CN1722843A - Digital signal conversion method - Google Patents

Abstract

An inputted digital signal of a first format (DV video signal) is restored to a variable-length code by having its framing cancelled by a de-framing section 11, then decoded by a variable-length decoding (VLD) section 12, inversely quantized by an inverse quantizing (IQ) section 13, and inversely weighted by an inverse weighting (IW) section 14. Then, required resolution conversion in the orthogonal transform domain (frequency domain) is carried out on the inversely weighted video signal by a resolution converting section 16. After that, the video signal having the resolution converted is weighted by a weighting (W) section 18, then quantized by a quantizing (Q) section 19, coded by variable-length coding by a variable-length coding (VLC) section 20, and outputted as a digital signal of a second format (MPEG video signal).

Description

Digital signal conversion method

Patent application of the present invention is dividing an application of following application for a patent for invention:

Application number: 98801684.2; The applying date: on November 05th, 1998; Denomination of invention: digital signal conversion method and digital signal transfer unit

Technical field

The present invention relates to by using the conversion process of orthogonal transform such as discrete cosine transform (DCT) the digital signal of compressed encoding, relate in particular to digital signal conversion method and digital signal transfer unit, be used to change the resolution between the compressed video signal of different-format.

Background technology

Traditionally, discrete cosine transform (DCT) (a kind of orthogonal transform coding) has been used as coded system, is used for compressed encoding still image data and dynamic menu data effectively.When this class of disposal has been carried out the digital signal of orthogonal transform, must change resolution or transform-based sometimes.

For example, if have the household digital video format for example the digital signal of first orthogonal transform of the resolution of 720 * 480 pixels to be converted into the digital signal of second orthogonal transform of the resolution of 360 * 240 pixels with so-called MPEG1 form, then first signal is carried out inverse orthogonal transformation, so that recover the signal of spatial domain, carry out conversion process then such as interpolation and desaturation (thinning), so that carry out orthogonal transform once more, be secondary signal with first conversion of signals like this.

In this mode, common situation be orthogonal transform digital signal by inverse transformation once so that recover primary signal, handle by the transform operation of needs then, and then carry out orthogonal transform.

Figure 28 shows the example structure that is used for the digital signal of having carried out DCT is carried out the conventional digital signal processing apparatus of above-mentioned conversion of resolution.

In this conventional digital chromacoder, the vision signal (hereinafter referred to as the DV vision signal) of so-called " DV form " (form of a kind of household digital vision signal), be used as the digital signal input of first form, the vision signal (hereinafter referred to as the MPEG vision signal) that meets the form of so-called MPEG (Motion Picture Experts Group) standard, the digital signal that is used as second form is exported.

Separate framing (de-framing) part 51 and be used to eliminate the framing of DV vision signal.Separate in the framing part 51 at this, be resumed according to the DV vision signal of so-called DV form framing and be variable-length codes.

52 pairs of length-changeable decoding (VLD) parts are carried out length-changeable decoding by separating the vision signal that framing part 51 reverts to variable-length codes.By with fixed rate compression, its data volume is reduced to 1/5 of about primary signal like this, and is encoded by variable length code, so that improve efficiency of data compression with the data of DV form compression.The decoding that length-changeable decoding part 52 is carried out corresponding to this variable length code.

53 pairs of vision signals by 52 decodings of length-changeable decoding part of re-quantization (IQ) part are carried out re-quantization.

Contrary weighting (IW) part 54 is carried out contrary weighting, and this contrary weighting is the inverse operation to the weighting of being undertaken by the vision signal of re-quantization part 53 re-quantizations.

Vision by utilizing the people is to the not very sensitive characteristic of the distortion of high frequency side, and ranking operation is the value for the DCT coefficient of the higher frequency components that reduces vision signal.Like this, the number with high frequency coefficient of 0 value increases, and variable length code efficient can be enhanced.The result is in some cases, can reduce the arithmetical operation amount of dct transform.

55 pairs of vision signals by contrary weighting part 54 contrary weightings of inverse discrete cosine transform (IDCT) part are carried out inverse DCT (inverse discrete cosine transform), like this DCT coefficient are reverted to the data of spatial domain, i.e. pixel data.

Then, 56 pairs of the conversion of resolution parts vision signal that reverts to pixel data by inverse discrete cosine transform part 55 is carried out the conversion of resolution that needs.

57 pairs of vision signals by conversion of resolution part 56 conversion of resolution of discrete cosine transform (DCT) part are carried out discrete cosine transform (DCT), so once more vision signal are converted to orthogonal transform coefficient (DCT coefficient).

58 pairs of conversion of resolution of weighting (W) part and the vision signal that is converted to the DCT coefficient are weighted.Above-mentioned the same of this weighted sum.

Quantize the vision signal that (Q) part 59 quantizes by 58 weightings of weighting part.

Then, 60 pairs of vision signals that quantized by quantized segment 59 of variable length code (VLC) part are carried out variable length code, and output result signal is as the MPEG vision signal.

Above-mentioned " MPEG " is the abbreviation of the Motion Picture Experts Group of ISO/IEC JTC1/SC29 (International Standards Organization/International Electrotechnical Commission, 1/ sub-committee 29 of joint technical committee).The ISO11172 standard is as the MPEG1 standard, and the ISO13818 standard is as the MPEG2 standard.In these international standards, at multimedia multiplexing part ISO11172-1 and ISO13818-1 by standardization, at video section ISO11172-2 and ISO13818-2 by standardization, at audio-frequency unit ISO11172-3 and ISO13818-3 by standardization.

According to ISO11172-2 or ISO13818-2 as the image compression encoding standard, on picture (frame or field) basis, pass through to use picture being correlated with at time or direction in space, picture signal is compressed coding, and use relevant on direction in space realizes by using DCT to encode.

In addition, this orthogonal transform is extensively adopted by various types of image information compressed encodings such as JPEG (joint image coding expert group) institute such as DCT.

Usually, orthogonal transform is transformed into orthogonal transform domain by the primary signal with time domain or spatial domain, such as frequency domain, makes compressed encoding have high compression efficiency and outstanding reproducibility.

Above-mentioned " DV form " is used for data volume with digital video signal and is compressed to approximately 1/5, is recorded on the tape with component.The DV form is used for household digital video-unit and some specialty digital video apparatus.This DV form has been realized effective compression of vision signal by with discrete cosine transform (DCT) and variable length code (VLC) combination.

Simultaneously, orthogonal transform and inverse orthogonal transformation for such as discrete cosine transform (DCT) need a large amount of calculating usually.Thereby the problem of bringing is that the conversion of resolution of above-mentioned vision signal can not be carried out effectively.Equally, owing to the increase along with amount of calculation, mistake is accumulated, and has brought the problem of Signal Degrade.

Of the present invention open

Consider the aforesaid state of present technique, an object of the present invention is to provide a kind of digital signal conversion method and digital signal transfer unit, can make the effective and less deterioration of conversion process by reducing arithmetic processing amount by the data volume of the handled signal of conversion of resolution that is used to be transformed into different-format such as conversion of resolution.

In order to solve aforesaid problem, a kind of digital signal conversion method according to the present invention comprises: the data extract step, from the relevant block of the digital signal of first form of the orthogonal transform coefficient piece that comprises scheduled unit, extract a part of orthogonal transform coefficient, thus the component part piece; The inverse orthogonal transformation step on the basis of described part piece, is carried out inverse orthogonal transformation to the orthogonal transform coefficient that constitutes each part piece; Part piece Connection Step connects the described part piece of being handled by inverse orthogonal transformation, thereby constitutes the new piece of described scheduled unit; With the orthogonal transform step, on described basis, carry out orthogonal transform to described new, thereby produce second digital signal of the described new orthogonal transform block that comprises described scheduled unit.

Equally, in order to solve aforesaid problem, a kind of digital signal conversion method according to the present invention comprises: the inverse orthogonal transformation step, to the digital signal of first form of the orthogonal transform coefficient piece that comprises scheduled unit, carry out inverse orthogonal transformation based on described; The piece partiting step is divided each piece of the digital signal of described first form of being handled by inverse orthogonal transformation; The orthogonal transform step based on the piece of dividing, is carried out orthogonal transform to the orthogonal transform coefficient of the piece that constitutes each division; With the data amplification procedure, the piece that is interpolated into each orthogonal transform with the orthogonal transform coefficient of predetermined value to be constituting described scheduled unit, thereby produces the digital signal of second form.

Equally, in order to solve aforesaid problem, a kind of digital signal transfer unit according to the present invention comprises: decoding device, the digital signal of first form of the orthogonal transform coefficient that comprises scheduled unit of being used to decode; The re-quantization device is used for the digital signal of described decoding is carried out re-quantization; The conversion of resolution device is used for extracting a part of orthogonal transform coefficient from the adjacent block of the orthogonal transform coefficient piece of the scheduled unit of the digital signal of described re-quantization, thus the component part piece, conversion resolution; Quantization device is used to quantize the described digital signal of being handled by conversion of resolution; And code device, be used to the digital signal of described quantification of encoding, thereby produce the digital signal of second form.

Equally, in order to solve aforesaid problem, a kind of digital signal transfer unit according to the present invention comprises: decoding device, the digital signal of first form by using the orthogonal transform compressed encoding that is used to decode; The re-quantization device is used for the digital signal re-quantization to described decoding; The conversion of resolution device is used for the orthogonal transform coefficient of predetermined value is interpolated into each predetermined block of the digital signal of described re-quantization, thereby constitutes described scheduled unit, conversion resolution; Quantization device is used to quantize the described digital signal of being handled by conversion of resolution; And code device, be used to the digital signal of described re-quantization of encoding, thereby produce the digital signal of second form.

Equally, in order to solve aforesaid problem, according to a kind of digital signal conversion method of the present invention, the digital signal of first form that is used for comprising the orthogonal transform coefficient piece of scheduled unit is converted to the digital signal of second form of the new orthogonal transform coefficient block that comprises another scheduled unit.In the method, be included in the data amount information in the digital signal of described first form by utilization, control the data volume of the digital signal of described second form.

Equally, in order to solve aforesaid problem, according to a kind of digital signal transfer unit of the present invention, be used for to comprise that the digital signal of first form of the orthogonal transform coefficient piece of scheduled unit is converted to the digital signal of second form of the new orthogonal transform coefficient block that comprises another scheduled unit, this device comprises: decoding device, the digital signal of described first form that is used to decode; The re-quantization device is used for the digital signal of described decoding is carried out re-quantization; Chromacoder is used to follow the format conversion of the digital signal of described re-quantization, carries out signal processing; Quantization device is used to quantize the described digital signal of being handled by signal processing; The data volume control device is used to control the data volume of described quantization device; And code device, its data volume that is used to encode is controlled and the digital signal of quantification by described data volume control device, thereby produces the digital signal of described second form.

Equally, in order to solve aforesaid problem,, be used for the digital signal of first form is converted to the digital signal of second form according to a kind of digital signal conversion method of the present invention.This method comprises: decoding step, the digital signal of described first form of decoding; The conversion of signals step is converted to the digital signal of the decoding of described first form digital signal of described second form; Coding step, the digital signal of described second form of encoding; With the weighted step, integrally carry out the weighting of digital signal of described second form of contrary weighted sum of the digital signal of described first form.

Equally, in order to solve aforesaid problem,, be used for the digital signal of first form is converted to the digital signal of second form according to a kind of digital signal transfer unit of the present invention.This device comprises: decoding device, the digital signal of described first form that is used to decode; Chromacoder is used for the digital signal of the decoding of described first form is converted to the digital signal of described second form; Code device, the digital signal of described second form that is used to encode; With the weighted device, be used for integrally carrying out the weighting of digital signal of described second form of contrary weighted sum of the digital signal of described first form.

Equally, in order to solve aforesaid problem,, the input information signal of the compressed encoding of band motion detection with the decoding of motion compensation, is carried out signal conversion processes to this decoded signal according to the present invention.Then, based on the motion vector information of described input information signal, the signal of conversion of band motion detection is carried out compressed encoding handle.

Equally, in order to solve aforesaid problem,, the input information signal is carried out partial decoding of h handle according to the present invention, described input information signal is by comprising that the predictive coding of being with motion detection and the compressed encoding of orthogonal transform coding handle, thereby obtains the signal of the decoding of orthogonal transform domain.Then, the signal of the decoding of orthogonal transform domain is carried out signal conversion processes,, the signal of described conversion is handled with the compressed encoding of motion compensated prediction by using motion detection based on the motion vector information of described input information signal.

Equally, in order to solve aforesaid problem,, the input information signal is carried out partial decoding of h handle according to the present invention, described input information signal is handled by the compressed encoding that comprises the predictive coding of being with motion detection and orthogonal transform coding, thereby obtains the signal of orthogonal transform domain.Then, this signal is carried out signal conversion processes, by adding the motion vector information of changing based on the motion vector information of described input information signal, the signal of described conversion is carried out compressed encoding handle.

Equally, in order to solve aforesaid problem, according to the present invention, the digital signal of first form of decoding, the digital signal of described first form have the dynamic mode/static schema information that adds in advance, and the signal of decoding is carried out signal conversion processes.Then, to each predetermined block of the signal of described conversion, according to described dynamic mode/static schema information, whether difference carries out the inter-frame difference coding.Based on the result of difference, the signal of the described conversion of encoding, output is by the digital signal of second form of the encoding process of use frame-to-frame differences.

Equally,,, the digital signal of first form is carried out partial decoding of h handle according to the present invention in order to solve aforesaid problem, thus the signal of acquisition orthogonal transform domain.Signal to described orthogonal transform domain carries out signal conversion processes, and to each predetermined block of the signal of described conversion, according to the maximum of the absolute value of the frame-to-frame differences of the signal of described conversion, whether difference carries out the inter-frame difference coding.Based on the result of difference, the signal of the described conversion of encoding is exported the digital signal of second form.

In addition, in order to solve aforesaid problem, according to the present invention, to the signal that comprises the intraframe coding of handling by intraframe coding with by the digital signal of first form of the forward prediction code signal of the forward of band motion detection and bidirectional interframe predictive encoding process and bi-directional predictive coding signal, the signal and the forward prediction code signal of intraframe coding carried out inverse orthogonal transformation.Based on inverse orthogonal transformation output, generation will be added to the motion compensation output of the signal of the forward prediction encoded signals of partial decoding of h and bi-directional predictive coding.Orthogonal transform is carried out in motion compensation output, orthogonal transform output is added to the forward prediction encoded signals of described partial decoding of h and the signal of bi-directional predictive coding.Signal based on addition output is carried out compressed encoding, export the digital signal of second form.

Brief description of drawings

Fig. 1 is the block diagram of example structure that the digital signal transfer unit of first embodiment according to the invention is shown.

Fig. 2 is illustrated in the principle of the conversion of resolution of orthogonal transform domain.

Fig. 3 is illustrated in the principle of the conversion of resolution of orthogonal transform domain.

Fig. 4 A to 4C schematically shows the digital signal conversion by first embodiment according to the invention, and the DV vision signal is converted into the state of MPEG vision signal.

Fig. 5 illustrates the relation between DV form and the mpeg format.

Fig. 6 illustrates the basic calculating process that conversion of resolution is handled.

Fig. 7 A and 7B illustrate DV form " static schema " and " dynamic mode ".

Fig. 8 is illustrated in the process of conversion process in " static schema ".

Fig. 9 A to 9C is the block diagram that illustrates according to the example structure of the digital signal transfer unit of second embodiment of the present invention.

Figure 10 is illustrated in the process of conversion process in the amplification of image.

Figure 11 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 3rd embodiment of the present invention.

Figure 12 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 4th embodiment of the present invention.

Figure 13 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 5th embodiment of the present invention.

Figure 14 is the block diagram that the example structure of digital signal transfer unit according to a sixth embodiment of the present is shown.

Figure 15 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 7th embodiment of the present invention.

Figure 16 is illustrated among the 7th embodiment of the present invention, when the DV vision signal is converted into mpeg signal, is used to set the flow chart of basic process of quantizer ratio of each macro block (MB) of every frame.

Figure 17 is illustrated among the 7th embodiment of the present invention, applies the flow chart of the basic process that feeds back to next frame by using predetermined quantitative device ratio.

Figure 18 illustrates the block diagram of example structure that is used for the MPEG vision signal is converted to the conventional digital chromacoder of DV vision signal.

Figure 19 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 8th embodiment of the present invention.

Figure 20 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 9th embodiment of the present invention.

Figure 21 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the of the present invention ten embodiment.

Figure 22 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 11 embodiment of the present invention.

Figure 23 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 12 embodiment of the present invention.

Figure 24 is illustrated in motion compensation and the motion estimation process in the orthogonal transform domain among the 12 embodiment of the present invention, shows the state that macro block B stretches on a plurality of macro blocks of reference picture.

Figure 25 is illustrated in motion compensation and the motion estimation process in the orthogonal transform domain among the 12 embodiment of the present invention, shows the conversion process of reference macroblock.

Figure 26 is illustrated in motion compensation and the motion estimation process in the orthogonal transform domain among the 12 embodiment of the present invention, shows the transfer process of reference macroblock.

Figure 27 is the block diagram that illustrates according to the example structure of the digital signal transfer unit of the 13 embodiment of the present invention.

Figure 28 is the block diagram that the example structure of traditional digital signal transfer unit is shown.

Realize best mode of the present invention

Below with reference to description of drawings the preferred embodiments of the present invention.

At first, with the structure of describing according to digital signal transfer unit of the present invention, then will be with reference to this structrual description according to digital signal conversion method of the present invention.

Fig. 1 illustrates the example structure as the major part of the digital signal transfer unit of first embodiment of the present invention.Although conversion of signals is an example with the conversion of resolution, in fact conversion of signals is not limited to conversion of resolution, and various types of signal processing such as format conversion and Filtering Processing can be used.

In this digital signal transfer unit, the vision signal (hereinafter referred to as the DV vision signal) of above-mentioned what is called " DV form " is used as the input of first digital signal, and the vision signal (hereinafter referred to as the MPEG vision signal) that meets the form of MPEG (Motion Picture Experts Group) standard is used as the output of second digital signal.

Separate framing part 11 and be used to eliminate the framing of DV vision signal.Separate in the framing part 11 at this, be resumed according to the DV vision signal of predetermined format (so-called DV form) framing and be variable-length codes.

12 pairs of length-changeable decoding (VLD) parts are carried out length-changeable decoding by separating the vision signal that framing part 11 reverts to variable-length codes.

13 pairs of vision signals by 12 decodings of length-changeable decoding part of re-quantization (IQ) part are carried out re-quantization.

Contrary weighting (IW) part 14 is carried out contrary weighting, and this contrary weighting is the inverse operation to the weighting of being undertaken by the vision signal of re-quantization part 13 re-quantizations.

If carry out conversion of resolution, as the example of signal conversion processes, conversion of resolution part 16 conversion of resolution to needing being undertaken by the vision signal of contrary weighting part 14 contrary weightings in orthogonal transform domain (frequency domain) then.

18 pairs of vision signals of being handled by conversion of resolution of weighting (W) part are weighted.

Quantize the vision signal that (Q) part 19 quantizes by 18 weightings of weighting part.

Then, 20 pairs of vision signals that quantized by quantized segment 19 of variable length code (VLC) part are carried out variable length code, and output result signal is as the MPEG vision signal.

Can make similar according to each parts of above-mentioned digital signal transfer unit of the present invention shown in Figure 1 in the structure of each parts of traditional digital signal transfer unit shown in Figure 28.

Yet digital signal transfer unit according to the present invention is different from traditional digital signal transfer unit part and is, before conversion of resolution part 16 and inverse discrete cosine transform (IDCT) part and discrete cosine transform (DCT) part be not provided afterwards.

That is, in the conventional digital chromacoder, the orthogonal transform coefficient of the digital signal of first form of input so that be resumed data into spatial domain (on frequency axis), is carried out the translation operation that needs by inverse orthogonal transformation then.Therefore, carry out orthogonal transform once more, so that data are reverted to orthogonal transform coefficient.

On the contrary, in digital signal transfer unit according to the present invention, in orthogonal transform coefficient territory (frequency domain), carry out the translation operation of needs of orthogonal transform coefficient of the supplied with digital signal of first form, before being used to carry out and inverse orthogonal transformation device and orthogonal converter be not provided afterwards such as the device of the conversion process of conversion of resolution.

With reference now to Fig. 2 and 3, the principle that the conversion of resolution in the conversion of resolution part 16 is handled is described.

In Fig. 2, the input orthogonal transform matrix produces the orthogonal transform matrix Ts that part 1 produces the orthogonal transform that expression carried out in advance to supplied with digital signal 5 _(k)Inverse matrix Ts _(k) ^-1, give transformation matrix with this inverse matrix and produce part 3.The output orthogonal transformation matrix produces the inverse-transform matrix Td that part 2 produces the inverse orthogonal transformation that will carry out the output digital signal corresponding to expression _(L) ^-1Orthogonal transform matrix Td (L), give transformation matrix with this orthogonal transform matrix and produce part 3.Transformation matrix produces part 3 and produces the transformation matrix D that is used to carry out such as the conversion process of the conversion of resolution in the frequency domain, gives conversion of signals part 4 with this transformation matrix.4 conversions of conversion of signals part have been transformed into the supplied with digital signal 5 of frequency domain by orthogonal transform, keep the orthogonal transform domain such as frequency domain simultaneously, produce output digital signal 6.

Especially, as shown in Figure 3, by using orthogonal transform matrix Ts _(k), signal (primary signal) A of original time domain (or spatial domain) is switched to frequency domain, to produce frequency signal B ₁(corresponding to supplied with digital signal 5).This frequency signal B ₁Tapered to N/L (perhaps amplifying) by conversion of signals part 4, to produce frequency signal B ₂(corresponding to output digital signal 6).By using inverse-transform matrix Td _(L) ^-1, this frequency signal B ₂By inverse orthogonal transformation, to produce the signal C of time domain.

In the example of Fig. 3, each conversion block of one dimension primary signal A is by orthogonal transform, each conversion block length is k, the adjacent block of the m of the resulting conversion block of a frequency domain unit, promptly has the L (continuous frequency signal of the length of=k * m), be converted into and have N that (piece of the length of N＜L), promptly integral body tapers to N/L.

In the following description, has the orthogonal transform base vector that length is n e ₁, e ₂..., e _nThe matrix (orthogonal transform matrix) that is arranged at corresponding line is represented as T _(n), its inverse-transform matrix is represented as T _(n) ^-1In this is described, xRepresent the x vector.In this case, each matrix is n rank forward matrixes (forwardmatrix).For example, the one dimension dct transform matrix T of n=8 ₍₈₎Represent by following equation (1).

$T_{\left(8\right)}=\left(\begin{array}{c}\underset{\‾}{e_1}\\ \underset{\‾}{e_2}\\ \underset{\‾}{e_3}\\ \underset{\‾}{e_4}\\ \underset{\‾}{e_5}\\ \underset{\‾}{e_6}\\ \underset{\‾}{e_7}\\ \underset{\‾}{e_8}\end{array}\right)=\frac{1}{2}\left(\begin{array}{cccccccc}1/\sqrt{2}& 1/\sqrt{2}& 1/\sqrt{2}& 1/\sqrt{2}& 1/\sqrt{2}& 1/\sqrt{2}& 1/\sqrt{2}& 1/\sqrt{2}\\ \cos (\mathrm{\π}/16)& \cos (3\mathrm{\π}/16)& \cos (5\mathrm{\π}/16)& \cos (7\mathrm{\π}/16)& \cos (9\mathrm{\π}/16)& \cos (11\mathrm{\π}/16)& \cos (13\mathrm{\π}/16)& \cos (15\mathrm{\π}/16)\\ \cos (2\mathrm{\π}/16)& \cos (6\mathrm{\π}/16)& \cos (10\mathrm{\π}/16)& \cos (14\mathrm{\π}/16)& \cos (18\mathrm{\π}/16)& \cos (22\mathrm{\π}/16)& \cos (26\mathrm{\π}/16)& \cos (30\mathrm{\π}/16)\\ \cos (3\mathrm{\π}/16)& \cos (9\mathrm{\π}/16)& \cos (15\mathrm{\π}/16)& \cos (21\mathrm{\π}/16)& \cos (27\mathrm{\π}/16)& \cos (33\mathrm{\π}/16)& \cos (39\mathrm{\π}/16)& \cos (45\mathrm{\π}/16)\\ \cos (4\mathrm{\π}/16)& \cos (12\mathrm{\π}/16)& \cos (20\mathrm{\π}/16)& \cos (28\mathrm{\π}/16)& \cos (36\mathrm{\π}/16)& \cos (44\mathrm{\π}/16)& \cos (52\mathrm{\π}/16)& \cos (60\mathrm{\π}/16)\\ \cos (5\mathrm{\π}/16)& \cos (15\mathrm{\π}/16)& \cos (25\mathrm{\π}/16)& \cos (35\mathrm{\π}/16)& \cos (45\mathrm{\π}/16)& \cos (55\mathrm{\π}/16)& \cos (65\mathrm{\π}/16)& \cos (75\mathrm{\π}/16)\\ \cos (6\mathrm{\π}/16)& \cos (18\mathrm{\π}/16)& \cos (30\mathrm{\π}/16)& \cos (42\mathrm{\π}/16)& \cos (54\mathrm{\π}/16)& \cos (66\mathrm{\π}/16)& \cos (87\mathrm{\π}/16)& \cos (90\mathrm{\π}/16)\\ \cos (7\mathrm{\π}/16)& \cos (21\mathrm{\π}/16)& \cos (35\mathrm{\π}/16)& \cos (49\mathrm{\π}/16)& \cos (63\mathrm{\π}/16)& \cos (77\mathrm{\π}/16)& \cos (91\mathrm{\π}/16)& \cos (105\mathrm{\π}/16)\end{array}\right)---\left(1\right)$

In Fig. 3, pass through to use orthogonal transform matrix Ts when _(k)When the size of orthogonal transform block that is orthogonally transformed into the supplied with digital signal 5 of frequency domain is k, that is, when basic length equaled k, the input orthogonal transform matrix produced part 1 and produces inverse orthogonal transformation matrix T s _(k) ^-1, the output orthogonal transformation matrix produces part 2 and produces and have basic length L (the orthogonal transform matrix Td of=k * m) _(L)

At this, produce the inverse orthogonal transformation matrix T s that part 1 produces by the input orthogonal transform matrix _(k) ^-1, the contrary processing corresponding to the orthogonal transform in producing supplied with digital signal 5 is handled produces the orthogonal transform matrix Td that part 2 produces by the output orthogonal transformation matrix _(L),, be about to contrary processing that conversion of signals is handled to the inverse orthogonal transformation in the time domain corresponding in the output digital signal of decoding by 4 conversions of conversion of signals part.These orthogonal transform matrixs produce the base vector that part 1 and 2 can produce random length.

It can be that same orthogonal transform matrix produces part that orthogonal transform matrix produces part 1 and 2.In this case, orthogonal transform matrix Ts _(k)And Td _(L)Become the orthogonal transform matrix of same-type, except their basic length differs from one another.Each different orthogonal transform system all exists orthogonal transform matrix to produce part.

Then, transformation matrix produces part 3 by produced the inverse orthogonal transformation matrix T s of m unit of part 1 generation by the input orthogonal transform matrix in the diagonal arrangement _(k) ^-1, produce L rank forward matrix A, shown in following equation (2).When the basic length of output digital signal 6 equaled N, transformation matrix produced part 3 and takes out orthogonal transform matrix Td _(L)The low frequency base vector of N unit, produce the matrix B of the capable L row of N.

Yet, in expression formula, work as Td _(L)When representing by following base vector, e ₁, e ₂..., e _NIt is the low frequency base vector of N unit.

Then, calculation equation

D＝α·B·A …(5)

To produce the matrix D of the capable L row of N.This matrix D is a transformation matrix, is used for economy (or magnification ratio) conversion resolution with N/L.In this equation, α is scalar value or vector value, is the coefficient of level correction (levelcorrection).

The signal transformation part 4 of Fig. 2 is with the supplied with digital signal B of frequency domain ₁M piece be collected as one group, this signal is divided into the first piece (meta-block) (one of them first piece comprises m piece) with L size, as shown in Figure 3.If supplied with digital signal B ₁Length be not the multiple of L, this signal uses puppet (dummy) data such as 0 to replenish and fill up, and becomes the multiple of L.So first piece of Chan Shenging by Mi (i=0 wherein, 1,2 ...) represent.

The principle that above-mentioned conversion of resolution is handled is described in detail in the PCT/JP98/02653 that was submitted on June 16th, 1998 by this assignee.

Digital signal conversion method below with reference to first embodiment of structrual description of above-mentioned digital signal transfer unit.

Fig. 4 A to 4C schematically shows by digital signal conversion according to an embodiment of the invention, and the DV vision signal is converted into the processing in the MPEG vision signal.This processing is mainly undertaken by the conversion of resolution part 16 in the digital signal transfer unit of embodiments of the invention shown in Figure 1.

In the following description, one dimension DCT coefficient block is used as example.But the processing of bidimensional DCT coefficient can be carried out similarly.

At first, four DCT coefficients of lower frequency side are taken out from adjacent block (i) and (i+1) each, and piece (i) and (i+1) each comprises 8 DCT coefficients of the digital signal of first form is shown in Fig. 4 A.That is, at 8 DCT coefficient a0 of piece (i), a1, a2, a3 ..., among the a7, have only 4 DCT coefficient a0 of lower frequency side, a1, a2, a3 is removed, and draws the part piece of the decreased number to 1/2 with DCT coefficient, similarly, at 8 DCT coefficient b0 of piece (i+1), b1, b2, b3 ..., among the b7, have only 4 DCT coefficient b0 of lower frequency side, b1, b2, b3 is removed, and draws the part piece of the decreased number to 1/2 with DCT coefficient.Concentration of energy was in the characteristic of the low frequency of DC and AC when taking-up was based on vision signal by frequency inverted in the computing of the DCT of lower frequency side coefficient.

Then, each the part piece that comprises 4 DCT coefficients is carried out 4 inverse discrete cosine transforms (4 IDCT), obtain the pixel data of reduction like this.These pixel datas are represented as the pixel data p0 among Fig. 4 B, p1, p2, p3 and pixel data p4, p5, p6, p7.

Then, comprise that each part piece of the pixel data of the reduction of being handled by inverse discrete cosine transform is connected, have piece with the identical size of original block with generation.That is, pixel data p0, p1, p2, p3 and pixel data p4, p5, p6, p7 is connected, and comprises the new piece of 8 pixel datas with generation.

Then, the new piece that comprises 8 pixel datas is carried out 8 discrete cosine transforms, generation comprises 8 DCT coefficient c0 like this, c1, and c2, c3 ..., the piece (j) of c7 is shown in Fig. 4 C.

By above-mentioned process, by with the number desaturation of the orthogonal transform coefficient (DCT coefficient) of each predetermined block unit to half, vision signal can be converted into the vision signal of different resolution form.Will be to 1/4 the time when the number of DCT coefficient by desaturation, above-mentioned processing is carried out twice continuously.

For example, above-mentioned conversion of resolution is handled the conversion that can be applied to from the DV form to the MPEG1 form.

Referring now to Fig. 5 relation between DV form and the mpeg format and the format conversion between these forms are described.

Especially, situation for the vision signal that meets TSC-system formula shown in Figure 5, the vision signal of DV form is the compressed video signal with resolution of 720 * 480 pixels, and the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 1: 1.The vision signal of MPEG1 form is the compressed video signal with resolution of 360 * 240 pixels, and the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 2: 0.Thereby, in this case, handle by above-mentioned conversion of resolution according to the present invention, can be reduced to 1/2, can be reduced to 1/4 at the number of the DCT coefficient of the vertical direction of aberration (C) signal in the level of brightness (Y) signal and the number of vertical direction DCT coefficient.

Because odd-numbered line and even number line alternately adopt the value of 4: 2: 0 and 4: 0: 2,4: 2: 0 ratio is represented the value of odd-numbered line or even number line.

On the other hand, for the situation of the vision signal that meets pal mode, the vision signal of DV form is the compressed video signal with resolution of 720 * 576 pixels, and the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 2: 0.The vision signal of MPEG1 form is the compressed video signal with resolution of 360 * 288 pixels, and the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 2: 0.Thereby, in this case, handle by above-mentioned conversion of resolution according to the present invention, can be reduced to 1/2, can be reduced to 1/2 at the number of the DCT coefficient of the level of C signal and vertical direction in the level of Y-signal and the number of vertical direction DCT coefficient.

Above-mentioned conversion of resolution is handled the conversion that can be applied to similarly from the DV form to the MPEG2 form.

For the situation of the vision signal that meets the TSC-system formula, the vision signal of MPEG2 form is the compressed video signal with resolution of 720 * 480 pixels, and the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 2: 0.Thereby, in this case, can be reduced to 1/2 at the number of the DCT coefficient of the vertical direction of C signal, can be doubled at the number of the DCT coefficient of the horizontal direction of C signal, and not carry out the conversion process of Y-signal.To the method for this amplification, will describe later on.

For the situation of the vision signal that meets pal mode, the vision signal of MPEG2 form is the compressed video signal with resolution of 720 * 576 pixels, and the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 2: 0.Thereby, in this case, do not need Y-signal or C signal are carried out conversion process.

Fig. 6 illustrates the basic calculating process that above-mentioned conversion of resolution is handled.

Particularly, by four the DCT coefficient a0s of connection from two adjacent blocks taking-ups of the digital signal of first form of input, a1, a2, a3 and four DCT coefficient b0, b1, the piece that comprises 8 DCT coefficients that b2, b3 draw, be multiplied by one (8 * 8) matrix, this matrix is included in cornerwise two inverse discrete cosine transform matrixes (IDCT4), and each is (4 * 4) matrix, and other component is 0.

It is long-pending to be multiplied by one once more and to be the discrete cosine transform matrix (DCT8) of (8 * 8) matrix, just produced to comprise 8 DCT coefficient c0, and c1, c2, c3 ..., the new piece of c7.

In digital signal conversion method according to the present invention, handle owing in DCT territory (frequency domain), carry out conversion of resolution, be unnecessary at inverse DCT before the conversion of resolution and the DCT after conversion of resolution.In addition, comprise (8 * 8) matrix of two (4 * 4) inverse discrete cosine transform matrixes (IDCT4) and amassing as transformation matrix D of (8 * 8) discrete cosine transform matrix by obtaining in advance at diagonal, the arithmetical operation amount can be reduced effectively.

The MPEG1 Video signal processing that is used for being converted to as the DV vision signal of the digital signal of first form as the digital signal of second form will describe in further detail.

The DV form has " static schema " and " dynamic mode ", switches according to the motion detection result of picture.For example, these patterns were distinguished by motion detection before the DCT of each (8 * 8) matrix of video-frequency band, carried out DCT according to the result who distinguishes with a certain pattern.The whole bag of tricks of motion detection can be considered.Particularly, can adopt than the absolute value sum of poor between the competiting place (inter-field diffrence) and the method for predetermined threshold.

" static schema " is the fundamental mode of DV form, wherein (8 * 8) pixel in a piece carried out (8 * 8) DCT.

(8 * 8) piece is made of a DC component and 63 AC components.

" dynamic mode " is used to avoid a kind of situation, if promptly carry out DCT when object motion, because interlacing scan has reduced compression efficiency by the dispersion of energy.Under this dynamic mode, (8 * 8) piece is divided into first (4 * 8) piece and (4 * 8) piece of second, and the pixel data of each (4 * 8) piece is carried out (4 * 8) DCT.Like this, be suppressed, can prevent that compression ratio from reducing in the increase of the high fdrequency component of vertical direction.

Each above-mentioned (4 * 8) piece is made of a DC component and 31 AC components.

Like this, in the DV form, block structure is different between static schema and dynamic mode.Thereby, in order to carry out similar processing to subsequent treatment, after the DCT of each (4 * 8) piece, the coefficient of the same order by obtaining each piece and with poor, constituted (8 * 8) piece of relative dynamic mode block.By this processing, the dynamic mode piece can be counted as by a DC component and 63 AC components and constitute, and is similar to the static schema piece.

Simultaneously, be converted in the vision signal of MPEG1 form, must only separate a field,, do not have the notion of field owing to the vision signal that the MPEG1 form only disposed for 30 frame/seconds in vision signal with the DV form.

Fig. 7 A schematically shows the processing that is used for isolation field in the DCT coefficient that will be converted to the MPEG1 form according to the DCT coefficient of " dynamic mode (2 * 4 * 8 DCT pattern) " of DV form.

First (4 * 8) piece 31a of the DCT coefficient block 31 of (8 * 8) be first coefficient and second coefficient and (A+B), and second (4 * 8) piece 31b of the DCT coefficient block 31 of (8 * 8) is poor (A-B) of two coefficient.

Thereby, by first (4 * 8) piece 31a and second (4 * 8) piece 31b addition,, can obtain comprising (4 * 8) piece 35a of the DCT coefficient of first (A) then with it with divided by 2 with the DCT coefficient block 31 of (8 * 8).Similarly, by from the piece 31a of first (4 * 8), deducting second (4 * 8) piece 31b, then with its difference divided by 2, can obtain comprising (4 * 8) piece 35b of the discrete cosine coefficient of second (B).That is, handle (8 * 8) piece 35 that can obtain having isolation field by this.

Then, to these one of, for example first DCT coefficient carries out above-mentioned conversion of resolution and handles.

Fig. 7 B is shown schematically in the processing that is used for isolation field under " static schema (8 * 8 DCT pattern) ".

In the DCT of (8 * 8) coefficient block 32, the DCT coefficient of first (A) and the DCT coefficient of second (B) are mixed.Like this, must carry out conversion process, by using described below separating treatment,, be used for (4 * 8) the piece 35b that obtains to comprise (4 * 8) the piece 35a of first (A) and comprise second (B) similarly by subtracting each other between (4 * 8) piece 35a and (4 * 8) piece 35b.

Fig. 8 is illustrated in the process of " static schema " midfield separating treatment.

At first, comprise 8 DCT coefficient d 0, d1, d2, d3 ..., the input of d7 is multiplied by 8 rank inverse discrete cosine transform matrixes (IDCT8), to recover pixel data.

Then, these data are multiplied by and are used for (8 * 8) matrix of separating, thereby will (8 * 8) piece be divided into first of upside with downside second, and each is (4 * 8) piece.

Then, these data multiply by (8 * 8) piece again, and this (8 * 8) piece is included in two the discrete cosine transform matrixes (DCT4) on the diagonal, and each is (4 * 4) matrix.

Like this, four DCT coefficient e0 of first have just been obtained comprising, e1, e2, four DCT coefficient f0 of e3 and second, f1, f2, eight DCT coefficients of f3.

Then, to these one of, for example first DCT coefficient carries out above-mentioned conversion of resolution and handles.

In digital signal conversion method according to the present invention,, be unnecessary at inverse DCT before the conversion of resolution and the DCT after conversion of resolution owing in DCT territory (frequency domain), carry out conversion of resolution.In addition, comprise (8 * 8) matrix of two (4 * 4) inverse discrete cosine transform matrixes (IDCT4) and amassing of (8 * 8) discrete cosine transform matrix by obtaining in advance at diagonal, the arithmetical operation amount can be reduced effectively.

Above-mentioned conversion of resolution is in order to reduce image.Hereinafter, the conversion of resolution that is used for enlarged image is handled and will be described as second embodiment.

The schematically illustrated state that is converted into the MPEG2 vision signal by digital signal conversion method according to the present invention, DV vision signal of Fig. 9 A to 9C.

Equally in the following description, one dimension DCT coefficient is used.But similarly processing can be carried out bidimensional DCT coefficient.

At first, the piece (u) that comprises 8 orthogonal coefficients (DCT coefficient g0 is to g7) shown in Fig. 9 A is carried out 8 inverse discrete cosine transforms (8 IDCT), recover 8 pixel datas (h0 is to h7) like this.

Then, comprise that the piece of 8 pixel datas is divided into two parts, thereby produce two part pieces that respectively comprise four pixel datas.

Then, two part pieces that respectively comprise 4 DCT coefficients are carried out 4 DCT, thereby produce two part pieces respectively comprising 4 DCT coefficients (i0 to i3 and j0 to j3).

Then, shown in Fig. 9 C, comprise that respectively each high frequency side of two part pieces of 4 pixel datas fills up as four DCT coefficients with 0.Like this, produced the piece that respectively comprises 8 DCT coefficients (v) and piece (v+1).

According to said process, in orthogonal transform domain, carried out the conversion of resolution between the compressed video signal of different-format.

Figure 10 shows in this case, the process of conversion process.

At first, comprise 8 DCT coefficient g0, g1, g2, g3 ..., the input of g7 is multiplied by 8 rank inverse discrete cosine transform (IDCT) matrixes, to recover 8 pixel datas.

Then, comprise that the piece of 8 pixel datas is divided into two parts, thereby produce two part pieces that respectively comprise four pixel datas.

Then, respectively comprise each of two part pieces of four DCT coefficients, be multiplied by and comprise that 4 discrete cosine transform matrixes are as upside (4 * 4) matrix and 0 matrix (4 * 8) matrix as downside (4 * 4) matrix.Like this, two part pieces comprising 8 DCT coefficients (i0 to i7 and j0 to j7) have been produced.

By such processing, obtain two DCT coefficient block from a piece.Thereby, can amplify resolution at frequency domain.

Under the situation of TSC-system formula, in order to be the MPEG2 form with the DV format conversion, needn't carry out the level and the vertical transition of brightness signal Y, be twice but must in the horizontal direction color difference signal C be put, in vertical direction color difference signal C is tapered to 1/2, as shown in Figure 5.Thereby it is being the conversion of resolution of MPEG2 form color difference signal C in the horizontal direction with the DV format conversion that above-mentioned processing and amplifying is used for.

Figure 11 illustrates the example structure according to the major part of the digital signal transfer unit of the 3rd embodiment of the present invention.The structure division identical with first embodiment represented with identical label.Be weighting part 18 and be combined into weighted part 21 by integral body with the different of structure of Fig. 1 against weighting parts 14.

Especially, weighted (IW*W) part 21 is integrally carried out contrary weighted sum weighting, contrary weighting is that weighting is the MPEG vision signal that is used for as the output digital signal of second form to the inverse operation of the weighting of carrying out as the DV vision signal of the supplied with digital signal of first form.

Utilize this structure, because the weighted of the contrary weighted of the incoming video signal of first form and the outputting video signal of second form can integrally be carried out, compared by the situation of carrying out respectively with weighted with contrary weighted, the quantity of calculating can be reduced.

In the digital signal transfer unit of the 3rd embodiment shown in Figure 11, provide weighted part 21 in the following stages of conversion of resolution part 16.Yet weighted partly may be provided in the level before the conversion of resolution part 16.

Figure 12 illustrates the digital signal transfer unit according to the 4th embodiment of the present invention, and wherein weighted part 22 is provided at conversion of resolution part 16 level before.The part of the structure of this digital signal transfer unit shown in Figure 12 can be similar to the appropriate section of the digital signal transfer unit of Figure 11.

The weighted and the above-mentioned weighted of weighting of contrary weighted sum second digital signal that is used for integrally carrying out the digital signal of first form can be carried out before or after such as the orthogonal transform of discrete cosine transform (DCT).This is because its arithmetical operation is linear operation.

With reference now to digital signal conversion method and the device of Figure 13 description according to the 5th embodiment of the present invention.

This digital signal transfer unit has: decoded portion 8, and the DV vision signal is used to decode; Conversion of resolution part 16 is used for handling carrying out conversion of resolution from the format conversion of the decoding of decoded portion 8 output; Difference part 7 is used for according to dynamic mode/static schema information, and whether difference carries out forward inter-frame difference coding to each the intended conversion module unit from 16 outputs of conversion of resolution part; With coded portion 9, be used for based on the difference result who distinguishes part 7 to encoding from the conversion of conversion of resolution part 16 outputs and exporting the MPEG vision signal, as shown in figure 13.

In the following description, adopt the digital video signal conversion equipment that constitutes by these parts.Certainly, various piece is carried out the processing according to each step of digital signal conversion method of the present invention.

In the DV vision signal in being input to this digital video signal conversion equipment, the mode flags (for example) as indication static schema/dynamic mode information is added to each DCT piece in advance.

In this digital video signal conversion equipment, difference part 7 is based on this mode flags, and whether difference carries out forward inter-frame difference coding to each the intended conversion module unit from 16 outputs of conversion of resolution part.This computing will be described in detail later.

Separate framing part 11 and extract the mode flags of indication static schema/dynamic mode information, this mode flags is supplied with difference part 7.

Separate and shuffle part 15 cancellation and shuffle, shuffling is in order to be a unit with the amount of information unification in the video-frequency band, and the length that is used for DV coding side is fixed.

Difference part 7 comprises adder 27 and I (I picture)/P (P picture) difference and determining section 28.Adder 27 will be added to conversion of resolution output as the reference DCT coefficient that is stored in the negative DCT coefficient in frame memory (FM) part 24, and frame memory (FM) part 24 will be described later.From adder 27 output and be provided to I/P difference and determining section 28, I/P distinguishes and determining section 28 also has been provided mode flags, this mode flags has been indicated static schema/dynamic mode of explaining framing part 11 by oneself.

Describe the computing of I/P difference and determining section 28 now in detail.From the conversion of conversion of resolution part 16 outputs 8 * 8 DCT coefficients are arranged, as a unit.Respectively have four DCT coefficient block of 8 * 8 DCT coefficients to be assigned to luminance signal, two DCT coefficient block are assigned to color difference signal, have always constituted a predetermined block by six DCT coefficient block like this.This predetermined block is called as macro block.

Simultaneously, the difference of supposition of P picture and former frame is asked for simply.Under the situation of rest image, reduced amount of information when poor asking for.Yet, under the situation of dynamic image, increased amount of information when poor asking for.Thereby, similarly be dynamic if publish picture from the mode flags difference of indication static schema/dynamic mode, then macro block is used as the I picture, so that do not increase amount of information.If it similarly is static that difference is published picture, can carry out efficient coding to make the P picture by asking for difference.

When from separate that framing partly sends here about all mode flags indication dynamic modes of six DCT coefficient block the time, I/P difference and determining section 28 are macro block use I picture.On the other hand, when the sign of indication dynamic mode only can be detected in one of six DCT coefficient block, I/P difference and determining section 28 were used the P picture for macro block.

If the sign of dynamic mode is added to four or more DCT coefficient block of six DCT coefficient block, can be with the I picture as macro block.Equally, when the sign of indication static schema is added to all six DCT coefficient block, can be with the P picture as macro block.

Be provided for coded portion 9 by I/P difference and determining section 28 based on the DCT coefficient that macro block is defined as the I/P picture.

Coded portion 9 has weighting (W) part 18, quantizes (Q) part 19, re-quantization (IQ) part 26, contrary weighting (IW) part 25, FM part 24, variable length code (VLC) part 20, buffer storage 23, rate controlled part 29.

The DCT coefficient that the conversion that 18 pairs of conducts of weighting (W) part provide through difference part 7 from conversion portion 16 is exported is weighted.

Quantizing 19 pairs of DCT coefficients by 18 weightings of weighting (W) part of (Q) part quantizes.Then, 20 pairs of variable length code (VLC) parts are carried out variable length code by quantizing the DCT coefficients that (Q) part 19 quantizes, and the data of mpeg encoded are offered buffer storage 23.

Buffer storage 23 is the transfer rate of the data of mpeg encoded fixedly, and the data of mpeg encoded are exported as bit stream.Rate controlled part 29 is controlled at the increase or the minimizing of the amount of information of generation in quantifications (Q) part 19, i.e. quantization step according to such as the increase of the buffer capacity of buffer storage 23 or the change information of minimizing.

26 pairs of re-quantization (IQ) parts are carried out re-quantization from quantizing the DCT coefficients that (Q) part 19 quantizes, and the DCT coefficient of re-quantization is offered contrary weighting (IW) part 25.Contrary 25 pairs of DCT coefficients from re-quantization (IQ) part 26 of weighting (IW) part carry out contrary weighting, and this contrary weighting is the inverse operation of weighting.DCT coefficient by contrary weighting (IW) part 25 contrary weighted is stored in the FM part 24, as reference DCT coefficient.

As mentioned above, in digital video signal conversion equipment shown in Figure 13, difference part 7 is that each macro block is distinguished I picture or P picture by using I/P difference and determining section 28 according to from separating the mode flags of indication dynamic mode/static schema that framing part 11 sends here.Thereby, comprising singly that at first the DV signal of I picture can be converted into the mpeg picture that uses I picture or P picture, can be utilized as the improved advantage of the compression ratio of the feature of MPEG vision signal.

According to a sixth embodiment of the present digital signal conversion method and device are described now.

Digital video signal conversion equipment according to the 6th embodiment is a kind of digital video signal conversion equipment, and difference part 7 wherein shown in Figure 13 is replaced by difference part 30 shown in Figure 14.

Particularly, the digital video signal conversion equipment has: decoded portion 8 is used for the DV signal is carried out that partial decoding of h is handled and obtains signal such as the orthogonal transform domain of DCT coefficient; Conversion portion 16 is used to format conversion that the DCT coefficient from decoded portion 8 is carried out signal conversion processes; Difference part 30 is used for the maximum according to the absolute value of the frame-to-frame differences of changing output, and whether difference carries out forward inter-frame difference coding to each the intended conversion module unit from conversion portion 16 outputs; With coded portion 9, be used for based on the difference result who distinguishes part 30 encoding from the conversion of conversion portion 16 outputs and exporting the MPEG vision signal.

When being extracted as the DCT coefficient of the conversion of exporting from the conversion of conversion portion 16 with from the difference between the reference DCT coefficient of FM part 24, difference part 30 compares this maximum and predetermined threshold with reference to the maximum of the absolute value of AC coefficient.Difference part 30 result based on the comparison distributes to each macro block with the I/P picture.

Difference part 30 has poor calculating section 31, maximum test section 32, rating unit 33, I/P determining section 35.

Difference calculating section 31 calculates from the DCT coefficient of conversion portion 16 conversions and poor between the reference DCT coefficient of FM part 24.Be provided for maximum test section 32 from the difference of difference calculating section 31 outputs, also be provided for I/P determining section 35.

The maximum of the absolute value of the AC coefficient of maximum test section 32 detection difference outputs.Basically, when bulk information was converted into the DCT coefficient, it is big that the AC coefficient becomes.On the other hand, when low-volume information was converted into the DCT coefficient, the AC coefficient diminished.

Rating unit 33 will be compared with the predetermined threshold that provides from terminal 34 from the maximum of the absolute value of maximum test section 32.When this predetermined threshold was selected suitably, the amount of information that is converted to the DCT coefficient can be distinguished according to the maximum of the absolute value of AC coefficient.

I/P determining section 35 is by using the comparative result of rating unit 33, and difference comes DCT coefficient poor of autodyne calculating section 31, and promptly the difference of amount of information is greatly or little.When distinguishing out this difference when big, I/P determining section 35 is distributed to the I picture macro block that comprises from the DCT coefficient block of the conversion of conversion portion 16.When distinguishing out this poor hour, I/P determining section 35 is distributed to the P picture come the macro block of autodyne calculating section 31.

That is, if peaked absolute value greater than threshold value, is just distinguished containing much information of going on business, adopting the I picture is macro block.On the other hand, if peaked absolute value is little with regard to distinguishing the amount of information of going on business less than threshold value, employing P picture is a macro block.

Like this, the DV conversion of signals that can comprise the I picture according to the digital video signal conversion equipment of the 6th embodiment at first is for using the mpeg picture of I picture or P picture, can utilize the improved advantage as the compression ratio of the feature of mpeg signal vision signal.

In the digital video signal conversion equipment shown in Figure 13 and 14, the DV signal and the MPEG1 vision signal that meet the TSC-system formula are used separately as input and output.Yet this digital video signal conversion equipment also can be applied in each signal of pal mode.

Above-mentioned conversion of resolution is handled the conversion that can be applied to similarly from the DV form to the MPEG2 form.

Undertaken by conversion portion 16 because conversion of resolution is handled, mainly described the conversion of resolution of reduction above.Yet it also is possible amplifying.Particularly, usually,, can amplify resolution with any magnification ratio by high fdrequency component being added in the supplied with digital signal of frequency domain.

When the MPEG2 vision signal was applied to digital broadcast service, signal was classified according to performance (profile) (function)/rank (level) (resolution).For example, the adaptable situation of the amplification of resolution is that main performance/high-level (MP@HL) vision signal that is used for the digital HDTV of the U.S. is converted into the DV signal.

The processing of the 6th embodiment also can be carried out with software mode.

With reference now to digital signal conversion method and the device of Figure 15 description according to the 7th embodiment of the present invention.The parts of the structure identical with the foregoing description are denoted by the same reference numerals.

Rate controlled part 40 is according to the data volume in the quantizer number (Q_NO) of explaining framing part 11 by oneself and class-mark (Class) the control quantized segment 19.

Figure 16 is the digital signal conversion method that illustrates by the 7th embodiment, the DV vision signal is being converted to the basic process of setting the quantizer ratio in the MPEG vision signal for each macro block (MB) of every frame.

At first, at step S1, for each macro block obtains quantizer number (Q_NO) and class-mark (Class).This quantizer number (Q_NO) is expressed by 0 to 15 value, is common to all six the DCT pieces in the macro block.Class-mark (Class) is expressed by 0 to 3 value, for each of six DCT pieces provides.

Then, at step S2, be that each DCT piece calculates quantization parameter (q_param) according to following processes.

Quantization table q_table[4]=9,6,3,0}

Quantization parameter q_param=Q_NO+q_table[class]

Particularly, quantization table has four kinds of values (9,6,3,0), corresponds respectively to class- mark 0,1,2,3.For example, when class-mark is 2, when quantizer number is 8, drawn quantization parameter 11 mutually corresponding to the quantification tabular value 3 of class-mark 2 and quantizer numbers 8.

Then, at step S3, the mean value of the quantization parameter (q_param) of six DCT pieces in macro block is calculated.

Then, at step S4, the quantizer ratio (quantizer_scale) of MPEG macro block is obtained according to following processes, and processing finishes.

Quantization table q_table[25]=32,16,16,16,16,8,8,8,8,4,4,4,2,2,2,2,2,2,2,2,2,2,

2，2}

quantizer_scale＝q_table[q_param]

Particularly, quantization table has 25 kinds of values (32 to 2), corresponding to the quantization parameter that calculates in the above described manner.Quantization table corresponding to quantization parameter value 0 is 32.Quantization table corresponding to quantization parameter value 1 is 16.Quantization table corresponding to quantization parameter value 5 is 8.For example, when the mean value of the quantization parameter of obtaining in the above described manner is 10, become the quantizer ratio value corresponding to the value 4 of quantization parameter value 10.By this process, the MPEG quantizer ratio (quantizer_scale) of dependence target (target) speed is calculated based on the quantization parameter (q_param) of each macro block in every frame.Relation between corresponding relation between class-mark and the quantization table and quantization parameter and the quantizer ratio is obtained by rule of thumb.

In digital signal transfer unit of the present invention shown in Figure 15, above-mentioned processing by rate controlled part 40 based on from separating quantizer number (Q_NO) that framing part 11 sends here and class-mark (Class) and carry out.

Figure 17 illustrates and is used for by using the quantizer ratio of setting according to said process feedback to be applied to the basic process of next frame.

At first, at step S11,, set the number of the target bit of every frame with the bit rate of setting according to said process.

Then, at step S12, the total number of the position that every frame produced is added up to.

Then, at step S13, the number of target bit and produced the position total number between poor (diff) calculated.

Then, at step S14, adjust the quantizer ratio according to result calculated.

Calculation expression in per step is as follows:

diff＝cont*diff(cont：constant)

q_param＝q_param±f(diff)

quantizer_scale＝q_table[q_param]

Particularly, will carry out normalization at the difference diff multiplication by constants cont that step S13 obtains.Normalized difference multiply by the function that experience is obtained, and is added to quantization parameter or deducts from quantization parameter.End value is used as quantization parameter.Value corresponding to this quantization parameter value is selected from the quantization table that 25 kinds of values are arranged, as the quantizer ratio of next frame.

By aforementioned process,, carry out the feedback between frame by calculate the new quantizer ratio of new quantizer ratio (quantizer_scale) and use next frame based on the quantization parameter of adjusting (q_param).

Digital signal conversion method and digital signal transfer unit according to the 8th embodiment of the present invention are described now.Although the DV form is converted into mpeg format in the foregoing embodiments, in the following embodiments, mpeg format is converted into the DV form.

With reference to Figure 18, the traditional device that is used for mpeg format is converted to the DV form is described at first.

Digital video signal conversion equipment shown in Figure 180 is made of mpeg decoder 70 and DV encoder 80.The mpeg decoder 70 MPEG video data that is used to decode, DV encoder 80 is used to export the DV video data.

In mpeg decoder 70, provide the parser 71 of the bit stream of MPEG2 video data, detection will offer length-changeable decoding (VLD) part 72 by the quantization DCT coefficient of variable length code coding according to the head of the bit stream of the quantization DCT coefficient of MPEG2 form framing.Equally, parser 71 is extracted motion vector (mv), and the motion vector that extracts is offered motion compensation (MC) part 77.

72 pairs of length-changeable decoding (VLD) parts are carried out length-changeable decoding by the quantization DCT coefficients of variable length code coding, and the result of length-changeable decoding is offered re-quantization (IQ) part 73.

Re-quantization part 73 is carried out re-quantization by being multiply by the quantization step that uses in the coding side by the quantization DCT coefficient of length-changeable decoding part 72 decodings.Like this, re-quantization part 73 obtains the DCT coefficient, and the DCT coefficient is offered inverse discrete cosine transform (IDCT) part 74.

74 pairs of DCT coefficients from re-quantization part 73 of inverse discrete cosine transform part are carried out inverse DCT, like this DCT coefficient are reverted to the data of spatial domain, i.e. pixel data.Particularly, by inverse DCT, to each piece calculating pixel value of comprising 8 * 8 pixels (brightness Y and aberration Cr, Cb).Under the situation of I picture, pixel value is its actual pixel value.Yet under the situation of P picture and B picture, pixel value is the difference between the corresponding pixel value.

Motion compensation portion 77 is stored in the image information among two frame memory FM of frame memory (FM) part 76 and is extracted by parser 71 by use motion vector mv produces motion compensation output, and this motion compensation output is offered adder 75.

Adder 75 is exported the difference that is added to from inverse discrete cosine transform part 74 with motion compensation, the decoded pictures data is offered discrete cosine transform (DCT) part 81 and the frame memory part 76 of DV encoder 80.

In DV encoder 80, the 81 pairs of decoded pictures data of discrete cosine transform part are carried out DCT and are handled, so that once more the decoded pictures data are converted to the data of orthogonal transform domain, that is, and the DCT coefficient, and this DCT coefficient offered quantification (Q) part 82.

Quantized segment 82 is considered visual characteristics, comes quantization DCT coefficient by using matrix table, and quantized result is offered variable length code (VLC) part 83 as the I picture of DV form.

Variable length code part 83 is handled the I picture that compresses the DV form by carrying out variable length code, and the I picture that compresses is offered framing part 84.

Framing part 84 will have been carried out the DV formatted data framing that variable length code is handled, and the bit stream of output DV video data.

Simultaneously, need a large amount of calculating usually, thereby the problem of bringing is that the format conversion of aforesaid video data can not be carried out efficiently such as the orthogonal transform of discrete cosine transform (DCT) and its inverse transformation.Because mistake is accumulated with the increase of amount of calculation, Signal Degrade also is a problem.

Like this, in order to address these problems, will digital video signal conversion equipment according to the 8th embodiment of the present invention be described with reference to Figure 19.

In digital signal transfer unit shown in Figure 19, the above-mentioned MPEG vision signal that meets mpeg format is imported as first digital signal, and the DV signal is exported as second digital signal.

Parser 111 references are extracted the movable information such as the image of motion vector mv and quantizer ratio as the head of the MPEG vision signal of the digital signal of first form.

Motion vector mv is sent to motion compensation (MC) part 115, and wherein motion compensation is carried out.Quantizer ratio (quantizer_scale) is fed to evaluation portion 123, and it will be described later.

The bit stream of 112 pairs of MPEG vision signals of length-changeable decoding (VLD) part carries out length-changeable decoding, and necessary information is extracted from this MPEG vision signal by parser 111.

113 pairs of MPEG vision signal re-quantizations of re-quantization (IQ) part by 112 decodings of length-changeable decoding part.

Then, the MPEG vision signal by re-quantization part 113 re-quantizations is transfused to adder 125.Motion compensated result from the motion vector mv of parser 111 also is input to this adder 125 from motion compensation portion 115.

Be fed to conversion of signals part 116 from the output of adder 125, it will be described later.This output also is input to motion compensation portion 115 by frame memory 114.116 pairs of conversion of signals parts are passed through the signal conversion processes of the vision signal of adder 125 inputs at orthogonal transform domain (frequency domain) execution such as the needs of conversion of resolution.

To shuffle by shuffling part 117 by the vision signal that conversion of signals part 116 is carried out the signal conversion processes that needs, be sent to buffer 118 and classified part 122 then.

The vision signal of delivering to buffer 118 is sent to quantification (Q) part 119, is quantized by this quantized segment 119.Then, this vision signal is by 120 variable length codes of variable length code (VLC) part.In addition, this vision signal is by framing part 121 framing, as the bit stream output of DV vision signal.

On the other hand, 122 pairs of classified parts are by shuffling the vision signal classification that part 117 is shuffled, and give evaluation portion 123 with sorting result as classified information.

Evaluation portion 123 is based on determining quantification number at quantized segment 119 from the classified information of classified part 122 with from the quantizer ratio (quantizer_scale) of parser 111.

For this structure, owing to can determine based on the data amount information that is included in as in the MPEG vision signal of the vision signal input of first form, be used for determining that the processing of data volume of the vision signal of second form that produced by conversion of signals can be simplified as the data volume of the DV vision signal of the vision signal of second form output.

It is MPEG1 vision signal and another is the situation of MPEG2 vision signal that above-mentioned the 7th and the 8th embodiment also can be applied to one of the digital signal of first form and digital signal of second form.

With reference now to digital signal conversion method and the digital signal transfer unit of Figure 20 description according to the 9th embodiment of the present invention.

This digital signal transfer unit is a kind of device, and the MPEG vision signal that is used for meeting the MPEG2 form is converted to the DV vision signal that meets the DV form.Suppose that these data are data of pal mode.

Be that the signal that meets MPEG2 form and DV form has the resolution of 720 * 576 pixels under the situation of signal of pal mode in vision signal, the ratio of the sample frequency of the sample frequency of luminance signal and two color difference signals equals 4: 2: 0.Thereby, needn't carry out the conversion of resolution processing for Y-signal or C signal.

In Figure 20, mpeg decoder 100 has parser 111, length-changeable decoding (VLD) part 112, re-quantization (IQ) part 113, adder 125, inverse discrete cosine transform (IDCT) part 131, frame memory (FM) part 132, motion compensation (MC) part 115, and discrete cosine transform (DCT) part 130.The formation of frame memory (FM) part 132 makes it be used as two forecast memories.

As what will be described in detail later, 131 pairs of I picture and P pictures by length-changeable decoding part 112 and re-quantization part 113 partial decoding of h of inverse discrete cosine transform part carry out the inverse discrete cosine transform processing.Motion compensation portion 115 is exported based on inverse discrete cosine transform and is produced motion compensation output.Discrete cosine transform is carried out in 130 pairs of motion compensation outputs of discrete cosine transform part.Adder 125 will be added to P picture and the B picture by length-changeable decoding part 112 and re-quantization part 113 partial decoding of h from the motion compensation output of discrete cosine transform part 130.

Hereinafter whole computing will be described.At first, parser 111 is with reference to the head of the MPEG2 video data of importing as bit stream, and the quantization DCT coefficient that will meet MPEG2 form framing reverts to variable-length codes, and this variable-length codes is offered length-changeable decoding part 112.Equally, parser 111 is extracted motion vector mv, and the motion vector that extracts is offered motion compensation portion 115.

The quantization DCT coefficient that 112 pairs of length-changeable decoding parts revert to variable-length codes carries out length-changeable decoding, and the result of length-changeable decoding is offered re-quantization part 113.

Re-quantization part 113 is carried out re-quantization and is handled by being multiply by the quantization step that uses in the coding side by the quantization DCT coefficient of length-changeable decoding part 112 decodings.Like this, re-quantization part 113 obtains the DCT coefficient, and this DCT coefficient is offered adder 125.The DCT coefficient that is obtained by length-changeable decoding part 112 and re-quantization part 113 is provided for adder 125 as output, and it is not reverted to pixel data by inverse discrete cosine transform, that is, and and as the partial decoding of h data.

Adder 125 also is provided with the motion compensation output from motion compensation portion 115, and it is by 130 orthogonal transforms of discrete cosine transform part.Then, adder 125 is added to motion compensation output the data of partial decoding of h in orthogonal transform domain.Adder 125 offers DV encoder 110 with addition output, also offers inverse discrete cosine transform part 131.

I picture in 131 pairs of addition outputs of inverse discrete cosine transform part or P picture are carried out inverse discrete cosine transform and are handled, thereby produce the data of spatial domain.The data of this spatial domain are the reference picture data of using for motion compensation.Should be stored in the frame memory part 132 for the reference picture data that motion compensation is used.

Motion compensation portion 115 produces motion compensation output by the motion vector mv that use is stored in the reference picture data in the frame memory part 132 and is extracted by parser 111, and this motion compensation output is offered discrete cosine transform part 130.

Discrete cosine transform part 130 will return to aforesaid orthogonal transform domain in the motion compensation output that spatial domain is handled, and then this motion compensation output be offered adder 125.

Adder 125 will be added to the DCT coefficient from the differential signal of the P of the partial decoding of h of re-quantization part 113 and B picture from the DCT coefficient that the motion compensation of discrete cosine transform part 130 is exported.Then, the data that are used as at the partial decoding of h of orthogonal transform domain from the addition of adder 125 output offer DV encoder 110 and inverse discrete cosine transform part 131.

Because the I picture from the partial decoding of h of re-quantization part 113 is the in-frame encoding picture signal, do not need the motion compensation addition to handle.The I picture of partial decoding of h is offered inverse discrete cosine transform part 131 by former state, also offers DV encoder 110.

DV encoder 110 comprises quantification (Q) part 141, variable length code (VLC) part 142 and framing part 143.

Quantized segment 141 quantizes the decoding output from the I picture in the orthogonal transform domain of mpeg decoder 100, P picture and B picture, and promptly the DCT coefficient offers variable length code part 142 with the DCT coefficient that quantizes.

The DCT coefficient of 142 pairs of quantifications of variable length code part carries out variable length code to be handled, and coded data is offered framing part 143.143 pairs of data framings of framing part from the compressed encoding of variable length code part 142, the bit stream of output DV video data.

By this way, when the MPEG2 video data that will be converted is the I picture, mpeg decoder 100 makes length-changeable decoding part 112 and re-quantization part 113 partial decoding of h MPEG2 video datas to orthogonal transform domain, and DV encoder 110 makes quantized segment 141 and variable length code part 142 part coding video frequency datas.Simultaneously, mpeg decoder 100 makes the 131 pairs of I pictures of inverse discrete cosine transform part carry out inverse discrete cosine transform, and result's I picture is stored in the frame memory part 132 reference picture as the P/B picture.

On the other hand, when the MPEG2 video data that will be converted is P picture or B picture, only being useful on the processing that produces motion compensation output carries out in spatial domain by using inverse discrete cosine transform part 131, be used to constitute except as being undertaken by using discrete cosine transform part 130 at discrete cosine transform domain, as mentioned above by the processing of the frame of the differential signal of the P picture of length-changeable decoding part 112 and re-quantization part 113 partial decoding of h or B picture.After this, carry out the part coding by DV encoder 110.

Especially, under the situation of P picture, the macro block in the position of being indicated by motion vector mv by from the I picture of being handled by inverse discrete cosine transform part 131 inverse discrete cosine transforms, extracts by the motion compensation process of motion compensation portion 115.Carry out discrete cosine transform by 130 pairs of these macro blocks of discrete cosine transform part and handle, be added to the differential signal of the DCT coefficient of P picture by adder 25 again as discrete cosine transform domain.This processing is based on, and the result of the discrete cosine transform that the addition result of spatial domain is carried out is equivalent to the result of the data addition of being handled by discrete cosine transform.This result is encoded by DV encoder 110 parts.Simultaneously, as the reference of next B picture, carry out inverse discrete cosine transform by 131 pairs of addition outputs from adder 125 of inverse discrete cosine transform part, result's data are stored in the frame memory part 132.

Under the situation of B picture, the macro block in the position of being indicated by motion vector mv is extracted from the P picture of being handled by inverse discrete cosine transform part 131 inverse discrete cosine transforms.Then, carry out discrete cosine transform by 130 pairs of these macro blocks of discrete cosine transform part and handle, in discrete cosine transform domain, be added on it as the DCT coefficient of the B picture of differential signal.Under two-way situation, from two reference frames, extract macro block, its mean value is used.

Its result is encoded by DV encoder 110 parts.Because the B picture does not become reference frame, does not need to carry out inverse discrete cosine transform by inverse discrete cosine transform part 131.

Handle for the I picture of decoding needs inverse discrete cosine transform (IDCT) and discrete cosine transform (DCT) traditionally, and only needs IDCT is for referencial use according to the digital video signal conversion equipment of the 9th above-mentioned embodiment.

For the P picture of decoding, it is essential that DCT for referencial use and IDCT handle.Yet,, and, do not need IDCT according to the digital video signal conversion equipment of this embodiment needs DCT only for decoding B picture needs DCT and IDCT traditionally.

Under typical MPEG2 data conditions, comprise an I picture, four P pictures and 10 B pictures with GOP number N=15, forward prediction picture spacing M=3.The amount of calculation of supposing the amount of calculation of DCT and IDCT is equal substantially, and when weighting was omitted, per 15 frame MPEG2 data were expressed from the next under the situation of conventional art

2×DCT×(1/15)+2×DCT×(4/15)+2×DCT×(10/15)

＝2×DCT

Under the situation of digital video signal conversion equipment shown in Figure 20, be expressed from the next

1×DCT×(1/15)+2×DCT×(4/15)+1×DCT×(10/15)

＝1.2666×DCT

Like this, can significantly reduce amount of calculation.DCT represents amount of calculation in these equatioies.

That is, in digital video signal conversion equipment shown in Figure 20, the data computation treating capacity that is used for the format conversion from the MPEG2 video data to the DV video data can be reduced significantly.

With reference now to the digital video signal conversion equipment of Figure 21 description according to the of the present invention ten embodiment.

In the tenth embodiment, adopted the MPEG video data that is used for to meet the MPEG2 form to be converted to the digital video signal conversion equipment of the DV video data that meets the DV form.Yet, suppose that the MPEG2 video data is for example compressed video signal of 1440 * 1080 pixels of high-resolution.

For example, when the MPEG2 vision signal was applied to digital broadcast service, signal was classified according to performance (function)/rank (resolution).Main performance/high-level (MP@HL) vision signal that is used for the digital HDTV of the U.S. has high-resolution, and as mentioned above, this signal is converted into the DV video data.

Thereby the structure of digital video signal conversion equipment shown in Figure 21 is, provides the conversion of signals part 140 that is used to carry out above-mentioned conversion process between the mpeg decoder 100 of Figure 20 and DV encoder 110.

This conversion of signals part 140 is by using the transformation matrix that produces based on inverse orthogonal transformation matrix and orthogonal transform matrix, the DCT coefficient from the dct transform domain of mpeg decoder carried out conversion of resolution handle.Described inverse orthogonal transformation matrix is corresponding to the orthogonal transform matrix that is used for the mpeg encoded data are carried out the DCT coding, and described orthogonal transform matrix is corresponding to the inverse orthogonal transformation matrix that is used to acquisition at the IDCT of the conversion of signals output signal of time domain coding.

The DCT coefficient as conversion of resolution output from this conversion of signals part 140 is provided for DV encoder 110.

110 pairs of DV encoders quantize and variable length code as the DCT coefficients of conversion of resolution output, framing DCT coefficient then, the bit stream of output DV video data.

Like this, in this digital video signal conversion equipment, the main performance in the MPEG vision signal/high-level (MP@HL) vision signal is by conversion of signals part 140 conversion of resolution, then by the DV encoder encodes to produce the DV video data.

Be similar to the digital video signal conversion equipment of Figure 20, about the I picture, the digital video signal conversion equipment of the tenth embodiment only needs IDCT for referencial use, and IDCT and DCT handle and need traditionally.

About the P picture, in conventional art, it is for referencial use to carry out DCT and IDCT.About the B picture, this digital video signal conversion equipment only needs DCT and does not need IDCT, and IDCT and DCT need traditionally.

That is, in digital video signal conversion equipment shown in Figure 21, the data computation treating capacity that is used for from high-resolution MPEG2 video data to the format conversion of DV video data also can be reduced significantly.

Handle as the conversion of resolution of being undertaken, mainly described the conversion of resolution of reduction by conversion of signals part 140.Yet it also is possible amplifying.Specifically, usually,, can amplify resolution with any magnification ratio by high fdrequency component being added in the supplied with digital signal of frequency domain.For example, carry out format conversion from the MPEG1 vision signal to the DV vision signal.

Above-mentioned processing also can be undertaken by software.

Simultaneously, in the compressibility of above-mentioned mpeg format or DV form,, can adopt the MIXED COMPRESSION CODING method of use in conjunction with the orthogonal transform coding of predictive coding for effective compressed encoding Still image data or dynamic image data.

Input information signal by MIXED COMPRESSION CODING method compressed encoding is being carried out after conversion of resolution handles, when carrying out orthogonal transform once more and during with the predictive coding of motion compensation, must estimated motion vector when carrying out the step that predictive coding again handles.

If carry out predictive coding once more and do not carry out the conversion of resolution processing with identical resolution, when predictive coding, can use motion vector.Yet if resolution is converted, slewing distortion is changed.Thereby the motion vector that uses in predictive coding step more also is changed.

Like this, when predictive coding step again, need estimated motion vector.Yet estimation of motion vectors needs the arithmetic processing amount.

In order to eliminate this problem, used digital signal transfer unit according to the 11 embodiment.In digital signal conversion method and device according to the 11 embodiment, by the input information signal of using use in conjunction with the MIXED COMPRESSION CODING compressed encoding of the orthogonal transform coding of predictive coding, handle by signal conversion processes in time domain or orthogonal transform domain such as conversion of resolution, return to orthogonal transform domain then and be used for the recompression coding, perhaps at the orthogonal transform domain compressed encoding.

The example of above-mentioned MIXED COMPRESSION CODING be by ITU-T (International Telecommunications Union-telecommunication standardization portion) recommend H.261 and H.263 and MPEG and DV coding standard.

H.261 standard is the image encoding standard of low bitrate, and its development mainly is for teleconference and video telephone by ISDN.H.263 be modified version H.261 for the GSTN video-telephone system.

With reference now to Figure 22, the 11 embodiment described, in the digital video signal conversion equipment of the 11 embodiment, the mpeg encoded data that meet mpeg format are transfused to, and are handled by the conversion of resolution as signal conversion processes, and the mpeg encoded signal of conversion of resolution is output.

This digital video signal conversion equipment has: decoded portion 210 is used for using motion compensation MC to decode to the bit stream of the mpeg encoded data that detect compressed encoding with motion vector (mv); Conversion of resolution part 160 is used for handling carrying out conversion of resolution from the decoding output of decoded portion 210; With coded portion 220, be used for motion detection based on the motion vector mv of the data that are added to mpeg encoded, compressed encoding is handled and the bit stream of the video data encoder of output resolution ratio conversion to carrying out from the conversion output image of conversion of resolution part 160, as shown in figure 22.

The digital video signal conversion equipment that is made of these parts will be described below.Certainly each member carries out the processing according to each step of digital signal conversion method of the present invention.

Decoded portion 210 comprises length-changeable decoding (VLD) part 112, re-quantization (IQ) part 113, inverse discrete cosine transform (IDCT) part 150, adder 151, motion compensation (MC) part 152 and frame memory (FM) part 153.FM part 153 is made of two frame memory FM, as forecast memory.

112 pairs of mpeg encoded data of VLD part are promptly decoded according to variable length code by motion vector with as the coded data that the variable length code of the quantization DCT coefficient of additional information obtains, and are extracted motion vector mv.IQ part 113 is carried out re-quantization and is handled by being multiply by the quantization step that uses in the coding side by the quantization DCT coefficient of VLD part 112 decodings.

150 pairs of IDCT parts are carried out inverse DCTs from the DCT coefficients of IQ part 113, thereby the DCT coefficient is returned to the data of spatial domain, i.e. pixel data.Particularly, by inverse DCT, to each piece that comprises 8 * 8 pixels calculate corresponding pixel value (brightness Y and aberration Cr, Cb).Under the situation of I picture, pixel value is its actual pixel value.Yet under the situation of P picture and B picture, pixel value is the difference between the corresponding pixel value.

MC part 152 is carried out motion compensation process by using the motion vector mv that is extracted by VLD part 112 to the image information in two frame memories that are stored in FM part 153, and this motion compensation output is offered adder 151.

Adder 151 will be added to the difference from IDCT part 150 from the motion compensation output of MC part 152, thus the picture signal of output decoder.The conversion of resolution that 160 pairs of decoded image signal of conversion of resolution part needing to carry out is handled.Conversion output from conversion of resolution part 160 is provided for coded portion 220.

Coded portion 220 comprises ratio conversion portion 171, estimation (ME) part 172, adder 173, DCT part 175, rate controlled part 183, quantize (Q) part 176, variable length code (VLC) part 177, buffer storage 178, IQ part 179, IDCT part 180, adder 181, FM part 182 and MC part 174.

Ratio conversion portion 171 carries out the ratio conversion according to the conversion of resolution speed of being used by conversion of resolution part 160 to the motion vector mv that is extracted by VLD part 112.For example, if the conversion of resolution speed of being used by conversion of resolution part 160 is 1/2, motion vector mv is switched to 1/2 ratio.

ME part 172 is by using the ratio transitional information from ratio conversion portion 171, and search is from the close limit of the conversion output of conversion of resolution part 160, thus the optimum movement vector of the resolution of estimation conversion.

Use when carrying out motion compensation by ME part 172 estimated movement vector by MC part 174.The conversion output image that is used for the estimation of motion vector by ME part 172 from conversion of resolution part 160 is provided for adder 173.

Adder 173 is calculated the reference picture described later and poor between the conversion output of conversion of resolution part 160, and this difference is offered DCT part 175.

DCT part 175 is by using the piece of 8 * 8 sizes, to carrying out discrete cosine transform by MC part 174 by the reference picture of motion compensation acquisition and the difference between the conversion output picture.About the I picture,, directly carry out the DCT arithmetical operation, and do not calculate the poor of interframe owing to carry out interframe encode.

Quantize (Q) part 176 and consider visible characteristic, the DCT coefficient from DCT part 175 is quantized by using matrix table.VLC part 177 is by using variable length code, and the DCT coefficient from the quantification of Q part 176 is compressed.

Buffer storage 178 is memories, is used to keep the constant bit rate of coded data, and this coded data is compressed by the 177 usefulness variable length codes of VLC part.From this buffer storage 178, the video data encoder of conversion of resolution is used as the bit stream output of constant bit rate.

Rate controlled part 183 is controlled the increase/reduction of the amount of information that produces in the Q part 176, i.e. quantization step according to the relevant change information of the increase/reduction of the buffer capacity of buffer storage 178.

IQ part 179 and IDCT part 180 constitute the local decode part together.IQ part 179 re-quantizations offer IDCT part 180 from the quantization DCT coefficient of Q part 176 with the DCT coefficient.180 pairs of DCT coefficients from IQ part 179 of IDCT part carry out inverse DCT, to recover pixel data, pixel data are offered adder 181.

Adder 181 will be added to the pixel data of exporting as from the inverse DCT of IDCT part 180 from the motion compensation output of MC part 174.Be provided for FM part 182 as the image information of exporting from the addition of adder 181.The image information that is stored in the FM part 182 is used motion compensation process by MC part 174.

MC part 174 is carried out motion compensation by using the optimum movement vector of being estimated by ME part 172 to the image information that is stored in the FM part 182, will offer adder 173 as the motion compensation output of reference picture.

Adder 173 is calculated poor between the conversion output picture of conversion of resolution part 160 and the reference picture, this difference is offered DCT part 175, as mentioned above.

DCT part 175, Q part 176, VLC part 177 and the computing as mentioned above of buffering memory 178.Finally, the video data encoder of conversion of resolution be used as bit stream with constant bit rate from the output of this digital video signal conversion equipment.

In this digital video signal conversion equipment, when by ME part 172 estimated motion vectors of coded portion 220, append to the motion vector of the vision signal macro block of initial compression, change in proportion according to the conversion of resolution speed in the conversion of resolution part 160 by ratio conversion portion 171, based on ratio transitional information from ratio conversion portion 171, search is from the close limit of the conversion output picture of conversion of resolution part 160, so that be the motion-compensated estimation motion vector, the estimation of the motion vector when replacing lacking any information.Like this, because amount of calculation in ME part 172 can significantly be reduced, microminiaturization and the minimizing of conversion process time that can implement device.

The 12 embodiment described now.In this embodiment, also adopted the digital video signal conversion equipment that is used for the MPEG vision signal is carried out conversion of resolution processing and output resolution ratio video signal converted.

This digital video signal conversion equipment has: decoded portion 211, be used for the mpeg encoded data of carrying out above-mentioned hybrid coding are handled by only carrying out prediction decoding with MC, and obtain the decoded data of orthogonal transform domain; Conversion of resolution part 260 is used for handling carrying out conversion of resolution from the decoded data of the orthogonal transform domain of decoded portion 211; With coded portion 221, be used for by the motion detection of use based on the motion vector information of the data of mpeg encoded, along with motion compensated prediction, carry out compressed encoding and handle, as shown in figure 23 exporting from the conversion of conversion of resolution part 260.

The digital video signal conversion equipment that is made of these parts will be described below.Certainly each member carries out the processing according to each step of digital signal conversion method of the present invention.

In this digital video signal conversion equipment, to compare with device shown in Figure 22, IDCT part 150 is unnecessary in decoded portion 210, DCT part 175 and IDCT part 180 are unnecessary in coded portion 220.That is, in this digital video signal conversion equipment, the decoded data in DCT territory is carried out conversion of resolution handle, its conversion output is encoded.

Orthogonal transform and inverse orthogonal transformation such as DCT need a large amount of calculating usually.Thereby above-mentioned conversion of resolution can not be carried out effectively.Equally, owing to the increase along with amount of calculation, mistake is accumulated, and signal may deterioration.

Thereby in the digital video signal conversion equipment of Figure 23, the IDCT part 150 of Figure 22, DCT part 174 and IDCT part 180 are eliminated.The function of conversion of resolution part 160 is changed.

Equally, for in the DCT territory from from the conversion DCT coefficient calculations activity described later (activity) of conversion of resolution part 160 with by using this activity estimated motion vector, an activity calculating section 200 is used to replace the ratio conversion portion 171 of Figure 22.

Conversion of resolution part 260 shown in Figure 23 provides addition output (DCT coefficient), and this addition output (DCT coefficient) is to be obtained by the DCT coefficient that the quantization DCT coefficient of VLD part 212 decodings obtains by re-quantization by being added to from the motion compensation output of MC part 252 by adder 251 by IQ part 213.

This conversion of resolution part 260 is by using a transformation matrix, the DCT coefficient from the dct transform domain of decoded portion 211 carried out conversion of resolution handle.Described transformation matrix is based on corresponding to the inverse orthogonal transformation matrix of the orthogonal transform matrix of the DCT coding that is used for the data of mpeg encoded are carried out and produces corresponding to the orthogonal transform matrix of the inverse orthogonal transformation matrix of the IDCT coding that is used to the conversion of signals output signal that obtains time domain.

The DCT coefficient as conversion of resolution output from conversion of resolution part 260 is provided for activity calculating section 200.Activity calculating section 200 is each macro block computer memory activity from the luminance component from the DCT coefficient of conversion of resolution part 260.Particularly, the feature of the maximum value calculation image of the AC value by using the DCT coefficient.For example, still less the existence of high fdrequency component has indicated smoothed image.

ME part 272 is estimated the optimum movement vector in the resolution of conversion based on the activity of being calculated by activity calculating section 200.Particularly, ME part 272 is based on the activity of being calculated by activity calculating section 200, and the motion vector mv that conversion is extracted by VLD 212 is so that estimated motion vector mv offers MC part 274 with estimated movement vector mv.ME part 272 is at the orthogonal transform domain estimated motion vector.This estimation at orthogonal transform domain will be described subsequently.

DCT coefficient from the conversion of resolution of conversion of resolution part 260 is provided for adder 273 by activity calculating section 200 and ME part 272.

Adder 273 is calculated described later with reference to DCT coefficient and poor between the DCT coefficient of the conversion of conversion of resolution part 260, and this difference is offered quantification (Q) part 276.

Q part 276 quantizes difference (DCT coefficient), and the DCT coefficient that quantizes is offered VLC part 277 and IQ part 279.

Rate controlled part 283 is controlled at the increase/minimizing of the amount of information of generation in the Q part 276, i.e. quantization step according to the change information from the increase/minimizing of the buffer capacity of the activity information of activity calculating section 200 and relevant buffer storage 278.

VLC part 277 is by using variable length code, and compressed encoding offers buffer storage 278 from the quantization DCT coefficient of Q part 276 with the DCT coefficient that compresses.Buffer storage 278 keeps by the constant bit rate of VLC part 277 by the coded data of variable length code compression the video data encoder of conversion of resolution being exported as bit stream with constant bit rate.

279 pairs of quantization DCT coefficients from Q part 276 of IQ part carry out re-quantization, and the DCT coefficient is offered adder 281.Adder 281 will be added to from the DCT coefficient of IQ part 279 as re-quantization output from the motion compensation output of MC part 274.The DCT coefficient information of exporting as addition from adder 281 is provided for FM part 282.The DCT coefficient information that is stored in the FM part 282 is passed through motion compensation process by MC part 274.

MC part 274 is carried out motion compensation by using the optimum movement vector of being estimated by ME part 272 to the DCT coefficient information that is stored in the FM part 282, and motion compensation output is offered adder 281 as reference DCT coefficient.

Adder 273 is calculated from the DCT coefficient of the conversion of conversion of resolution part 260 and poor with reference between the DCT coefficient, this difference is offered Q part 276, as mentioned above.

Q part 276, VLC part 277 and the computing as mentioned above of buffering memory 278.Finally, the video data encoder of conversion of resolution is exported with constant bit rate from this digital video signal conversion equipment.

MC part 274 is similar to ME part 272 and carries out motion compensation at orthogonal transform domain by using optimum movement vector of being estimated by ME part 272 and the reference DCT coefficient that is stored in the FM part 282.

With reference now to Figure 24 to Figure 26, estimation and motion compensation in the orthogonal transform domain are described.In Figure 24, the macro block of the picture A that solid line indicates to compress, dotted line is represented the macro block of reference picture B.When using motion vector, making the picture A that must compress and reference picture B to overlap each other, as shown in figure 24, the border of macro block may not overlap.Under the situation of Figure 24, the macro block B ' that compress is partly at four macro block B of reference picture B ₁, B ₂, B ₃And B ₄Last stretching, extension.Thereby the macro block that does not have reference picture B can not obtain reference picture B at the DCT of the position of macro block B ' coefficient one by one corresponding to macro block B '.Thereby, must be by the conversion macro block B ' DCT coefficient of four macro blocks of the reference picture B that stretches of part thereon, with the DCT coefficient of the reference picture B that obtains the part that macro block B ' is positioned at.

Figure 25 schematically shows the process of this conversion process.Because the macro block B of reference picture B ₁Bottom left section overlap on the upper right portion of macro block B ', by conversion macro block B ₁The DCT coefficient produce macro block B ₁₃, as hereinafter described.Similarly, because the macro block B of reference picture B ₂Lower right-most portion overlap on the upper left of macro block B ', by conversion macro block B ₂The DCT coefficient produce macro block B ₂₄, as hereinafter described.Similarly handle macro block B ₃And B ₄Carry out, thereby produce macro block B ₃₁And B ₄₂By four macro block B in conjunction with such generation ₁₃, B ₂₄, B ₃₁And B ₄₂, can obtain the DCT coefficient of the reference picture B of the part that macro block B ' is positioned at.

In brief, this processing can be expressed by following equation (6) and (7).

B′＝B ₁₃+B ₂₄+B ₃₁+B ₄₂ …(6)

DCT(B′)＝DCT(B ₁₃)+DCT(B ₂₄)+DCT(B ₃₁)+DCT(B ₄₂) …(7)

With reference now to Figure 26, the conversion of the DCT coefficient of macro block is described.Figure 26 illustrates and is used for passing through from original block B in spatial domain ₄Deng calculating obtain part macro block B ₄₂Mathematical Modeling.Particularly, at upper left side B ₄Be extracted, insert 0, be displaced to the lower right side.That is, illustrate from piece B ₄The B that calculating by following equation (8) obtains ₄₂

B ₄₂＝H ₁×B ₄×H ₂

${\mathrm{<figure-callout\; id="1"\; label="H"\; filenames="A20051008944600461.png,A20051008944600471.png"\; state="\{\{state\}\}">H</figure-callout>}}_1=\left[\begin{array}{cc}0& 0\\ \mathrm{<figure-callout\; id="0"\; label="Ih"\; filenames="A20051008944600471.png,A20051008944600481.png"\; state="\{\{state\}\}">Ih</figure-callout>}& 0\end{array}\right],{\mathrm{<figure-callout\; id="2"\; label="H"\; filenames="A20051008944600471.png,A20051008944600481.png"\; state="\{\{state\}\}">H</figure-callout>}}_2=\left[\begin{array}{cc}0& \mathrm{<figure-callout\; id="0"\; label="Iw"\; filenames="A20051008944600471.png,A20051008944600481.png"\; state="\{\{state\}\}">Iw</figure-callout>}\\ 0& 0\end{array}\right]---\left(8\right)$

In this equation, Ih and Iw are from piece B ₄The identification code of the matrix that comprises the capable and h row of h that extracts and comprise the identification code of the matrix with w * w size of the capable and w row of w with h * h size.As shown in figure 26, at first and B ₄Synthetic first matrix (pre-matrix) H ₁, h row are extracted and are transformed into the end.For subsequently and B ₄Synthetic H ₂, a w is capable to be extracted and to be transformed into the right side.

Based on equation (8), B ₄₂The DCT coefficient can be according to following equation (9) directly from B ₄The DCT coefficient calculations.

DCT(B ₄₂)＝DCT(H ₁)×DCT(B ₄)×DCT(H ₂) …(9)

This equation is applied to all sub-pieces (subblocks), and summation is calculated.Like this, newly the DCT coefficient of piece B ' can be directly from original block B ₁To B ₄The DCT coefficient obtain, represented as following equation (10).

$\mathrm{DCT}\left(B^{\&prime;}\right)=\underset{i=1}{\overset{4}{\mathrm{\&Sigma;}}}\mathrm{DCT}\left(H_{i1}\right)\&times;\mathrm{DCT}\left(B_i\right)\&times;\mathrm{DCT}\left(H_{i2}\right)---\left(10\right)$

H _I1And H _I2The DCT coefficient can be calculated and be stored in the memory in advance so that constitute the table memory.Can carry out estimation and motion compensation by this way, even in orthogonal transform domain.

Then, at coded portion 221, when by ME part 272 estimated motion vectors, append to the motion vector of the vision signal macro block of initial compression, based on the activity of calculating by activity calculating section 200, estimate the estimation of the motion vector when replacing lacking any information from the close limit of the conversion of conversion of resolution part 260 output by search.

As mentioned above, in the decoded portion 211 of the digital video signal conversion equipment of this embodiment, data to mpeg encoded are handled with the prediction decoding of motion compensation, data to described mpeg encoded, comprise that the predictive coding of being with motion detection and the hybrid coding of orthogonal transform coding are performed, and promptly carry out re-quantization after length-changeable decoding.Then, carry out motion compensation, the decoded data in DCT territory is carried out conversion of resolution to obtain to remain on the decoded data in DCT territory.Thereby, can directly carry out conversion of resolution at orthogonal transform domain, decoding (inverse orthogonal transformation) is just unnecessary to time domain or spatial domain.Like this, calculating is simplified, and can realize the high-quality conversion of less mistake in computation.Even, in coded portion 221, when by ME part 272 estimated motion vectors, append to the motion vector of the vision signal macro block of initial compression, estimate the estimation of the motion vector when replacing lacking any information by the search close limit based on the activity of calculating from conversion of resolution output.Thereby, because the amount of calculation of ME part 272 can significantly be reduced, microminiaturization and the minimizing of conversion process time that can implement device.

The 13 embodiment described now.In this embodiment, also adopted and be used for the mpeg encoded data are carried out the signal conversion processes handled such as conversion of resolution and the digital video signal conversion equipment of output video coded data.

This digital video signal conversion equipment has: decoded portion 340, be used for the mpeg encoded data of carrying out above-mentioned hybrid coding are handled by carrying out partial decoding of h, and obtain the data of orthogonal transform domain; Conversion portion 343 is used for handling carrying out conversion of resolution from the data of the orthogonal transform domain of decoded portion 340; With coded portion 350, be used to add a motion vector based on the motion vector information of the data of mpeg encoded, handle carrying out compressed encoding, as shown in figure 27 from the conversion output of conversion portion 343.

Decoded portion 340 comprises VLD part 341, IQ part 342.These VLD parts 341 and IQ part 342 have the VLD part 112 that is similar to Figure 21 respectively and the structure of IQ part 113, and computing similarly.The characteristic of this decoded portion 340 is not carry out motion compensation.

Particularly,, carry out conversion of resolution, do not carry out motion compensation by 343 pairs of DCT coefficients of conversion portion as difference information about P picture and B picture.The DCT coefficient of the conversion that obtains by conversion of resolution is quantized by Q part 345, Q part 345 by rate controlled part 348 by rate controlled.The DCT coefficient is exported from buffering memory 347 with constant rate of speed then by 346 variable length codes of VLC part.

In this case, the motion vector conversion portion 344 of coded portion 350 refreshes ratio (rescale) according to conversion of resolution speed to the motion vector mv that is extracted by VLD part 341, and the motion vector that will refresh ratio offers VLC part 346.

The motion vector mv that VLC part 346 will refresh ratio is added to the quantization DCT coefficient from Q part 345, carries out variable length code and handles.VLC part 346 offers coded data buffer storage 347 then.

As mentioned above, in digital video signal conversion equipment shown in Figure 27, owing to do not carry out motion compensation in decoded portion 340 and coded portion 350, calculating can be simplified, and the burden of hardware can be reduced.

In above-mentioned digital video signal conversion equipment, can carry out rate transition.In brief, this digital video signal conversion equipment can be applied to the conversion of the transfer rate from 4Mbps to 2Mbps, and resolution does not change.

Although described the structure of device in the above-described embodiments, the digital signal conversion method of the application of the invention can be configured to software with related device.

According to the present invention, the input information signal of the compressed encoding of band motion detection with the decoding of motion compensation, is carried out signal conversion processes to the signal of decoding.To the signal of this conversion, carry out tape base and handle in the compressed encoding of the motion detection of the motion vector information of input information signal.When the conversion of resolution processing as this signal conversion processes is employed, the signal of conversion to be handled with the compressed encoding of motion compensation, this motion compensation is based on by handle the information that ratio converting motion Vector Message obtains according to conversion of resolution.Especially, the motion vector information that needs when compressed encoding is changed in proportion according to conversion of resolution speed, the search close limit.Thereby the amount of calculation when estimation of motion vectors can significantly be reduced, miniaturization and the minimizing of conversion process time that can implement device.

Equally,, the input information signal of having carried out the compressed encoding that comprises the predictive coding of being with motion detection and orthogonal transform coding is carried out partial decoding of h, thereby obtain the decoded signal of orthogonal transform domain according to the present invention.Then, the decoded signal to orthogonal transform domain carries out signal conversion processes.To the signal of this conversion, to handle with the compressed encoding of motion compensated prediction, the motion detection based on the motion vector information of input information signal is used in this motion compensation.When the conversion of resolution processing is used as this signal conversion processes application, signal to conversion is handled with the compressed encoding of motion compensation, and this motion compensation is based on the information by obtaining according to the activity converting motion Vector Message that obtains from the conversion of resolution processing.Thereby, can estimate the motion vector information that needs when the compressed encoding by the search close limit, amount of calculation can significantly be reduced.Like this, miniaturization and the minimizing of conversion process time that can implement device.Equally,, do not need inverse orthogonal transformation to handle, do not need decoding (inverse orthogonal transformation) to time domain or spatial domain owing to can carry out signal conversion processes at orthogonal transform domain.Thereby calculating is simplified, and can carry out the high-quality conversion of less mistake in computation.

In addition,, the input information signal of having carried out the compressed encoding that comprises the predictive coding of being with motion detection and orthogonal transform coding is carried out partial decoding of h, thereby obtain the decoded signal of orthogonal transform domain according to the present invention.Then, the decoded signal to orthogonal transform domain carries out signal conversion processes.To the signal of this conversion,, carry out compressed encoding and handle by adding based on the motion vector information of the motion vector information conversion of input information signal.Thereby, when the conversion of resolution processing is used as this signal conversion processes application,, the signal of changing is carried out the compressed encoding processing by adding the information of passing ratio conversion according to the motion vector information acquisition of conversion of resolution processing.

That is, owing to can estimate the motion vector information that adds when the compressed encoding by the search close limit, the amount of calculation when estimation of motion vectors can significantly be reduced.Equally, owing to can carry out signal conversion processes, do not need inverse orthogonal transformation to handle at orthogonal transform domain.In addition and since decoding and when coding without motion compensation process, can further reduce amount of calculation.

Claims (8)

1. digital signal conversion method comprises:

The data extract step is extracted a part of orthogonal transform coefficient from the relevant block of the digital signal of first form of the orthogonal transform coefficient piece that comprises scheduled unit, thus the component part piece;

The inverse orthogonal transformation step on the basis of described part piece, is carried out inverse orthogonal transformation to the orthogonal transform coefficient that constitutes each part piece;

Part piece Connection Step connects the described part piece of being handled by inverse orthogonal transformation, thereby constitutes the new piece of described scheduled unit; With

The orthogonal transform step on described basis, is carried out orthogonal transform to described new, thereby produces second digital signal of the described new orthogonal transform block that comprises described scheduled unit.

2. digital signal conversion method as claimed in claim 1, wherein said orthogonal transform is discrete cosine transform, the digital signal of described first form is to use the vision signal of variable length code with pre-determined constant speed compressed encoding, and the digital signal of described second form is the vision signal with the variable bit rate compressed encoding.

3. digital signal conversion method as claimed in claim 1, wherein in described data extract step, at the discrete cosine transform coefficient of lower frequency side is to extract from the relevant block of the digital signal of described first form, and the number of the number of the discrete cosine transform coefficient of the number of the discrete cosine transform coefficient of the horizontal component of luminance signal, the horizontal component of color difference signal and the discrete cosine transform coefficient of its vertical component is reduced.

4. digital signal conversion method as claimed in claim 1, wherein when a frame of the digital signal of described first form is made of two frames,

In described data extract step, be used to separate the separating of discrete cosine transform coefficient of the discrete cosine transform coefficient of row of the odd field that constitutes described frame and the row of the even field that constitutes described frame, generation comprise these one of the piece of discrete cosine transform coefficient.

5. digital signal conversion method as claimed in claim 1, the digital signal of wherein said first form is the compressed video signal that has the ratio of the sample frequency of the sample frequency of the resolution of 720 * 480 pixels and luminance signal and color difference signal to equal 4: 1: 1, and the digital signal of described second form is the compressed video signal that has the ratio of the sample frequency of the sample frequency of the resolution of 360 * 240 pixels and luminance signal and color difference signal to equal 4: 2: 0.

6. digital signal conversion method as claimed in claim 1, the digital signal of wherein said first form is the compressed video signal that has the ratio of the sample frequency of the sample frequency of the resolution of 720 * 480 pixels and luminance signal and color difference signal to equal 4: 2: 0, and the digital signal of described second form is the compressed video signal that has the ratio of the sample frequency of the sample frequency of the resolution of 360 * 240 pixels and luminance signal and color difference signal to equal 4: 2: 0.

7. digital signal conversion method as claimed in claim 1, wherein in described data extract step, in the orthogonal transform coefficient of described lower frequency side is to extract from the relevant block of the digital signal of described first form, and the number of the discrete cosine transform coefficient of the vertical component of color difference signal is reduced to 1/2.

8. digital signal conversion method as claimed in claim 7, the digital signal of wherein said first form is the compressed video signal that has the ratio of the sample frequency of the sample frequency of the resolution of 720 * 480 pixels and luminance signal and color difference signal to equal 4: 1: 1, and the digital signal of described second form is the compressed video signal that has the ratio of the sample frequency of the sample frequency of the resolution of 720 * 480 pixels and luminance signal and color difference signal to equal 4: 2: 0.

Images