GB2351629A

GB2351629A - Video processing apparatus for showing encoder effects

Info

Publication number: GB2351629A
Application number: GB9909778A
Authority: GB
Inventors: Matthew Beard
Original assignee: Snell and Wilcox Ltd
Current assignee: Snell Advanced Media Ltd
Priority date: 1999-04-28
Filing date: 1999-04-28
Publication date: 2001-01-03
Also published as: AU3981200A; WO2000067490A1; GB9909778D0

Abstract

To assist an operator in optimising the processing in channels which each receive a respective frequency range of an input video signal, a display is provided in which the separate encoding effects of each channel can be observed in separate display regions.

Description

2351629 timisation of asljfleo Processor This invention concerns the

subjective optimisation of a video process in which ' the signal is decomposed into a plurality of procesing: channels and then reassembled.

0 ln a video process such as noise reduction lt is often advantageous to decompose the signal into a number of signal components, such as frequency sub-bands or wavelets, which are separately processed and then combined to give a processed output signalSuch processes are ftequently optimised by an operator who views the output signal on a display device and adjusts the processing parameters in the various channels to improve the quality of the process. It may, however, be difficult for the 0Derator to optirnise the setting of a particular parameter by observing the reassembled output Agnal.

The invention -seeks to overcome this difficulty by providing means by which the individual processing effects of at least two, and preferably all, of the processing channels are displayed to the operator simultaneously on the same video display for the purpose of optimising the individital contributions of each of the displayed channels to the overall process.

Preferably the operator is presented with a processed picture display in which a number of separate, defined areas of the picture are processed by different processing channels and only one processing channeJ is active in each displayed area.

in a further aspect of the invention each displayed area shows the processed output from its respective channel alone, wii bout any cntribution from other channels.

In another mode of operation each displayed area shows the difference between the input and the output of its respective processing channeL In a yet further aspect the video signal is modified during the filter optirnisation by a picture content replication device so that the content of at least one of the defined picture areas is replicated andreplaces the original content of at least one other of the defined picitite areas., and the replication does not alter the scale or orientation of the picture content.

An example of the invention will nosy be described with reference to the drawings in which:

Figure 1 shows the block diagram of a video process in accordance with the invention.

Figure 2 shows a displayed picture divided into 9 areas corresponding to 9 spatial flequency bands.

Figures 3 and 4 show examples of displayed pictures in accordance with the TVst Tuning mode of the invention.

Figure 5 shows an example of a displayed picture in accordance with the Tbird Tiining mode of the invention.

Referring to Figure 1, an input video signal (1) is fed to a sync separator (2), a decomposing filter (3), and a subtractor (4). The sync separator extracts horizontal and vertical timing information and relates this to a number of defined areas within the picture. For example if the active line and field periods are each divided into three equal tune periods, nine rectangular picture areas are be identified. The area to which the. current video data (in a digital system) or signal value (in an analogue system) corresponds is identified and output as an area address signal which is passed to a filter control block (5).

The decomposing filter (3) breaks down tbe video signal into several parallel component signals %vhich may be added together to give the Original signal. For example the spectrum of the signal could be split into a number of frequency subbands on the basis of spatial frequency or signal frNuency. Each of the components (nine in the example shown in the figure) is fed to one of a set of processing blocks (6), each of which carries out some processing on its component and outputs a processed component to a summing device (7).

4:1 Bach processing block can operate in one of three modes according to a control signal from (be filter control block (5). The modes are:

a) Process the component. b) Pass the component unprocessed. c) Give no output.

In normal operation all processing blocks are in mode (a) and the output of the,suiliming device (7) passes through the changeover switch (8) to give a processed Output (9).

In addition to this normal mode one of three ru-ning modes may be selected in 1-csponse to an appropriate control signal to the filter controller (5).

in Lhe 1-116L 11111Mg mode only one processing cbannel is in mode (a) at any one time and all other channels are in mode (c). The filter control block (5) puts the channel corresponding to the current area address into mode (a) and sets all the others to mode (c). When the processed output signal (9) is displayed, the individual channels are 0 seen acting alone in their designated picture areas without any contribution from other channels.

in the second tuning mode one channel is in niode (a) at any one time and all the others are in niodt (b). This is achieved by the filter control block (5) setting all the channels Dot corresponding to the current area address into mode (b). Here the displayed picture always shows the full spectrum but only one component is processed in each area.

In the. third tuning mode only one processing channel is in mode (a) at any one time, the other channels are in mode (b) and the changeover switch (8) is set to route the output of the, subtractor (4) to the output (9). The subtractor (which includes a compensating delay to bring its two inputs into time coincidence prior to the subtraction) fornis the difference between the unfiltered input to the system and the filtefed output from the surnming device (7). here each area of the displayed picture shows what is removed by the processing channel corresponding to that area.

in the case of a noise reduction process the operator would adjust the processing parameters in the second tuning mode to removo noise but avoid loss of picture detail. In the [bird mode the operator would adjnst the processing parameters to maximise tl)C allIOLInt of noise shown in each area without allowing any picture detail to show.

2 Figures 2 - 5 show examples of what the operator would see when optimising a noise reduction process carried out on a hypothetical video signal containing a broad spectrum of horizontal and vertical picture detail evenly distributed over the whole picture area. In these examples the signal is decomposed into nine bands on the basis of horizontal ind vertical spatial frequency.

Figure 2 shows how the frequency bands could be allocated to different picture areas. In the Figure the frequency components are represented by grids of lines for siMplicity, though in practice, sbitisoidal variations in luminance would be seen. It can be seen that the spatial frequency bands are arranged so that the higher horizontal frequency bands are on the left of the picture and the higher vertical frequency bands are at the top.

M the input signal were; degraded by random noise, but the filter blocks (9) failed to remove the noise, the display in the first tuning mode would be as shown in Figure 3. (Again, for simplicity, the noise is represented by grey dots whereas in practice random-amplitude, sinusoidal spatial frequency components would be seen.) If the operator successfully optimised all the filter blocks apart flom two of the low horizontal frequency bands, the display would be as shown in Figure 4. No noise is visible in the correctly optin-dsed areas, only wanted picture detailHowever, the top ri,.:,,ht hand block shows insufficient noise reduction, and, the block irnmediately below it shows excessive processing which has not only removed the noise but has also attenuated the wanted detail.

Figure 5 shows this same situation in the third timing mode. The correctly optin-tised areas show [lie input rioise at fi.ill amplitude but no video detail. The block with insufficient noise reduction (top ri- bi) shows some. (reduced-amplitude) noise, and the cl 4;ovei-filicred" block sho.,,vs full amplitude noise together with reduced-amplitude video detail.

A possible disadvantage is that the nature of the material in a particular picture area may not be such as to enable the optirrCisation of the processing channel corresponding to that area, perhaps because there is no material which lies within the passband of the relevant decomposing filter.

This difficulty may be overcome by modifying the video input to the processor during tuning so that pail of ihe picture is repeated (at the same scale and orientation) in the areas corresponding to several (preferably all) processing channels and the operator is c:

able to optirWse more than one channel processing the same picture content.

The replication process can be carried out using the well known technology of digital video effects (DVE) systems in which the video signal is written into, and read out from, a store under the control of read and write. address signals. The sequence of write addresses is derived from the video synchronising information so that a particular address value corresponds to a particular picture element within the video frame. The read address sequence is arranged to repeat the addresses corresponding to part of the frarne so that a part of the picture appears replicated when the output from 1he store is, displayed.

it must be recognised that the invention has ben described only by way of example and many variations are possible within the concepts described.

3

Claims

1. Video processing apparatus where the signal is decomposed into a plurality of components which are processed in separate channels and recombined to form a processed output, characterised in that means are providcd by which the individual processing effects of at least two, of the processing channels can be displayed to the operator simultaneously on the same video display for the purpose of optimising the individual contributions of each of the displayed channels to the overall process.

2. Video processin apparatin; in accordance with Claim 1 where a processed output sigual may be displayed to show a numberof separate, defined areas of the picture processed by different processing channels and only one processing channel is active in each displayed area.

3. Video processing apparatus in accordance with Claim 2 in which each area shows 0 the processed output from its respective channel alone, without any contribution from other channels.

4. Video processing apparatus in accordance with Claim 2 where each defined area rp of the picture shows the difference between the input and the output of its respective processing channel.

c)

5. Video processing apparatus in accordance with any of the preceding claims where, during the tuning process, the video processor includes picture content replication C:> means by which part of the picture is replicated in the picture areas corresponding to at least two processing channels.

6. Video noise reduction apparatus in accordance with any of the preceding claims.

4