CN1294821A

CN1294821A - Signal processing appts. and method

Info

Publication number: CN1294821A
Application number: CN99804422A
Authority: CN
Inventors: J·－C·波斯; V·赫林
Original assignee: Koninklijke Philips Electronics NV
Current assignee: Koninklijke Philips NV
Priority date: 1998-11-27
Filing date: 1999-11-16
Publication date: 2001-05-09
Anticipated expiration: 2019-11-16
Also published as: EP1057345A1; CN1178519C; WO2000033582A1; KR20010034437A; JP2002532028A

Abstract

A signal-processing arrangement comprising an examining circuit (EXAM), and adjustable filter (FIL), and a signal processor (PRC). The signal processor (PRC) may be, for example, a video encoder for encoding a sequence of pictures in accordance with an MPEG standard. The examining circuit (EXAM) examines a signal (S) to be processed so as to obtain a distortion indication (IND) the distortion indication (IND) indicates to which extent distortion (DIST) would be introduced if the signal (S) were processed (PRC[S]) by means of the processor (PRC). An adjustable filter (FIL) filters the signal (S) in dependence on the distrotion indication (IND) so as to obtain a filtered signal (SF). The signal processor (PRC) processes the filtered signal (SF). Thus, the signal-processing arrangement filters the signal to be processed in a pro-active manner so as to counter distortion which might otherwise be introduced by processing the signal. Accordingly, a satisfactory signal quality can be obtained. For example, in a video-encoding application, the filter can reduce details contained in a series of pictures. This allows the series of pictures to be coded with a sufficient precision without introducing block effects which might otherwise occur if the pictures were not filtered. Moreover, since the filter is pro-actively adjusted, it filters the series of pictures more evenly than if the filter were retro-actively adjusted. Accordingly, there will be relatively little variation in resolution from one picture to another, which contributes to a satisfactory overall picture quality.

Description

Signal processing apparatus and method

Invention field

The present invention relates to signal processing technology, for example encode according to the image sequence of Motion Picture Experts Group (MPEG) definition standard.

Background technology

Patent documentation JP-A-09,009,260 has described a moving picture encoding device.Its literal that is attended by english abstract is quoted as proof.View data for the AD conversion, in variable low pass spatial filter, do the filtering of low frequency component according to the 2-D data module unit, carrying out high-frequency coding in compressor reducer 13 handles, temporarily be stored in the FIFO memory, the packed data memory space in the FIFO memory is monitored by the FIFO supervising device.The FIFO stored information is from the monitoring data that is used for each data block unit of compressor reducer and data block multidate information, this information is input to a band controller, at each data block cut-off frequency is set, and outputs to variable low pass spatial filter from band controller.

Summary of the present invention

An object of the present invention is to provide and realize the more signal processing of good quality.

The present invention takes following several aspect.Signal Processing can cause distortion.For example, image sequence of coding is exactly a kind of signal processing form that can cause distortion.In order to be no more than a definite boundary that obtains amount of coded data, may be with some image of quite low precision encoding.If image is with quite low precision encoding, then image may be in distortion sensuously, promptly data block effect in the image occurs in decoding end.

Might offset by handling the distortion that signal causes by the signal filtering before handling.Yet, if a signal is filtered, also can distortion be arranged sensuously, some signal component is attenuated even is deleted.Therefore, if signal is filtered before handling, on the effect be do between distortion that is causing by filtering and the distortion that causes by processing compromise.For example, if an image is filtered, then the details that comprises of the filtering image that therefore obtains is lacked than original image.The result is, the image of filtering can be with the precision encoding higher than original image, sees some restriction of existence from the angle of the amount of coded data that obtains simultaneously.Yet the definition that filtering image has is lower than original image.Therefore, image of filtering must guarantee that one between encoding precision and image definition is traded off before coding, because the both influences the gross mass of image.

In background technology, the spatial low-pass filter of filtered image data is used as a function of the amount of coded data that comprises at the FIFO output state and is adjusted before image encoding.Therefore, do not prevent to produce at first quite a large amount of coded datas.If really produced quite a large amount of coded datas, then will the image encoded data will be by distortion on a sizable degree.This is because view data or filtering or must be with quite low precision encoding to a considerable extent, and perhaps the two is taken into account, in order to avoid make the amount of coded data of acquisition surpass a boundary.Therefore, will the image encoded data will have a quality of being on duty mutually, perhaps because definition is low, perhaps because wherein some data block effect, or because these two.

According to the present invention, a checking circuit check will be by the signal of a signal processor processes, so that obtain a distortion indication.This distortion indication show if this signal by this signal processor processes with the distortion level of introducing.A tunable filter is according to indicating this signal of filtering according to this distortion, so that obtain the signal of a filtering.The signal of this this filtering of signal processor processes.

Therefore, this tunable filter is adjusted in a forward action mode in the present invention, and the spatial low-pass filter in background technology is to adjust in the acting in opposition mode.Therefore, in the present invention by handle caused distortion and make between the caused distortion by filtering compromisely do not have essence to postpone, and in background technology, wherein have this some compromise delay inherently.Therefore, of being implemented between the distortion of two above-mentioned types of the present invention is better compromise.The result is that the present invention realizes a better quality.

For example, the present invention can be applied to the coding of an image sequence.Check one with image encoded or its part if suppose this checking circuit.Thus, can show whether will be obtained quite a large amount of coded datas if this image so encodes.As above illustrated, this may cause the distortion of one or more image that will be encoded subsequently.Owing to the invention enables this image will be according to this indication and filtered, in fact produce a large amount of coded datas so can prevent the coding of this image.Therefore, this will help the quality improvement of one or more subsequent picture.Therefore in force, the present invention is implemented in the image of inspection and the quality reciprocity between one or more subsequent picture.Therefore, the present invention realize a satisfaction with the uniform images quality.

The present invention and can selecting is used to realize that the supplementary features of advantage of the present invention will be from following description with reference to accompanying drawing and obvious.

Brief description of the drawings

Fig. 1 is the rendering that illustrates as the essential characteristic of the present invention that requires in the claim 1.

Fig. 2 is the block diagram of the example of a video encoder of explanation;

Fig. 3 is the block diagram that illustrates according to a video coding apparatus of the present invention;

Fig. 4 is a curve chart at the first FILTER TO CONTROL characteristic of video coding apparatus shown in Figure 3.

Fig. 5 is a curve chart at the second FILTER TO CONTROL characteristic of video coding apparatus shown in Figure 3.

The detailed description of accompanying drawing

The at first use of description references mark.Similar means is represented by same alphanumeric codes in whole accompanying drawing.Under the situation of a plurality of similar means shown in the single accompanying drawing, just numeral is added on these alphanumeric codes, so that kindred organization can distinguish each other mutually.If the quantity of similar means is a running parameter, then numeral will be between round parentheses.In description and claim, if can, the Any Digit in the reference number then can be omitted.

With solid line essential characteristic of the present invention is shown among Fig. 1.A signal processing apparatus comprises a checking circuit EXAM, preprocessor FIL and signal processor PRC.The signal S that checking circuit EXAM check will be handled is so that obtain a distortion indication IND.This distortion indication IND shows, if utilize this processor P RC to handle (PRC[S]) this signal S, distortion DIST will be introduced into.Preprocessor FIL handles this signal S according to distortion indication IND, anticipates signal SF so that obtain one.Signal processor PRC handles this pretreated signal SF.

Feature shown in Fig. 1 can be applied to for example coding of an image sequence.In such application, signal S is a data flow of the image sequence that expression will be encoded.This signal processor PRC is a video encoder, and it can make one or more image fault on sizable degree.This will be illustrated in greater detail as follows.

Fig. 2 is the block diagram of the example of a video encoder of explanation.Video encoder provides a mpeg data stream OUT, has the bit rate R that determines of the image data stream of response presentation video sequence IN.This image data stream IN can obtain by for example an analog video signal being transformed into a digital video signal.This video encoder comprises following functional block: input buffer IBUF, subtracter SUB, memory MEM, motion estimation machine and compensator MEC, discrete cosine transformer DCT, anti-phase discrete cosine transformer IDCT, quantizer Q, inverse quantizer IQ, adder ADD, variable length coder VLC, output state OBUF and controller CON.

Video encoder shown in Fig. 2 most image of encoding in the following manner.Input buffer IBUF is temporary will image encoded.In fact, will image encoded be subdivided into the 16 macrodata pieces of taking advantage of 16 pixels.Suppose that memory MEM comprises a decoding scheme of coded image in advance, hereinafter referred to as image formerly.At each the macrodata piece in the image that will be encoded, estimation and compensator MEC in front in the image one similar 16 of search take advantage of 16 macrodata pieces.This estimation and compensator MEC are provided to subtracter SUB to this similar macrodata piece that it has been found that.Subtracter SUB provides the macrodata piece of a difference by this similar macrodata piece that is provided by motion estimation and compensator MEC is provided from the macrodata piece that will encode.This discrete cosine transformer DCT certainly is transformed into the macrodata of this difference the data block of a coefficient of frequency.

Quantizer Q divides each coefficient of frequency by the value that is proportional to a quantization parameter QP, and division result is trimmed to nearest integer.Therefore, quantizer Q provides a data block of sampling frequency coefficient.Inverse quantizer IQ and anti-phase discrete cosine transformer ICDT provide a decoding scheme of the macrodata piece of this difference according to the data block of sampling frequency coefficient.That adder ADD has been found that this estimation and compensator MEC, the similar macrodata piece in the image is added to the decoding scheme of the macrodata piece of this difference in front.Therefore, this adder ADD provides a decoding scheme of the macrodata piece of encoding, and its decoding scheme is stored in this memory MEM.Therefore, in case be encoded with all macrodata pieces in this image, then this memory MEM will comprise the decoding scheme of the current image that just is being encoded.Therefore, when an image is encoded, be implemented in a decoding scheme that comprises the earlier figures picture in this memory MEM in this way.

Variable length coder VLC is transformed into a series of variable length codes to the data block of sampling frequency coefficient.The data block of sampling frequency coefficient generally includes the some coefficient of frequencies with null value, and this number of zeros has higher value along with quantization parameter QP and be higher.The result is, if quantization parameter QP has high value, then the series of this variable length code comprises quite few bit, and vice versa.This variable length code is temporarily stored among this output state OBUF.Output state OBUF exports the variable length code of the macrodata piece that belongs to encoded with bit rate R.

Controller CON is according to following criterion control quantization parameter QP.Being included in amount of coded data among the output state OBUF, to be called the output state degree of filling should be between a minimum and peak level.This means that if this output state OBUF is fuller, then quantization parameter QP will have higher value usually.If this output state OBUF is empty, then quantization parameter QP will have lower value usually.

In more detail, this controller CON calculates the aim parameter of be used for being encoded image or its a part of coded data.The aim parameter of this coded data depends on output state degree of filling F.That is, if this output state OBUF is fuller, then this target will be lower, and vice versa.Controller CON is used for the value of quantization parameter QP according to the aim parameter of coded data and the empirical Calculation of encoding recently.These nearest coding experiences comprise: the value of the quantization parameter QP in data volume that comprises in recent coded image part and the coding that is applied in this part.The coded portion of supposing recent image comprises quite a large amount of coded datas.From then on controller CON infers that this quantization parameter QP should have the value lower slightly than nearest value, so that prevent to surpass the aim parameter of coded data.

If the part of this image or this image is to have the quantization parameter QP coding of quite high value, video encoder then shown in Figure 2 may make image fault arrive sizable degree.If this quantization parameter QP has quite high value, then the difference amount of sampling frequency coefficient may have quite low value.And many coefficient of frequencies in practice will be rounded is null value.The result is that in a decoding end, image is with distortion to a considerable extent.For example, comprise that the target of quite a lot of details or its some parts will utilize sizable, uniform data block to represent.This is called the data block effect below phenomenon.

Blocking effect may appear at for example when the sequence in the image exists scenery to change.Generally speaking, first image of a new scenery will be different from the image of front scenery largely.Therefore, if first image of this new scenery of encoding, many data blocks of coefficient of frequency will comprise the many coefficient of frequencies with essence value.If these data blocks quantize with the quantization parameter QP with nominal value, then obtain quite a large amount of coded datas.For anti-situation here, controller CON will provide the quantization parameter QP of a high value.Explain that as above this can cause blocking effect.

Example according to a video coding apparatus of the present invention shown in Figure 3.This video coding apparatus is recaptured the feature shown in Fig. 1.The view data data flow PDS of its response expression one image sequence provides a mpeg data stream data flow MDS.This video coding apparatus comprises a pseudo-video encoder PENC, filter controller FCON, delay circuit DEL, tunable filter FIL and video encoder ENC.This video encoder ENC and pseudo-video encoder PENC are similar to the video encoder of discussing among above Fig. 2.Yet the controller CON that comprises in this puppet video encoder does not control this quantization parameter QP in the mode that reference Fig. 2 describes.This will be illustrated in greater detail as follows.

Video coding apparatus shown in Fig. 3 is operated in the following manner.This puppet video encoder PENC image of encoding, so as to draw show truly will the coded image degree of difficulty complexity indication CI.Filter controller FCON controls this tunable filter FIL according to this complexity indication CI.Delay circuit DEL compensates this puppet video encoder PENC and encodes the time of this image cost and the time that this filter controller FCON controls this tunable filter FIL cost.The image that tunable filter FIL will be encoded with a cut-off frequency FC filtering of being determined by complexity indication CI.The encode image of this filtering of video encoder ENC.

This puppet video encoder PENC can have variety of way that complexity indication CI is provided.For example, this puppet video encoder can utilize this image of encoding of the quantization parameter QP with a definite value of determining.If the amount of this coded data is quite big, just illustrate that then this image is difficult to coding, vice versa.Therefore, the amount of the coded data that obtains by this image of encoding can constitute this complexity indication CI.This puppet video encoder PENC can also utilize this image of encoding of the quantization parameter QP according to the aim parameter control of the coded data set up in encoder ENC.If this quantization parameter QP is presented a quite high value so that satisfy this target, can say that then this image is difficult to coding, vice versa.Therefore, quantization parameter QP can constitute this complexity indication Cl.Also might use amount parameter QP and the value that obtains of the product of amount of coded data indicate as complexity.In this case, the mode of setting up a value being used for quantization parameter QP with this puppet video encoder PENC has no relations.

In fact, complexity indication Cl is a distortion indication.If complexity indication Cl has high value, mean that then the image that may occur subsequently is difficult to coding.Video encoder ENC can provide the quantization parameter QP of quite high value, so that prevent to produce a quite a large amount of coded data.In this case, such as explained above, blocking effect in the image may appear in decoding end.This video encoder ENC can also give abundant low value of this quantization parameter QP, so that prevent this blocking effect.Yet in this case, such as explained above, will produce quite a large amount of coded datas, influence is the quality of one or more images subsequently.Therefore, if complexity indication CI has high value, this encoder ENC this image of so encoding, then or the distortion of image own, or one or more distortion of subsequent picture, or these two all distortion.

Fig. 4 illustrates can be by a FILTER TO CONTROL characteristic of filter controller FCON application.Fig. 4 is a curve chart, has the trunnion axis of expression complexity indication Cl and the vertical axis of the cut-off frequency PC of this adjustable filter FIL of expression.This curve chart illustrates, if complexity indication Cl is lower than threshold value Cth, then cut-off frequency has quite high value, and it does not indicate the function of Cl in fact as this complexity.Therefore, if an image is encoded quite easily, then it is filtered to relatively little degree, and the details of therefore preserving image in fact.Yet if complexity indication CI is higher than this threshold value, cut-off frequency reduces along with complexity indication Cl increment.Therefore, if image quite is difficult to coding, it will be according to the degree of difficulty of this image encoding and filtered to a degree.That is, if image quite is difficult to coding, its will be according to the degree of difficulty of this image encoding and on a degree details of this image of deletion.This makes this video encoder ENC with the image of filtering like this of encoding of the quantization parameter QP with abundant low value, so that prevent the blocking effect in the image itself, and the amount of coded data that obtains simultaneously is low fully, so that realize the satisfactory quality of subsequent picture.

Fig. 5 illustrates another FILTER TO CONTROL characteristic, and it can cooperate with control characteristic shown in Figure 4 selectively uses.The curve chart of Fig. 5 has a trunnion axis of presentation video number and represents the vertical axis of cut-off frequency FC.As described above, picture number N represents to set up the image of complexity indication Cl.The complexity indication Cl that supposes this image has suitable high value, means that then the image that may occur subsequently is difficult to coding.The result is that when picture number N was filtered, spatial filter FIL had quite low cut-off frequency FClow.The result is that Fig. 5 illustrates the image cut-off frequency filtering to increase gradually along with the picture number increase equally that has numbering N+1, N+2, N+3 subsequently.

When the scenery change occurring, the FILTER TO CONTROL characteristic shown in Fig. 5 is useful.Suppose that picture number N is first image of new scenery.Because first image of new scenery is with quite low cut-off frequency FClow filtering, so many details are with deleted.That is, in decoding end, first image of new scenery will have quite low resolution.Because this cut-off frequency FC little by little increases with the picture number in the new scenery, so this definition will little by little improve along with the image of back.The attribute of the progressive match people vision of this definition.The people trends towards at first noting the basic configuration and the color of fresh target, and just notes the details of fresh target afterwards.Therefore, under the situation that scenery changes, the FILTER TO CONTROL characteristic shown in Fig. 5 is the definition of influence sensation on a relatively little degree only.The more important thing is that the quantization parameter QP that first image of the feasible new scenery that will encode of this control characteristic has is high fully, so that avoid blocking effect, and does not produce quite a large amount of coded datas.Therefore, the control characteristic shown in Fig. 5 helps the gross mass of a satisfaction.

Accompanying drawing that more than illustrates and description are not limitations of the present invention.Obviously, a lot of possibilities is arranged within the scope of appended claims.In this respect, make following summary saying.

Have by various unit and realize the concrete a lot of methods that constitute spread function or function element.In this respect, accompanying drawing is graphic completely, and each only represents a possible embodiment of the present invention.Therefore, though a difference in functionality element that illustrates as different frames, this means that never some function element of eliminating or all function element may be as the possibilities of single actual cell realization.

Though the video coding apparatus shown in Fig. 3 uses a pseudo-video encoder PENC that a distortion indication CI of complicated indication form is provided, and can provide a distortion indication with other circuit.For example, can use be based upon be encoded image and one formerly the absolute difference between the respective pixel of image and circuit.If absolute difference and high, then this image is difficult to coding, so this image possibility distortion itself or one or more image possibility distortion subsequently, or both's distortion.Therefore, absolute difference and be distortion indication.

Though this tunable filter FIL filtering image and serve as basis control in this video coding apparatus shown in Figure 3 with image one by one, each several part that can also filtering image and be that the basis adjusts with the image section.

Tunable filter FIL with respect to shown in Fig. 5 also notes following content.Can use the filter of any type in principle.For example tunable filter FIL can be a spatial filter or a horizontal filter.That is, can be a two-dimensional filtering or one-dimensional filtering device.This tunable filter FIL can be used as for example mean filter, Gaussian filter or a median filter realization.

This tunable filter FIL can also realize as a wavelet field filtering.This wavelet field filter can be a benchmark with a data block decomposition algorithm that is used for jpeg compressed image.In this case, this filter is carried out rim detection and smothing filtering, so that preserve the detection edge in this wavelet field.This rim detection can be a benchmark with a threshold value of determining in the pseudo-video encoder PENC shown in Fig. 3.Therefore, this threshold application is to the complexity of the present image that will encode.

Those image-regions of blocking effect, this wavelet field filter counter mass effect wherein may appear by a filtering.Therefore, wavelet field filter degree of reducing the definition of an image is less than the degree that reduces of a low pass filter.Because the wavelet field filter, promptly pursues an application preferably by image applications, so can the further counter mass effect of enough jpeg algorithms.Such algorithm provides the data block decomposability of a satisfaction and quite simple.

In above discussion, signal processing that to be not limited to by a definite frequency response be feature that term " filter " is very clear.Term " filter " should broadly explain, so that comprise various types of preliminary treatment of the distortion that will reduce otherwise introduced by processor.

Any reference marker in the claim should not be interpreted as the restriction to claim.

Claims

1. method of utilizing signal processor (PRC) to handle a signal (S), the method comprising the steps of:

The signal (S) that will handle of check (EXAM) so that obtain a distortion indication (IND), shows if this signal (S) is utilized this processor (PRC) and handles (PRC[S]) with the distortion level (DIST) that is introduced into;

Anticipate (FIL) this signal (S) according to this distortion indication (IND), so that obtain a pretreated signal (SF); With

Utilize this processor (PRC) to handle (PRC[SF]) this pretreated signal (SF).

2. signal processing apparatus that comprises signal processor (PRC) is characterized in that this signal processing apparatus further comprises:

A checking circuit (EXAM), be used to check will be processed signal (S), so that obtain distortion indication (IND), show if this signal (S) is utilized this processor (PRC) and handle (PRC[S]) the distortion level (DIST) that is introduced into;

A preprocessor (FIL), be used for according to this distortion indication this signal of (IND) preliminary treatment (S), so that obtain a pretreated signal (SF), this signal processor (PRC) is coupled to this preprocessor (FEL), be used for handling (PRC[SF]) this pretreated signal (SF).

3. the signal processing apparatus of claim 2 is characterized in that this preprocessor (FIL) comprises a tunable filter, and its filtering characteristic is adjusted according to this distortion indication (IND).

4. method for video coding comprises step:

The video data that check (PENC) will be encoded so that obtain a complexity indication (CI), shows the degree of difficulty of the video data that will be encoded;

According to this this video data of complexity indication (CI) filtering (FIL), so that obtain the video data of filtering; With

The video data of coding (ENC) this filtering.

5. a video coding apparatus comprises:

A checking circuit (FIL) is used to check the video data that will be encoded, so that obtain a complexity indication (CI), shows the degree of difficulty of the video data that will be encoded;

A tunable filter (FIL) is used for according to this this video data of complexity indication (CI) filtering (FIL), so that obtain the video data of filtering; With

A video encoder (ENC), the video data of this filtering that is used to encode.