GB2312116A

GB2312116A - Removal of redundant fields in standards conversion

Info

Publication number: GB2312116A
Application number: GB9607352A
Authority: GB
Inventors: Alois Martin Bock
Original assignee: Digi Media Vision Ltd
Current assignee: Digi Media Vision Ltd
Priority date: 1996-04-09
Filing date: 1996-04-09
Publication date: 1997-10-15
Also published as: GB9607352D0

Abstract

A video signal converted from a first lower frame rate, such as 24 frames per second to a higher frame rate such as 30 frames per second using the 3:2 pulldown conversion technique includes repeated redundant video fields in some of the frames. This invention is concerned with reversing the conversion process and relates specifically to the removal of these redundant fields. It consists of generating difference signals representing differences between each frame of the video signal and the next in succession. The difference signals are accumulated over a field period (there being two video fields per frame) and the frame difference signals are distributed in rotation to 5 averaging means. The averaged signals constitute a pattern in which repeated fields are indicated by a low average signal. The pattern is detected by a logic circuit which blocks the redundant video fields from the video signal.

Description

Improvements in or relating to the orocessina of video signals.

The present invention relates to a method and apparatus for converting an input video signal, having frames which occur at a first frame rate and include video fields some of which are repeated, into an output video signal having a lower frame rate than the input video signal and from which the repeated fields have been removed.

The frame rate of motion picture films is 24 frames per second. To convert motion picture film images into video image signals film scanning systems are used which scan the film frames sequentially to generate film frame image signals. These film frame image signals are then converted into a standard video frame rate such as the standard 30 frame per second rate used in a 525 line 30 Hz video signal. The technique for the conversion from the film frame rate of 24 frames per second to a video frame rate of 30 frames per second is known as 3:2 pulldown. In this technique, the successive film frames are converted alternately into two and three video fields.

Where three video fields are generated from a film frame, one of the fields is a repeat of another of the three fields. The pulldown sequence therefore repeats every four film frames to produce five video frames. Each video frame includes two video fields but in the sequence of five video frames there are two repeats of video fields.

There are circumstances in which it would be advantageous to reverse the pull-down operation. One such case occurs when the video signal is to be digitally encoded in compressed format. Since the 3:2 pulldown operation repeats two video fields in every 5 frames, the video signal contains 20% redundant information. By eliminating the repeated fields and encoding only 24 video frames per second, transmission bandwidth can be saved.

The present invention provides for the elimination of redundant video fields from a video signal. According to the present invention there is provided a method of converting an input video signal, having frames which occur at a first frame rate and include video fields some of which are repeated, into an output video signal having a lower frame rate than the input video signal and from which the repeated fields have been removed, the method comprising the steps of; generating a difference signal representing differences between each frame of the input signal and the next in succession, accumulating each difference signal over a pre-set portion of each frame period, distributing the accumulated frame difference signals in rotation to a predetermined plurality of averaging means to generate a pattern of averaged signals indicative of the repeated video fields, and removing those video fields from the input video signal indicated by the said pattern of averaged signals.

This has the advantage that the repeated video fields can be reliably identified in the input video signal through the averaging process even when there is little or no movement in the video sequence. The removal of the repeated fields from the input video signal is therefore accomplished without the loss of field information.

Further, according to the present invention there is provided apparatus for converting an input video signal, having frames which occur at a first frame rate and include video fields some of which are repeated, into an output video signal having a lower frame rate than the input video signal and from which the repeated fields have been removed, the apparatus comprising; means to generate a difference signal representing differences between each frame of the input signal and the next in succession, means to accumulate each difference signal over a pre-set portion of each frame period, means to distribute the accumulated frame difference signals in rotation to a predetermined plurality of averaging means to generate a pattern of averaged signals indicative of the repeated video fields and means to remove from the input video signal the repeated video fields indicated by the said pattern of averaged signals.

The invention will now be described, by way of example, with reference to the accompanying drawings in which: Figure 1 is a schematic diagram to explain the 3:2 pulldown technique; and Figure 2 shows a block diagram of a system embodying the present invention to remove repeated video fields from a video signal.

Referring now to Figure 1, there is shown a sequence 20 of film frames which occur at a frame rate of 24 frames per second. The film frames are scanned in sequence to generate a sequence of video frames 21 each of which includes two video fields as indicated schematically by the dotted lines dividing the video frames. The frame rate of the video frames is 30 frames per second. In order to provide each of the video frames with two video fields, the successive film frames are converted alternately to two video fields and three video fields (3:2 pulldown process) as indicated schematically by the arrows 22. As a result, it will be observed that a pulldown sequence of 4 film frames is converted to 5 video frames and in those 5 video frames there are two repeats of video fields (one top field and one bottom field). It will be apparent that the resulting video frame sequence includes 20% redundant information.

Referring now to Figure 2, there is shown in block diagram form, a system to convert an input video signal at a frame rate of 30 frames per second to an output video signal at a frame rate of 24 frames per second by the removal of redundant video fields such as are generated by the 3:2 pulldown technique illustrated in Figure 1.

There are two fundamental problems in reliably detecting that a 3:2 pulldown technique has been used to generate a video signal. Firstly, in a video sequence which contains little or no movement, successive fields of the same parity are very similar. Secondly, repeated fields may not be completely identical due to noise or small amounts of jitter in the scanning process.

In the system of Figure 2, the input video signal at 30 frames per second is applied to an input terminal 23 and applied to a spatial low-pass filter 24 where the signal is subject to spatial low-pass filtering in both the horizontal and the vertical direction. This reduces the effect of small amounts of jitter which have been introduced by the scanning process. The signal is then passed to a noise coring block 25 which eliminates signal differences less than a predetermined value (thus quantising and rounding the signal) in order to minimise the effects of signal noise in the video signal.

The cored difference values are then applied to a frame store 26 where the absolute values of the differences in the signal are derived by a difference circuit 27 and the absolute difference signal so generated is accumulated over an entire field period in a field accumulator 28.

A frame switch 28 distributes the accumulated field signals in rotation to 5 averaging circuits 30 which each generate a sliding average corresponding to the five frame positions in a 3:2 pulldown sequence as represented in Figure 1. The sliding average in each circuit 30 corresponds to one field only in each frame (top or bottom) and therefore by comparing the sliding averages with one another in decision logic 31 it is possible to determine whether the input video signal at the terminal 23 has been generated by a 3:2 pulldown process and to identify the relative position of the repeated fields in the five frames of each sequence. Thus if one of the five sliding average values is significantly (e.g. has less than 50%) lower than the four others, this value will represent a repeated field. In other words, the decision logic 31 is effective to detect a pattern of averaged signals indicative of the repeated video fields. The decision logic 31 produces a gating signal applied to a field store 32. The field store 32 is operated to remove the repeated fields from the input video signal applied to the store 32 by way of a compensating delay 33. The output video signal at 24 frames per second is transmitted to an output terminal 34 from the field store 32.

It will be apparent that the field accumulator 28, frame switch 29 and averaging circuits 30 may be operated to generate a pattern representing either the top or the bottom repeated field in each sequence of 5 frames in the input video signal. The decision logic 31 would be arranged to detect the appropriate video field pattern being generated. In the aiternative, a pattern of average values could be generated which represents both the top and the bottom repeated field in each sequence of 5 frames in the input video signal.

A further application of the invention lies in the conversion of an input video signal of 25 frames/second to an output video signal of 24 frames per second, assuming that the input signal was originally from a 24 framelsecond film. In the sequence of 25 video frames/second there is a repeat of one top video field and a repeat of one bottom video field . the need is therefore to detect the 2 repeated video fields in each 25 input frames. this detection can be achieved according to this invention by adapting the circuit of Figure 2 to include 25 averaging circuits instead of 5 and enabling the decision logic circuit 31 to detect the pattern of repeated video fields in 25 video frames.

Claims

1. A method of converting an input video signal, having frames which occur at a first frame rate and include video fields some of which are repeated, into an output video signal having a lower frame rate than the input video signal and from which the repeated fields have been removed, the method comprising the steps of; generating a difference signal representing differences between each frame of the input signal and the next in succession; accumulating each difference signal over a pre-set portion of each frame period, distributing the accumulated frame difference signals in rotation to a predetermined plurality of averaging means to generate a pattern of difference signals indicative of the repeated video fields; and removing those video fields from the input video signal indicated by the said pattern of averaged signals.

2. A method according to claim 1, wherein the input video signal has been generated by a 3:2 pulldown technique to produce 2 repeated video fields for each sequence of 5 frames.

3. A method according to claim 2, wherein each difference signal is accumulated over one field period.

4. A method according to claim 2 or 3, wherein the said averaging means are 5 in number and generate a pattern of 5 averaged signals.

5. A method according to any preceding claim, wherein the input video signal is subjected to filtering and noise coring to remove jitter and noise.

6. Apparatus for converting an input video signal, having frames which occur at a first frame rate and include video fields some of which are repeated, into an output video signal having a lower frame rate than the video frame and from which the repeated fields have been removed, the apparatus comprising; means to generate a difference signal representing differences between each frame of the input signal and the next in succession, means to accumulate each difference signal over a pre-set portion of each frame period, means to distribute the accumulated frame difference signals in rotation to a predetermined plurality of averaging means to generate a pattern of averaged signals indicative of the repeated video fields, and means to remove from the input video signal the repeated vodeo fields indicated by the said pattern of averaged signals.

7. Apparatus as claimed in claim 6, which is adapted to convert an input video signal which has been generated by a 3:2 pulidown technique to produce 2 repeated video fields for each sequence of 5 frames.

8. Apparatus as claimed in claim 7, wherein the accumulating means accumulates each difference signal over one field period.

9. Apparatus as claimed in claim 6 or 7, wherein the said averaging means are 5 in number and generate a pattern of 5 averaged signals.

10. Apparatus as claimed in any of claims 6 to 9, wherein filter and noise coring means are provided to remove jitter and noise from the input video signal.