EP0939387B1 - Room supervision device - Google Patents

Description

The present invention relates to a device for room surveillance, with at least an image sensor and at least one presence or motion detector and with one Control and evaluation electronics to which the at least one image sensor and the at least one presence or motion detector are connected.

In a device of this type described in EP-A-0 772 168, the presence or Motion detector exclusively to reduce the power consumption of the preferred battery powered device by turning on the image sensor when needed. something similar applies to the video surveillance system described in GB-A-2 309 133, whose image sensor connected to a VCR. Here the presence or motion detector enables better utilization of the magnetic tape used for image storage, by turning on the VCR only when needed.

These known devices therefore provide stored image information that must be evaluated by an operator, which is known to be extraordinary is monotonous and tiring. Since there is no distinction between the recorded events Tolerability and reportability takes place, can not speak of intelligent surveillance become.

In a device described in DE-U-297 18 312, which has an image sensor, a stage for Still image creation and still image storage, a stage for creating difference images and one Contains the stage for image analysis, the analysis result is based on predefined criteria of reportability compared, and if the result is positive, a message is sent to a central station. While this facility allows for intelligent surveillance by checking the signals of the image sensor for tolerance and reporting reliability are carried out automatically, so that a message is only sent to the head office if a positive analysis result is available.

But since the image sensor relies on lighting in the visible or in the near infrared and therefore provide no evaluable images below a certain minimum lighting , it is questionable whether this facility offers the desired security or, with others Words whether it is sufficiently robust.

The invention is now to provide a device of the type mentioned at the outset, which is intelligent surveillance, optimal security against deception and sufficient Robustness guaranteed.

According to the invention, this object is achieved in that the control and evaluation electronics has a processing stage for local signal evaluation, in which a Combined evaluation of the signals of the at least one image sensor and the at least one A presence or motion detector is used to determine the mean lighting of the monitored room and that in the combined evaluation of the Signals the weighting of the signal of the at least one presence or motion detector (6) depends on the average lighting and with decreasing average lighting increases.

The device according to the invention is therefore a two- or multi-criteria motion detector at least one image sensor. This detector points towards both known dual detectors as well as significant advantages over pure image sensors. Compared to the dual detectors, where no or only a very rough spatial resolution is possible, so that A distinction between humans and animals can often not be made, is the inventive Detectors much more robust. The image sensor also offers the option of Classification of objects based on their geometry and movement and thus the intelligent Monitoring and verification and storage of events and later Accessibility from these.

Compared to a pure image sensor, the detector according to the invention has the advantage that it even in poor lighting conditions the performance of a presence or Provides motion detector and is therefore fully functional. In addition, the presence or Motion detector the image processing algorithms when interpreting difficult situations support.

A first preferred embodiment of the device according to the invention is characterized in that that prior to the combined evaluation of the signals, a separate pre-evaluation both the signals of the at least one image sensor and that of the at least one a presence or motion detector.

A second preferred embodiment of the device according to the invention is thereby characterized in that the at least one image sensor by a CMOS image sensor, preferably an active pixel sensor is formed.

The formation of the at least one image sensor as a CMOS image sensor has, among other things the advantage of an order of magnitude lower power consumption than CCD cameras and the ability to access individual pixels. The latter allows pictures with reduced resolution or only read out interesting areas of the image, whereas with CCD cameras the pixels only can be read out line by line.

A third preferred embodiment of the device according to the invention is thereby characterized in that a means of determining the distance of one of the at least a presence or motion detector detected object and that the Distance signal of this means is fed to the processing stage.

Fig. 1

ein Blockschema eines erfindungsgemässen Einrichtung zur Raumüberwachung; und

Fig. 2

ein Flussdiagramm zur Erläuterung der Signalverarbeitung.

In the following the invention is explained in more detail using an exemplary embodiment and the drawings; it shows:

Fig. 1

a block diagram of an inventive device for room monitoring; and

Fig. 2

a flowchart to explain the signal processing.

The device for room monitoring shown in Fig. 1 consists essentially of a multi-criteria motion detector 1, hereinafter referred to as a detector, and one hereinafter referred to as electronics control and evaluation electronics 2 with a control stage 3 and a processing stage 4 for the local evaluation of the signals of the detector 1. Of course, this representation only shows the individual function blocks and not the specific ones apparatus construction of the facility again. For example, certain are usually Parts of the electronics are present in detector 1. The latter applies in particular if the detector signals are processed or pre-evaluated in detector 1.

The detector 1 consists of at least one image sensor 5 and at least one following Presence or motion detector 6 referred to as P / B detector. The image sensor 5 is included a means for measuring the lighting in the monitored room. Both that Image sensor 5 and the P / B detector 6 are each followed by a preprocessing stage 7 or 8, but also included in detector 1 or, if applicable, in processing level 4 can be. The signals reach the processing stage from the preprocessing stage 7 and 8 4, which also the illumination signal of the means for illumination measurement contained in the image sensor 5 is fed.

As an option, the detector 1 can have a distance meter 9 for determining the distance from the P / B detector 6 events included. Processing level 4 is autonomous and enables local decisions and / or the display of those recorded by the image sensor 5 Pictures; it is preferably also for transferring the images to a distant one Central 10 equipped, where, for example, additional verification can then take place.

The P / B detector 6 is a motion or presence detector based on one of the known detection principles Passive infrared, active infrared, microwaves, ultrasound or any Combination of them. The image sensor 5 is in the range of visible light, in the near infrared or sensitive to long-wave infrared (heat radiation); he can on a usual imaging Process (CCD [CCD: charge-coupled device], CID [CID: charge injection device] or CMOS [CMOS: complementary metal oxide semiconductor = complementary metal oxide semiconductor structure]) based.

Preferably, a special CMOS image sensor, a so-called APS [APS: active pixel Sensor] used, which is characterized by a very low power consumption and by the possibility of access to individual pixels. In addition, in such APS additional application-specific analog or digital functions, for example simple Image processing algorithms such as filters or exposure control can be easily integrated. Regarding APS, the article "A 128 x 128 CMOS Active Pixel Image Sensor for Highly Integrated Imaging Systems "by Sunetra K. Mendis, Sabrina E. Kennedy and Eric R. Fossum, IEDM 93-538 and "128X128 CMOS Photodiode-Type Active Pixel Sensor With On-Chip Timing, Control and Signal Chain Electronics "by R.H. Nixon, S.E. Kemeny, C.O. Staller and E. R. Fossum in SPIE Vol. 2415/117.

The image sensor 5 is directed towards the space to be monitored, detects it in terms of image technology, digitizes the image and stores it as a reference image in a memory. If the the image sensor 5 forming APS consists of 128 by 128 pixels, for example, would then be used a wide-angle lens at a distance of 15 m in front of the image sensor 5 one pixel of approximately 12 by 12 cm. Such a grid is suitable, human and distinguish animal shapes relatively reliably from each other.

The possibility of recognizing a person at a distance of 15 m is very advantageous, and the area of the monitored room at this detection distance of about 15 by 15 m quite realistic. In the active state of the device, the image sensor 5 then makes intervals a picture of the monitored room from fractions of a second is saved Images for a certain time and compare them with the reference image and / or with each other. The storage of the images is preferably controlled in such a way that the images sent by the detector 1 immediately preceding and / or immediately following the detected event until saved further and that the remaining images automatically after the specified time to be deleted.

To still make a usable picture even in poor lighting conditions , the image sensor 5 is sensitive to light and has a large dynamic range (sufficient recognizability of details with strong light / dark contrast) optimized. The functions integrated on the APS chip allow an automatic electronic Integrate closure with a dynamic of 1: 1000.

The P / B detector 6 essentially serves to detect the potential weaknesses of the image sensor 5 compensate. This includes in particular that the image sensor 5 is below a critical one Illumination provides no image information, and that also a strong change in the image information is often not caused by an intruder compared to the reference image. you think, for example, of abrupt changes in lighting that by switching the street lighting on / off, passing vehicles with the lights on Headlights, thunderstorms or the like can be caused. In these and the like In cases where the signal from the P / B detector 6 is taken into account, the robustness of the Detector 1 increased significantly. This is taken into account by a combined evaluation the signals of the image sensor 5 and the P / B detector 6 in the processing stage 4 realized.

It has been shown that it is advantageous to use the signals from the image sensor 5 and the PIB detector 6 to subject the combined evaluation to a separate preliminary evaluation, which in preprocessing stages 7 and 8. As already mentioned, the preprocessing stages 7 and 8 in the respective sensor / detector 5 or 6 or in the processing stage 4 be integrated. During the pre-evaluation, the signals from the P / B detector 6 in converted a format suitable for combined evaluation with the signal of the image sensor 5 and classified according to their strength. If the detector 1 contains a distance meter 9, then this becomes corresponding in the presence of a signal from the P / B detector 6 Starch from processing level 4 activated via control level 3 and delivers to the processing level information about the distance of the event just detected or Object from detector 1. With the help of this distance information, the signal from the image sensor can be used 5 clear conclusions on the size of a detected object and thus on its size Nature, such as humans or animals.

In the case of the image sensor 5, the pre-evaluation is in the form of hardware and / or in the form of a processor core integrated on the APS chip. When pre-evaluating the signals, the number of compared the modified image of the reference image, its clustering and characteristics of the pixel cluster certainly. The reference image is constantly updated, whereby Changes are only adopted in the reference image after checking their stability. The reference image tracking can be done by including the signals of the P / B detector 6 run more stable.

At the input of processing stage 4 there is a signal of the P / B classified according to strength Detector 6, an image signal of the image sensor 5 with information about the number of changes Pixels and the characteristics of the pixel clusters and possibly a signal from the distance meter 9, the distance of the event triggering the signal of the P / B detector 6 indicates. In addition, the processing stage 4 continuously receives information about the image sensor 5 medium lighting in the monitored room and weighted in the combined evaluation the signals the signals of the lighting-independent P / B detector 6 the weaker the more the lighting is.

The combined evaluation of the signals gives a decision at the output of processing stage 4 Alarm or non-alarm, this decision essentially the image content, the overall lighting, the information from the P / B detector 6 and the change and / or history of these parameters, hereinafter referred to as global criteria, are taken into account. Plausibility links are very helpful here: if, for example, lighting and change the image content quickly and strongly, but the signal of the P / B detector 6 is weak, then no alarm is given, but the stability of the new image is checked and in it searched for movements. If the P / B detector 6 reports no movement and only If there was a change in lighting and / or image, it was almost certain to become one Illumination switched on or off so that there is no alarm. Whether it's actually a change of lighting has taken place, can easily be checked based on the stability of the new image become.

• Geringe Anzahl geänderter Pixel absolut oder pro Häufungsbereich: Unterkritische Änderung, die nicht für ein Alarmereignis genügt. Es findet keine weitere Auswertung statt.
• Mittlere Anzahl geänderter Pixel absolut oder pro Häufungsbereich: Es findet eine Detailanalyse statt.
• Grosse Anzahl geänderter Pixel absolut oder pro Häufungsbereich: Es erfolgt zuerst eine Überprüfung der globalen Kriterien und sofern diese keine vernünftige Erklärung liefert, anschliessend eine Detailanalyse.

The image changes can be divided into three categories:

• Small number of changed pixels absolute or per cluster: subcritical change that is not sufficient for an alarm event. There is no further evaluation.
• Average number of changed pixels in absolute terms or per cluster: A detailed analysis takes place.
• Large number of changed pixels in absolute terms or per cluster: First, the global criteria are checked and, if this does not provide a reasonable explanation, then a detailed analysis.

The detailed analysis is carried out both statically and dynamically. In the first case, the cluster areas examined for size, topology and orientation, in the second case regarding the change in size, shape and position of the cluster areas.

Essentially, an attempt is made to move or move from the fixed reference image Filter out objects and then assign them a clear, alarm-relevant category. For example, the difference between humans and animals is the height / aspect ratio of the frequency range in question, the vertical and the horizontal predominates in animals. From the front there is a distinction between people and Animal very difficult; but you can assume that the detected object is not frontal moved towards the detector 1 but sidesteps and is therefore easier to see. In addition to these quantitative criteria for static detailed analysis, one can use dynamic Detailed analysis of typical movements of humans and animals and save and compare the changes detected by the image sensor 5 with these patterns.

2 shows a simple example of signal processing in a device for room monitoring without distance meter 9 (Fig. 1). For simplicity's sake the representation assumed that the preprocessing stages 7 and 8 in the processing stage 4 (FIG. 1) are integrated, so that the preprocessing of the signals of the image sensor 5 and the P / B detector 6 not locally in the relevant sensor or detector, but instead takes place in processing stage 4.

As shown, following the preprocessing of the sensor signals, a decision is made as to whether the room brightness is sufficient for the image sensor 5. If this is not the case, ie if the room brightness is very low, then only the signal of the P / B detector 6 is examined. In this investigation, the latter signal is compared with a threshold value P ₁ , the size of which approximately corresponds to the alarm threshold of a P / B detector of the type in question. If the PB detector 6 is a passive infrared detector, then the threshold value P _{1 is selected in} accordance with the alarm threshold of such a passive infrared detector. If the signal of the P / B detector 6 is greater than P ₁ , an alarm is triggered, if not, then the processing of the relevant sensor signal is terminated and the next processing cycle is carried out.

If the room brightness is sufficient for the image sensor 5, then the number of pixels changed compared to the reference image is determined. If this number is zero or very small, then the signal of the P / B detector 6 is compared with a threshold value P ₂ with P ₂ > P ₁ . If the signal of the P / B detector 6 is greater than P ₂ , an alarm is triggered, if not, then the processing of the relevant sensor signal is terminated. The signals can possibly also be analyzed over a longer time interval.

If the number of pixels changed from the reference image is of medium size, then a detailed analysis with consideration of the lighting situation is carried out with the following Examination whether an object from both detectors (image sensor 5 and P / B detector 6) was detected. If this is the case, an alarm is triggered, if not, then the processing of the relevant sensor signal canceled.

If the number of pixels changed compared to the reference image is large, then it is examined whether there has been a strong change in lighting, which is done on the basis of the lighting measurement by the image sensor 5, and in addition the signal of the P / B detector 6 is smaller, a threshold value P ₃ , with P ₃ «P ,. If both conditions are met, the processing of the relevant sensor signal is terminated. If both conditions are not met, then a detailed analysis is carried out taking into account the lighting situation, followed by an investigation as to whether an object was detected by both detectors (image sensor 5 and P / B detector 6). If this is the case, an alarm is triggered, if not, the processing of the relevant sensor signal is terminated.

A much improved distinction between is achieved in the manner described Humans and pets and humans and insects, making two of the main ones Sources of false alarms have been eliminated. Of course, this distinction suddenly improved by the distance signal of the distance meter 7, because then a clear It is possible to draw conclusions about the size of the detected object.

Claims (15)

1. Space monitoring device, comprising at least one image sensor (5) and at least one presence or movement detector (6), and comprising a control and evaluation electronics system (2) to which the at least one image sensor (5) and the at least one presence or movement detector (6) are connected, characterized in that the control and evaluation electronics system (2) comprises a processing stage (4) for local signal evaluation in which a combined evaluation of the signals of the at least one image sensor (5) and of the at least one presence or movement detector (6) is performed, in that the average illumination of the monitored space is determined, and in that, in the combined evaluation of the signals, the weighting of the signal of the at least one presence or movement detector (6) is dependent on the average illumination and increases with decreasing average illumination.
2. Device according to claim 1, characterized in that a separate preliminary evaluation (7, 8) of the signals of the at least one image sensor (5) as well as of the signals of the at least one presence or movement detector (6) is performed prior to the combined evaluation of the signals.
3. Device according to claim 2, characterized in that the said preliminary evaluation (7, 8) is performed in the respective sensor or detector (5 or 6) or in the processing stage (4).
4. Device according to claim 1 or 2, characterized in that the at least one image sensor (5) consists of a CMOS image sensor, preferably an active pixel sensor.
5. Device according to one of claims 1 to 4, characterized in that a memory is provided for the images recorded by the image sensor (5), and in that this memory is controlled so that it stores those images that, in combination with the signal of the presence or movement detector (6), have resulted in an alarm decision.
6. Device according to claim 5, characterized in that, in addition, the images immediately preceding and/or following the said images are stored.
7. Device according to claim 5 or 6, characterized in that an interface is provided for reading out the stored images, and in that the reading-out is preferably performed with a PC.
8. Device according to one of claims 5 to 7, characterized in that a means is provided for transferring the said stored images to a terminal that is spatially separated from the monitoring device.
9. Device according to one of claims 1 to 8, characterized in that a means (9) is provided for determining the distance of an object detected by the at least one presence or movement detector (6), and in that the distance signal of this means is supplied to the processing stage (4).
10. Device according to claim 4, characterized in that, in the preliminary evaluation (7) of the signals of the at least one image sensor (5), the number of changed pixels compared to a reference image and/or their accumulation and/or their distribution over the image is determined.
11. Device according to one of claims 1 to 10, characterized in that, if there is insufficient illumination for the operation of the image sensor (5), only the signal of the presence or movement detector (6) is evaluated.
12. Device according to claim 11, characterized in that the signal of the presence/movement detector (6) is compared with a first threshold value (P₁) whose magnitude corresponds approximately to the alarm threshold of a presence or movement detector of the relevant type.
13. Device according to one of claims 1 to 12, characterized in that, if the illumination is sufficient for the operation of the image sensor (5), the signal of the presence or movement detector (6) is compared with threshold values (P₂, P₃) which are dependent on the number of changed pixels in the image recorded by the image sensor (5), compared to a reference image.
14. Device according to claim 13, characterized in that, in the case of only very few changed pixels compared to the reference image, the signal of the presence or movement detector (6) is compared with a second threshold value (P₂) lying above the first threshold value (P₁), this analysis being able to be performed over a relatively long time interval.
15. Device according to claim 13, characterized in that, in the case of many changed pixels compared to the reference image, the signal of the presence or movement detector (6) is compared with a third threshold value (P₃) lying significantly below the first threshold value (P₁), and in that a further processing of the sensor signals is performed only if the signal of the presence or movement detector (6) does not simultaneously lie below the third threshold value (P₃) and a large change in illumination has been recorded by the image sensor (5).

Images