GB2031686A - Movement detection - Google Patents

Abstract

A movement detection system including a detector arrangement (205) for detecting the movement between a plurality of picture points from a selected portion of the picture. A control arrangement (206, 250) or (300, 206) determines that movement has occurred if at least some of the plurality of picture points have moved as detected by the detector. The movement detection is described for use in a noise reduction system. <IMAGE>

Description

SPECIFICATION Movement detection The present invention is concerned with video movement detection, which detection is typically required in various systems such as noise reduction, picture zoom and standards conversion.

In previous movement detectors, using for exam pie, a single picture point from two frames to deter mine if movement has occurred, it can be difficult for the detector to discriminate between movement and picture noise.

The present invention is concerned with providing enhanced movement detection.

According to the invention there is provided a video movement detector system comprising detector means for detecting movement between a plural- ity of picture points; and control means for determining that movement has occurred if at least some of the plurality of picture points have moved as detected by the detector means.

Further according to the invention there is provided a method of detecting picture movement in video information comprising detecting movement between a plurality of picture points and determining that movement has occurred if at least some of the picture points have movement detected thereon.

The invention will now be described by way of example with reference to the accompanying drawings in which: FIGURE 1 shows a known noise reduction system, FIGURE 2 shows an alternative system, FIGURE 3 shows a simple movement detector, FIGURE 4 shows an area of picture to be examined, FIGURE 5 shows one arrangement for improved detection, and FIGURE 6 shows an alternative improved detector.

As stated above movement detection is required in various systems such as noise reduction systems, picture zoom systems and standards converters.

In a noise reduction system for example where a portion of the picture information is adding to incoming picture information to effectively reduce the noise content in the picture, when movement has occurred it is necessary to reduce the amount of information fed back when the moving part of the picture is being processed otherwise smearing and other effects are visible when observing the video output on a T.V. screen.

The basic mechanism for noise reduction is the digital integration of video data within a frame store.

An example of a noise reduction system is shown in Figure 1 as disclosed in more detail in British Patent Application 42751/76.

The signal paths are digital signal paths. Each of the signal paths carrying video data will be capable of carrying at least 8 bit wide video data operating at a clock rate typically 15 MHz.

New video data is received by subtractor 230 which also receives old data from store 22. The subtractor output is received by coefficient unit 231 for multiplying the subtractor output by a selected coefficient K1 1. The output of unit 231 is added to the old video data in adder 232. The adder output is available for receipt by the store 22.

Port 1 is an input port to the digital frame store capable of carrying a full frame of television information where each storage location is widerthan 8 bits.

In a practical system a total of 12 bits may be used at each location in the video store. Such a video store would in practice have a capacity of approximately 6 megabits. The video store has a minimum of a single input port and a single output port. More typically the video store will have three ports (as shown), one of which is an input port whilst the second two are output ports. The arrangement of the store will enable the input and output to run nonsynchronously with respect to television field and line rates. Port 2 will be capable of running sync hronouslywith port 1 so that video data stored at the same picture location from a previous field may be accessed at the same time as new information for the same picture point is available in the next field.

Thus access to the top left hand picture element of field 1 is available from port 2 at the same time that the top left hand element of field 3 is available on the new video data input.

A video store of a type suitable for such a system is described in detail in British application 6585/76 for example.

An alternative noise reduction system is shown in Figure 2 and this time whilst the subtractor 230 receives old and new data as before, the adder 232 receives new data at one input (rather than old data) and receives the output from coefficient unit 231 as before at its other input.

In a simple noise reduction system the values of the coefficient would be fixed so that the portions of the previous data and incoming data would have a predetermined relationship.

However, a conflict exists between the requirements for noise reduction in which the maximum integration time should be used and the requirement that the picture shall maintain movement portrayal without distortion. The latter-movement portrayal requires the minimum integration time. Thus it is desirable to provide an adaptive mechanism which is sufficiently intelligent to adjust the coefficients for variable integration time depending upon picture content. This is where the requirement for movement detection arises.

If the incoming video information is stationaryfor example a test card transmission - coefficient kl 1, may be set to a value for optimum noise reduction, so that a large proportion of the previous picture information is added to the current information, picture point by picture point.

If the total picture is moving, for example during the "pan" of a camera then the coefficient k1 1 will require setting at a different value if movement por trayal is not to be distorted by the effect of the noise reduction system.

If part of the picture is stationary whilst part remains moving in order to obtain effective noise reduction over the stationary part of the picture whilst allowing movement portrayal without distortion, it becomes necessary to modify k1 1 on a picture point by picture point basis.

The first step towards a more effective coefficient modification system which adapts to picture move ment is the comparison of changes which have occurred between successive data values stored at the same picture location in sequential pictures, and such a mechanism for determining that changes are occurring in the picture is a movement detector.

A simple movement detector is shown in Figure 3 and comprises a subtractor 205 which receives the old and new data and its output is passed to comparator 206.

Each picture point in a new picture is subtracted from the data previously stored at the same location and the difference signal is applied to the comparator. If the difference signal exceeds a threshold level the picture is deemed to have moved and coefficient k11 is switched to a level which would be appropriate for movement Whilst the change remains within the threshold level a coefficient suitable for noise reduction of still images are utilised.

A read only memory (ROM) could be provided in place of the comparator and the subtractor 205 is used in a similar mannerto the earlier arrangement.

The mode of operation of a ROM and construction thereof is well known. The difference signal is now used to address the read only memory which contains various coefficient values of k1 1 in fixed storage locations. Large movements are differentiated from small movements and a sliding scale of coefficients is provided on a real time basis.

Movement detection employing the subtraction between old video data and new video data has been described on a single picture point basis. It has been found however that a more effective method of determining the difference between noise and true movement may be made if an area of picture is examined.

Figure 4 shows the area examination system. In the example given 9 picture points are arranged in the form of a square with a single picture point in the centre. Line N has picture points P1 to P3. To determine the total difference at picture point P5 a contribution from all surrounding picture points is made. This is effected in one embodiment by independently assessing each picture point in the area for threshold and using a logic gate (e.g. a majority logic gate) for detecting the threshold decision so that if a proportion are found to exceed the threshold movement is signalled to have taken place.

In this system, four out of nine is a typical practical setting to indicate movement. An example of such a system is illustrated in Figure 5. The difference signal is provided by the subtractor 205 as before and passes to comparator 206. The output passes to the majority logic gate 250. Only two of the nine inputs to the gate are shown in detail, although each input will have its associated subtractor and com parator.

A further method of utilising the area voting system does not make use of majority logic. In this method, the total area is integrated, so that the aver age difference over the area is compared with a threshold. The object is to differentiate between noise and movement Such an arrangement is shown in Figure 6 and comprises a subtractor 205 and comparator 206 as before but with an integrator 300. The data from current and previous picture points for the given area are subtracted picture point by picture point and the difference integrated in the integrator 300. The integrator output is received by comparator 206 which effectively designates that movement has occurred when the integrator output exceeds a predetermined threshold level.

The system has a particular advantage in a colour television system utilising NTSC or PAL colour subcarrier. Utilising the arrangement shown for area movement detection the system may be made insensitive to residual subcarrier. The design of a decoding circuit to eliminate residual subcarrier is simplified if a scheme is utilised which is generally insensitive to the presence of residual subcarrier components.

Although the area described has comprised 9 picture points, other numbers from one or more lines can be used. The larger the number of picture points integrated, the closer the noise is averaged towards zero, whereas movement remains unattenuated.

On the other hand, there is a disadvantage of using too large an area due to the emergence of a noise "halo" around moving subjects. The noise "halo" approximates to the size of the area linear dimensions in both the horizontal and vertical axis. It is over that dimension that it becomes impossible to detect the difference between noise and movement utilising the area voting system.

In an effort to increase the detection's system abilits to differentiate between noise and movement without increasing the area over which the decision is made, a combination of two fields are utilised. In this way, it becomes possible to double the number of points within the area (volume) close to the picture point and thus attenuate the noise without attenuating the movement.

The outputs from Figures 5 and 6 could be used to control further arrangements to vary the selected coefficients using ROM's for example as described above.

Claims (12)

1. A video movement detector system comprising detector means for detecting movement between a plurality of picture points; and control means for determining that movement has occurred if at least some of the plurality of picture points have moved as detected by the detector means.

2. A system as claimed in claim 1, wherein the detector means comprises a subtractor for detecting movement sequentially for each picture point within a selected portion and said control means comprises a comparator for determining whenever the output of said subtractor exceeds a predetermined threshold and gating means for detecting when the threshold has been exceeded by a predetermined number of the picture points within the selected portion.

3. A system as claimed in claim 1, wherein the detector means comprise a plurality of subtractors each for detecting movement for one picture point from a selected picture portion, and said control means comprise a comparator for each of the sub tractors to determine when a preselected threshold has been exceeded and gating means for detecting when the threshold has been exceeded from a predetermined number of the picture points within the selected portion.

4. A system as claimed in claim 1, wherein the detector means comprises at least one subtractor for detecting the difference between picture points and integrating means for integrating the differences for each of the picture points within the selected portion and said control means comprises a comparator for determining when the integrator output level exceeds a predetermined threshold.

5. A system as claimed in any one of claims 1 to 4, wherein the plurality of picture points are selected from a portion of one field and the movement is detected between picture points from a corresponding portion from a different field.

6. A noise reduction system including a movement detector as claimed in any one of claims 1 to 5, wherein the noise reduction system includes subtractor means for subtracting incoming and earlier picture information, modifier means for modifying the output of said subtractor means in dependence on a selected coefficient dependent on detected movement and adder means for adding the modified picture information to either incoming or earlier picture information.

7. A method of detecting picture movement in video information comprising detecting movement between a plurality of picture points and determining that movement has occurred if at least some of the picture points have movement detected thereon.

8. A method as claimed in claim 7, including subtracting picture points from corresponding portions from different fields to detect movement.

9. A method as claimed in claim 8, including integrating the differences from each of the subtracted picture points within the preselected picture portion.

10. A method as claimed in claim 8, including comparing each of the differences with a predetermined threshold to determine that movement has occurred if the threshold is is exceeded on at least some of the picture points.

11. A video movement detector substantially as described herein and with reference to Figures 4 to 6 of the accompanying drawings.

12. A method of video movement detection substantially as described herein.