[0056]

According to the present invention, stuck bugs can be detected, so that false reports caused by stuck bugs can be eliminated, and this is effective in providing a more reliable security service.

Further, according to the present invention, since dirt on the monitoring window can be detected, it is effective in the maintenance of the image sensor, and the reliability of the image sensor is improved.

[Brief description of the drawings]

FIG. 1 is a side sectional view of one embodiment of the image sensor of the present invention.

FIG. 2 is a diagram illustrating a captured image obtained by an image sensor having the structure shown in FIG. 1;

FIG. 3 is a flowchart illustrating the operation of the image sensor of the present invention.

FIG. 4 is a flowchart illustrating logic for determining insects and dirt.

FIG. 5 is a flowchart for removing stuck insects of the image sensor of the present invention.

FIG. 6 is a matrix element correspondence table.

FIG. 7 is a diagram illustrating a state of an insect stuck to a monitoring window;

[Explanation of symbols]

 1. Housing 2. Surveillance camera 3. Substrate 4. Monitoring window 5. Reflecting mirror 6. 6. Surveillance camera field of view Turned field 13. Insects, dirt

 ────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Hiroyuki Otsuka 8-10-16 Shimorenjaku, Mitaka-shi, Tokyo Secom Co., Ltd. F-term (reference) 5C054 CC06 CE04 FC01 GB12 HA18 5C061 BB05 CC01 5C084 AA02 AA07 BB05 BB40 CC19 DD11 DD61 GG57

Claims (3)

[Claims] 1. An image capturing means for capturing an image of a monitoring area via a monitoring window to obtain a monitoring area image, and capturing an outer surface of the monitoring window by the image capturing means.
A reflecting mirror disposed at a position where a monitoring window image can be obtained; a predetermined reference image; and a change region appearing in the monitoring window image based on the captured monitoring region image and the monitoring window image. An image sensor comprising image processing means for determining a non-detection target and identifying and processing the detection target and the non-detection target. 2. The image processing means individually processes the monitoring window image and the monitoring region image, and based on the correspondence relationship between the positions appearing in the monitoring window image and the monitoring region image, respectively. The image sensor according to claim 1, wherein a detection target is determined. 3. The image processing means determines that the non-detection target is an insect when the non-detection target moves within a predetermined time,
3. The image sensor according to claim 1, wherein when the non-detection target is stopped for a predetermined time, the non-detection target is determined to be a device abnormality including dirt and obstruction of a visual field. 4.