GB2256110A - Measuring and correcting frame position errors in film scanning - Google Patents

Description

2 2.5) 611 U 1

DESCRIPTION

MEASURING AND/OR CORRECTING FRAME POSITION ERRORS IN FILM SCANNING The invention relates to a method of and apparatus for measuring frame position errors in film scanning. The invention also relates to apparatus for correcting frame position errors in film scanning.

When scanning films for television display or for recording on video tape, frame position errors, which consist of a vertical and a horizontal offset between the individual film frames, become disturbingly noticeable. The frame position errors are essentially produced by irregular transport in the longitudinal direction and by unwanted movements with respect to time of the film in the camera during recording. Vertical frame position errors are particularly disturbing.

Various methods and devices for measuring frame position errors are already known. For example, in accordance with GB-A-2 165 417 motion vectors are formed which are used for interpolation of video signals of consecutive pictures. There are, however, various exceptional situations such as, for example panning, zooming and scene transitions which impede the interpolation. Moreover, this known device requires a considerable number of components.

A further proposal known from GB-A-2 139 037 is the comparison of the video signals of two consecutive frames, in which the absolute value of the differences of the pixels is integrated. However, this proposal has the drawback that it can only detect frame position errors of the order of one line.

It is an object of the present invention to provide a method of and apparatus for measuring frame position errors in which frame position errors may be detected and measured with a minimal number of components.

The invention provides a method of measuring frame position errors in film scanning comprising the steps of comparing the pixels of one column of a frame with pixels of the same column of 2 a time adjacent frame, vertically offsetting the pixels of the time adjacent frame a given number of times and repeating the comparison each time, and deriving a frame position error signal from the results of the comparisons.

Such a method has the advantage that frame position errors of the order occurring in practice can be measured with a relatively low number of components.

In one embodiment of the invention the vertical offset value whose comparison value assumes an extreme value with respect to the other comparison values is used as the frame position error value.

In a further embodiment of the invention interpolation between the extreme value and at least one proximate comparison value is performed to determine the frame position error. Frame position errors which are smaller than one line width can then be measured and corrected in accordance with the measurement.

The comparison values may be determined by cross-correlation and the offset at which the comparison value assumes a maximum is defined as the frame position error. The cross-correlation has the advantage that digital signal processors particularly optimised for this purpose are available. Alternatively it is also possible to determine the comparison values by addition of differences of the pixels to be compared and to define the offset at which the comparison value assumes a minimum as the frame position error.

As the pixels of a column are evaluated the selection of this column is significant in that the precision of the measured frame position error is dependent on the picture content within this column. If, for example, vertical black/white transitions are absent in this column, the frame position error cannot be measured unambiguously. To mitigate this problem, in a further embodiment, the location of the column is adjustable. In conjunction with a programming unit which is often used with film scanners, the location of the column can be preprogrammed and, when a film is being scanned, it can be automatically adjusted in 3 conformity with the preprogramming operation and the motion of the film.

Difference values and frame position errors may be determined from a plurality of columns and one frame position error is selected from a plurality of detected frame position errors. Erroneous results which may occur when the contents of a column are unsuitable for detecting a frame position error can thereby be avoided. In the selection of the frame position error, frame position errors which strongly deviate from the majority of occurring values may be disregarded. Values which are within a predetermined tolerance field may be utilised for forming an average value.

To avoid misinterpretation of vertical motion as frame position errors, in a further embodiment a characteristic signal is derived which comprises information about the reliability of the detected frame position error being actually a frame position error. Preferably, the characteristic signal is derived by comparing consecutively detected frame position errors in such a way that in the case of small differences between consecutively detected frame position errors the characteristic signal assumes a value indicating a vertical motion but not a frame position error.

In a still further embodiment of the invention a scene transition signal is derived when the comparison values for all offsets performed from one frame to the next are above a predetermined threshold value in the case of an addition of differences, or below said threshold value - in the case of a cross-correlation.

The invention further provides apparatus for measuring frame position errors in film scanning comprising means for selecting the pixels of one column in a first frame and comparing them with pixels in a corresponding column of a time adjacent frame, means for vertically offsetting the pixels of the time adjacent frame a given number of times and comparing each time the offset pixels with the pixels of the first frame, and means for deriving a 4 frame position error signal from the results of the comparisons.

The invention still further provides apparatus for correcting frame position errors in film scanning comprising apparatus for measuring frame position errors as set forth in the preceding paragraph and a correction circuit comprising a first input for receiving an uncorrected video signal, a second input for receiving the frame position error signal, means for modifying the video signal in response to the error signal to create a corrected video signal, and an output from which the corrected video signal becomes available.

Embodiments of the invention will now be described by way of example, with reference to the accompanying drawings, in which:- Figure 1 shows a first embodiment in the form of a block diagram of an arrangement for performing a method according to the invention, Figure 2 is a block diagram of a second arrangement for performing a method according to the invention, Figure 3 is a diagrammatic representation of a television frame, Figure 4 shows various time diagrams to explain a method according to the invention, Figure 5 shows how the frame position error signal is formed from several comparison values, Figure 6 shows representations of comparison values at different conditions, Figure 7 is a block diagram of a further embodiment of apparatus according to the invention using a digital signal processor, Figure 8 is a block diagram of apparatus for correcting frame position errors and, Figure 9 shows an example of a correction filter for use in the apparatus shown in Figure 8.

Equivalent components have been given identical reference numerals in the different Figures. The embodiments are based on a measurement of frame position errors in the luminance signal and the result of the measurement is used for correcting the entire picture information. However the measurement of frame position errors in the chrominance signal could be carried out and the result used for correcting either the chrominance component or the entire picture information and such a method and apparatus is intended to be within the scope of the-claims.

In the arrangement according to Figure 1 the luminance signal Y is applied via an input 1 to latch 3 which is connected to a delay device 2 which delays the luminance signal Y by one frame period plus three line periods. By clocking the latch 3 with a signal H1, which occurs once in each line and has the duration of a pixel period, one pixel is applied to the point 5 from the signal Y in each line and another pixel which has been delayed by the delay device 2 is applied to the point 6. The pixel of the undelayed signal Y is then delayed by two line periods in each of the delay devices 7, 8, 9, to give pixel outputs delayed by two, four, and six line periods respectively.

There are four computing units 10, 11, 12, 13 which perform a crosscorrelation function between the pixel of the delayed luminance signal Ydel and the pixels of the undelayed luminance signal Y, the luminance signal delayed by two line periods, the luminance signal delayed by four line periods and the luminance signal delayed by six line periods. The cross-correlation produces comparison values VK1 to VK4 which are formed as the sum from all products Y and Ydel. The offset at which the correlation is largest is determined at 14 and frame position error is derived therefrom and produced at output 15. Details of the computation of the frame position error F will be described hereinafter with reference to Figures 3 and 4.

In the embodiment shown in Figure 2 an addition of the absolute values of the differences of the compatible pixels is performed in computing units 21, 22, 23, 24, instead of the cross-correlation operation performed in the embodiment of Figure 1. The comparison values VDI to VD4 thus produced are applied to a computing unit 25. The comparison signal at which the sum of 6 the absolute differences is smallest is detected by the computing unit 25 and represents the frame position error F which is produced at the output 26.

The devices shown in block diagrams in Figures I and 2 may be realised in many ways. For example, the formation of the comparison values and their further processing may be advantageously performed by a digital signal processor and the applied luminance signal may be delayed by suitable addressing of a memory.

When due to the required high processing speed, it is desirable or advantageous to have the control performed by dedicated hardware instead of by a programmed processor, a counter, from the start, counts down pixels of each line to the value provided for each column to be evaluated. Such a counter may be easily switched, so that a variable selection of the columns can be realised in a simple manner.

Figure 3 illustrates the selection of a column for the method according to the invention with reference to a television frame which is shown diagrammatically. The width b of the column represents one pixel.

The diagrams a to d in Figure 4 show the variation of the amplitudes A of the pixels of a column for an assumed picture content. Diagram a shows the contents of a column in the n7-th frame (7del), while the diagrams b to d show the n+1th frame (Y) produced by different offsets by the delay devices 7, 8, 9 (Figures 1 and 2). The dotted line represents the amplitude variation in the nth frame. In this example the surface difference between the n+1th and the nth frame in the diagram c is apparently smallest. The frame position error is thus within the range of this offset.

In connection with a minimal number of computations, the offset is realised in steps of two line periods each in the embodiments shown in Figures 1 and 2. However, the frame position error F can be detected with greater precision by an interpolation of the comparison values.

7 The selection of an approximation formula for determining the minimum when the differences of comparable pixels is found (or the maximum when applying the cross-correlation) is based on the assessment of the relation between frame position error and the result of the subtraction (or cross-correlation) for a given offset. It is assumed that the exact frame position error is in the proximity of the minimum difference and that the difference rises each time to the next value. The function describing the relation between offset and difference signal thus has a local minimum if the offset is equal to the frame position error and rises to both sides in a range of 1.5 lines.

The simplest function with which such a variation can be described is a parabola. As three values (the smallest difference and the two proximate values) are available for he approximation, a parabola may be used for determining the frame position error. However, any other appropriate function could be used. For example, the proportionality between frame position error and difference signal may be utilised, at which the relation between the comparison value and the offset in the proximity of the frame position error is linear. The associated approximation formula results from simple geometrical considerations.

Figure 5 shows an example for an interpolation in which the comparison values VD1, VD2 and VD3 were determined every two line periods as the sum of the absolute differences. In the computing unit 25 (Figure 2) a parabola is formed passing through these comparison values and the minimum is determined. The offset of this minimum constitutes the frame position error F and may or may not be an integral number of field periods.

The diagrams in Figure 6 show different comparison values dependent on the respective picture contents as a function of the offset V. A threshold value VDS is each time provided. In diagram a, one of the comparison values VD has a clear minimum.

As the greater part of the comparison values is below the threshold VDS, it can be concluded that a measurable frame 8 position error is present.

In diagram b, all comparison values are approximately equal. As there is no distinct minimum, it cannot be concluded that there is a frame position error. In diagram c, the comparison values are also similar so that no frame position error is detected, but all values are above the threshold value VDS. It can be concluded that there is a scene transition. A corresponding signal can be derived which is used for different tasks in film scanning. An example of application of such a signal is a colour correction which is preprogrammed for each scene.

In diagrams d and e, the comparison value having the largest offset is the smallest comparison value. As there is no information about the further variation of the comparison values, it cannot be concluded that there is a minimum. The frame position error is then outside the measuring range.

Figure 7 shows an embodiment of the arrangement of Figures 1 or 2 using a digital signal processor 31 and two write/read memories 32,33. The writelread memories 32,33 are formed as dual port RAMs so that the write and read procedures can be effected substantially independently of each other. A latch 35 succeeding the input 34 is clocked with the signal H' already described with reference to Figures 1 and 2 so that the amplitude of a pixel is present for one line period at the output of the latch 35. The pixels of each column are alternately written into the write/read memory 32,33 so that the digital signal processor 31 has access to the pixels of each column of two successive frames. An address logic 36 generates the addresses during writing, while the addresses to be read are determined by the digital signal processor 31 as a function of the computation requirements (cross-correlation, sum of the absolute differences).

Figure 8 shows a block diagram of an arrangement for correcting frame position errors, comprising an arrangement for recognising and measuring frame position errors and a correction filter 42. Moreover, a picture memory 43 is provided for 9 delaying the luminance and chrominance signals (Y,C) applied at 44,45 by one frame. This is necessary because a frame position error correction can only be performed if the contents of two frames have been compared. The arrangement 41 may, for example, be as described with reference to Figures 1 and 2 with the picture memory 43 performing the function as the delay unit 2.

The corrected signals Y' and Cl can be derived from the outputs 46,47.

In the embodiments it is assumed that the applied video signals are already present in a digital form. If this is not the case, an A/D conversion is required in advance. If the input signals are present as frames using a progressive scanning, the picture memory 43 may be arranged as a FIFO memory. However, if the input signals are arranged in accordance with the interlace scanning method, a correspondingly more complicated memory control is required. Since the frames are offset with respect to each other by an odd number of lines, the fields must be compared: the first field of the second frame must be compared with the second field of the first frame, and conversely. Also in this modification a frame memory is required with the addition of a second read control. The second read control is required to perform the correction in the frame. In this case the values of locally proximate lines should always be made available for the next correction filter.

A simplification which is, however, at the expense of the picture quality, is the halving of the memory location. In that case fewer than half the column contents can be utilised for correcting the frame position error. The frame position error correction could then only be performed in the field instead of in the frame.

Figure 9 shows a correction filter for one of the signals Y, C. The input 51 of the correction filter is connected to a series of line delay devices 52 to 56. A multiplication by filter coefficients ao to a5 is realised in multipliers 57 to 62.The output signals of the multipliers 57 to 62 are added in the adders 63 to 67 to produce the corrected signal Y' which can be taken from output 68. The filter shown in Figure 9 is a vertical interpolator which generates a new frame, which is offset by the frame position error, from the picture contents. The coefficients ao to aS which control the interpolation are dependent on the frame position error F. As the frame position error can only be measured in steps corresponding to a multiple of a fraction of a line, the picture is offset accordingly by means of the filter shown in Figure 9. A predetermined coefficient series is required for each relevant multiple, which series is applied to the filter from a memory which is not shown.

1 11

Claims (29)

CLAIM(S)

1. A method of measuring frame position errors in film scanning comprising the steps of comparing the pixels of one column of a frame with pixels of the same column of a time adjacent frame, vertically offsetting the pixels of the time adjacent frame a given number of times and repeating the comparison each time, and deriving a frame position error signal from the results of the comparisons.

2. A method as claimed in Claim I comprising using the vertical offset value whose comparison value assumes an extreme value with respect to the other comparison values as the frame position error value.

3. A method as claimed in Claim 1, in which an interpolation between the extreme value and at least one proximate value is performed to determine the frame position error.

4. A method as claimed in any of Claims I to 3 in which the comparison values are determined by means of cross-correlation and in that the offset at which the comparison value assumes a maximum is defined as the frame position error.

5. A method as claimed in any of Claims I to 3, in which the comparison values are determined by addition of differences of the pixels to be compared and the offset at which the comparison value assumes a minimum is defined as the frame position error.

6. A method as claimed in any preceding claim, in which the location of the column is adjustable.

7. A method as claimed in Claim 6, in which the location of the column is preprogrammable and, when a film is being scanned, is adjusted in conformity with the preprogramming operation and the motion of the film.

8. A method as claimed in any preceding claim in which comparison values and frame position errors are determined from a plurality of columns and in that one frame position error is selected from the plurality of detected frame position errors in 1 12 that frame.

9. A method as claimed in any preceding claim in which a characteristic signal is derived which comprises information about the reliability of the detected frame position error being actually a frame position error.

10. A method as claimed in Claim 9, in which the characteristic signal is derived by comparing consecutively detected frame position errors in such a way that in the case of small differences between consecutively detected frame position errors the characteristic signal assumes a value indicating a vertical motion instead of a frame position error.

11. A method as claimed in any preceding claim in which a scene transition signalis derived when the comparison values for all offsets performed from one frame to the next are above a predetermined threshold value in the case of an addition of differences, or below said threshold value in the case of a cross-correlation.

12. A method of measuring frame position errors in film scanning substantially as described herein with reference to the accompanying drawings.

13. Apparatus for measuring frame position errors in film scanning comprising means for selecting the pixels of one column in a first frame and comparing them with pixels in a corresponding column of a time adjacent frame, means for vertically offsetting the pixels of the time adjacent frame a given number of times and comparing each time the offset pixels with the pixels of the first frame, and means for deriving a frame position error signal from the results of the comparisons.

14. Apparatus as claimed in Claim 13 in which the frame position error signal is the vertical offset value whose comparison value assumes an extreme value with respect to the other comparison values.

15. Apparatus as claimed in Claim 13, comprising means for interpolating between the extreme value and at least one proximate comparison value, the frame position error signal being 13 the interpolated value.

16. Apparatus as claimed in any of Claims 13 to 15, in which the comparison values are determined by means of cross-correlation and the offset at which the comparison value assumes a maximum is defined as the frame position error.

17. Apparatus as claimed in any of Claims 13 to 15 in which the comparison values are determined by addition of differences of the pixels to be compared and the offset at which the comparison value assumes a minimum is defined as the frame position error.

18. Apparatus as claimed in any of Claims 13 to 17 comprising means for adjusting the location of the column.

19. Apparatus as claimed in Claim 18, in which the location of the column is preprogrammable and, when a film is being scanned, is adjusted in conformity with the preprogramming operation and the motion of the film.

20. Apparatus as claimed in any of Claims 13 to 19 in which difference values and frame position errors are determined from a plurality of columns and one frame position error is selected from a plurality of detected frame position errors.

21. Apparatus as claimed in any of Claims 13 to 20, in which a characteristic signal is derived which comprises information about the reliability of the detected frame position error being actually a frame position error.

22. Apparatus as claimed in Claim 21, in which the characteristic signal is derived by comparing consecutively detected frame position errors in such a way that in the case of small differences between consecutively detected frame position errors the charactistic signal assumes a value indicating a vertical motion instead of a frame position error.

23. Apparatus as claimed in any of Claims 13 to 22 in which a scene transition signal is derived when the comparison values for all offsets performed from one frame to the next are above a predetermined threshold value in the case of an addition of differences, or below said threshold value in the case of a cross-correlation.

24. Apparatus for measuring frame position errors in film scanning substantially as described herein with reference to Figures 1 and 3 to 6 or to Figures 2 to 6 or to Figures 3 to 7 of the accompanying drawings.

25. Apparatus for correcting frame position errors in film scanning comprising apparatus for measuring frame position errors as claimed in any of Claims 13 to 24 and a correction circuit comprising a first input for receiving an uncorrected video signal, a second input for receiving the frame position error signal, means for modifying the video signal in response to the error signal to create a corrected video signal, and an output from which the corrected video signal becomes available.

26. Apparatus as claimed in Claim 25 in which the modifying means comprises a correction filter.

27. Apparatus as claimed in Claim 26 in which the correction filter comprises a tapped delay line, the input, output, and each tap being connected to a first input of a respective multiplier, a summing arrangement for summing the outputs of the multipliers, and means for applying the frame position error signal to second inputs of the multipliers, the output of the summing means providing the corrected video signal.

28. Apparatus for correcting frame position errors in film scanning substantially as described herein with reference to Figure 8 or to Figures 8 and 9 of the accompanying drawings.

29. Any novel feature or novel combination of features disclosed herein either explicitly or implicitly whether or not it relates to the same invention as that claimed in any preceding claim.