GB2364608A - Video motion detector which is insensitive to global change - Google Patents

Abstract

A video motion detector having immunity to global change, achieved by a method of combining video readings taken on one frame of video to produce a characteristic value (CV) being a number which represents the intrinsic nature of that video frame. This CV is then stored and subsequently compared with a CV taken at a later time. If the amount of change is outside a permitted threshold an alarm is raised. The method of combining video readings to form a CV includes a method to divide the readings into two groups, then adding one group to the CV whilst subtracting the other group from the CV. Global changes, e.g. an increase in ambient brightness which would normally cause an increase in every video reading, are negated by this method. The video motion detector also has a method of controlling a video recorder, and raising an audible alarm. The simplicity of this method of globally-immune motion detection means that the device can be constructed with much less hardware than is usual, with the obvious commercial benefits.

Description

2364608 Title: Video Motion Detection The present invention relates to a

method and apparatus for detecting motion in a video picture.

According to the present invention there is provided a method of detecting motion in a video picture recorded by a video camera comprising the following steps: a) taking a video signal representing a first field, b) dividing at least part of the field into a number of zones c) applying a first algorithm to a factor related to the video signal in each zone to create a numerical value for each zone, d) applying a second algorithm to the value of each zone such that the numerical value of half the zones become negative and half remain positive, e) summing the positive and negative numerical values of all zones to provide a first field number, f) take a video signal representing a second field, g) repeating steps b) to e) to provide a second field number, h) comparing the first field number with the second field number, and i) generating an alarm signal if the there is a difference or substantial different between the first and second field numbers.

Preferably step c) comprises applying an algorithm to the video voltage value at a predetermined point on a scan line time base in each zone. Preferably the algorithm is applied by an analogue to digital converter applied to each voltage value to give a number.

In one embodiment the zones are provided by dividing the field into rows and columns to produce a number of rows of squares. Preferably 64 squares are provided by dividing the field into 8 rows and columns.

The invention also relates to an apparatus for performing the above method.

Preferably the apparatus further comprises a video controller to turn a video into a record mode upon generation of an alarm signal.

The invention also relates to a closed circuit television system, having a camera, a video recorder and an apparatus for performing the above method.

An embodiment of the invention will now be described with reference to the accompanying drawings in which: Figure I shows a schematic view of a first field divided into zones, Figure 2 shows a schematic view of the zones illustrating the effects of a second algorithm, applied thereto, and Figure 3 shows a block circuit diagram of an apparatus for performing the method of the invention.

Referring now to Figure 1, there is shown the image in a first field I generated by a video camera. The camera is recording the image a house. As is well known, the field is made up from a number of horizontal scan lines.

As shown, the field is divided into a plurality of zones by dividing the field into rows and columns to produce a number of rows of squares. In Figure 1, 64 squares are provided by dividing the field into 8 rows and columns.

Above the rows and columns is shown, by way of example, a graph of the video voltage plotted along the time base of a scan line. For each zone in every row, one voltage reading is taken along the time base at the mid point of each square in a row, i.e. at points A,B,C,D,E,F,G and H to create a factor relating to the video signal in each zone. The voltage at these points could be as follows for the first row of zones: A=2 B = 2.5 C = 3.5 D=4 E = 3.5 F = 2.5 G = 2.

As the speed each scan line is produced is very fast, the reading at point A for the first zone may be taken on the first scan line, and the reading at point B for the second zone may be taken say on the fourth scan line and so forth.

To the above factors related to the video signal is applied a first algorithm of 5 1. This is done by taking the voltage values and processing these through a standard 8 bit analogue to digital converter. This gives a numerical value for each zone in the first row as follows: A= 102 B = 127 C = 178 D = 204 E = 204 F = 178 G = 127 H = 102 If desired more than one reading can be taken in each row of zones, e.g. one reading at points A-H in each scan line, but obviously a more expensive processor would be required for this. Numerical values of each zone in the remaining rows are created in a similar manner.

As shown in Figure 2, a second algorithm is applied to the numerical value of each zone in all the rows such that the numerical value of half the zones become negative and half remain positive but in a certain sequence as shown in the pattern of Figure 2. Thus the value for each zone in the first row would be:

A= + 102 B = + 127 C = -178 D = -204 E = -204 F = -178 G = +127 H = +102 All the zone positive and negative numerical values are then summed to provide a first field number or first CV.

A video signal representing a second field generated by the video camera is taken This second field may be generated say half or one second after the first field. The second field may not necessarily be the next field after the first field generated by the camera, it is the time period between the first and second field which is important. The above steps of creating numerical values with a first algorithm for each zone is repeated, as well as applying a second algorithm and summing the total. This generates a second field number or second CV.

The first CV is compared with the second CV, and an alarm signal is generated if there is a difference or substantial different between the first and second CVs.

Referring now to Figure 3 there is shown a video camera 1, the output of which is connected to a standard video recorder 2 normally operated by a standard infra-red controller. The video camera may be used to generate images of the front of a house by pointing it at the front of a house as part of a security closed circuit television system.

The output of the camera is also connected to a video motion detector circuit 3 which performs the method described in Figures I and 2 above.

The output of camera I is first fed to a circuit I I to divide each field into 64 zones and take a voltage reading for each zone as described in Figure I to give an analogue voltage value for each zone.

The output of circuit I I is fed to 8 bit analogue to digital converter 12 to convert the analogue voltage values into a digital numerical value as described above.

The output of converter 12 is then processed by block 13 to apply the second algorithm mentioned above to generate the equal number of negative and positive numerical values for each zone and in the sequence shown in Figure 2, and to sum the total to create the Characteristic Value for each field.

Each time a Characteristic Value of a field is generated, block 14 compares this with a the Characteristic Value of the previous field which is stored in memory 15 and the output fed to an alarm signal generator 16. If the difference between a first CV and a second CV is below a predetermined value, then the second CV is passed for storage in memory 15. If the difference between a first CV and second CV is greater than a prcdeterTnined value then an alarm signal is generated which instructs an Infra Red Video controller 17 to turn on video 2.

In use therefore, camera I will normally generate a sequence of fields representing the image of the house it is pointing at. Because the house does not move, and assuming there are no moving objects in front of the house, each field will have a substantially identical CV. Thus no alarm signal will be generated by alarm signal generator 16. Video recorder will remain off and not record the image produced by the camera 1.

If an intruder should attempt to enter the house, as the intruder moves across the field of view of the camera 1, so the CV value of each field will change and this will cause an alarm signal to be generated and turn video recorder 2 on. The movement will be recorded by the video.

The apparatus of Figure 3 may be used for security purposes, for example monitoring a building such as a factory at night. In the morning the video can be played back and events which have generated movement will be recorded. This obviates the need for a person to watch hours of video playback without any movement. This also means that a three hour video tape will be sufficient to cover many days of surveillance.

A benefit of the invention is that it is insensitive to environmental changes such as changes of light intensity, but very sensitive to localised movements. The reason for this is that environmental changes produce changes to an entire image as viewed by the camera, whereas movement in partof the image is localised.

For example if the ambient brightness should increase (e.g. with the sun rising) all the numerical values for each zone will increase generally by the same amount, but because half are made positive and the other half made negative, the net effect to the CV in each of a sequence of fields will be generally unchanged. Also the benefit of continually updating the stored CV in memory 15 means that gradual changes such as shadow moving across the image is negated,

If there is localised movement, e.g. with an intruder walking from one zone to another as viewed by the camera, if that movement is from one "positive" zone to a "negative" zone then the difference to the total CV will be exaggerated as the net change will effectively be doubled. The sequence of positive and negative zones as shown in the pattern of Figure 2 has been found useful to create blocks of positive and negative zones which maximise the changes in CV values, and this has been found more effective rather than simply an arrangement of rows of alternative positive and negative zone. This pattern can be changed to suit different environments provided always generally half the zones are positive and half negative.

9 The predetermined value at which the alarm signal is generated (based on the difference between a first CV and second CV) may be adjustable to change the threshold at which the apparatus detects movement or not. This may be used as a sensitivity adjustment for the apparatus.

Also if after the video is turned on the apparatus detects a period of inactivity, the controller 17 may turn the video off.

In addition to turning the video on, the alarm signal may also be used generate an audible alarm such as a burglar alarm.

Instead of dividing the whole field into zones, it may be desirable to divide only part of the field into zones, i.e. just that area where it is desired to detect movement. For example, in Figure 1, it may be desirable just to divide the field into zones just around the location of the front door of the house, so as to discriminate against targets outside the area of the front door, e.g. so the drawing of a curtain in an upstairs window is not detected.

The time period between stored fields i.e. between each first and second fields may be adjusted for different applications. For example when detecting movement of a person a time period of half to one second has been found a workable value. If one was detecting movement of a fast co, r however, a time period much less than one second would be preferable. The time period is thus selected for each application depending on the likely time an image will move from one zone to another.

It is envisaged that the video motion detector 3 could be made in a separate housing for connection to a video camera and recorder.

The general principle of video motion detection however could be used in other applications.

It is also possible for the video controller 17 to include a circuit to adapt it to work with different infra red operating systems so that it can be used with any make of remote control video recorder. The video controller 17 could also be hard wired to control a video recorder.

In practice the video camera can be located remotely from the video motion detector 3 and video recorder 2.

A key advantage in detecting movement by providing just one CV per field is that the electronics can be simplified compared to known video movement detectors. This means that the invention may be produced at a lower cost.

Further modifications will be apparent to those skilled in the art without departing from the scope of the present invention.

Claims (9)

Claims:

1. A device for detecting motion in video images that is insensitive to global changes.

2. A device as claimed in Claim I where global changes such as ambient brightness change are rejected by virtue of dividing all video readings from one video frame into two (or more) groups, and combining these groups in such a way that any brightness change adding to the value of one group will detract from the value of another group thereby negating overall change in the frame.

3. A device as claimed in Claim 2 where a single characteristic value (CV) is derived to represent the position and contents and intrinsic nature of each inspected video frame, by combining the values of two or more groups of video readings in an algorithmic fashion whereby some groups (positive) are added to the CV and some groups (negative) are subtracted from the CV.

4. A device as claimed in Claim 3 where comparisons are made between the characteristic values (CV's) of temporally separated video frames, and if differences outside permitted thresholds are detected an alarm is raised, being audible or by other means, and also causes a recording of the video to occur.

5. A device as claimed in any preceding claim which clusters video readings into positive groups and negative groups, such that any object passing from an area covered by a positive group to an area covered by a negative group will cause a change in the CV equal to the sum of the changes in both groups, thereby increasing sensitivity to motion.

6. A device as claimed in any preceding claim which continually updates the stored characteristic value (CV) of the video picture such that the device is insensitive to slow changes such as the gradual movement of a shadow with the traverse of the sun.

7. A device as claimed in any preceding claim which automatically modifies the values of permitted thresholds of change in characteristic values, dependant upon the amount of background movement (such as leaves blowing in a breeze) that is occurring within the video image, thereby adjusting sensitivity and reducing the possibility of false alarm.

8. A device as claimed in any preceding claim that can display the area of coverage of motion detection on an attached video monitor or television, and can vary the area of coverage of motion detection within the video scene, either with direct operator control by use of a control fitted to the motion detector itself, or by some other means.

9. A device as claimed in any preceding claim including a method of reading and storing the output of a remote control handset for a video recorder, such that the invention can remotely cause a video recorder or other storage device to record an image when motion in the video image is detected by the invention, and stop recording after a period in which motion is no longer detected.