JP4636547B2 - Encoding device, program, and information recording medium - Google Patents

Description

  The present invention relates to a coding technique for signals having different frequency bands in the horizontal direction and the vertical direction, and more specifically to a coding technique for interlaced video signals.

  In a moving image photographing device such as a video camera, generally, a moving image is photographed by interlace scanning at intervals of 1/60 seconds. As an encoding method for interlaced video signals shot by such a video camera, field-based encoding for encoding field data and frame data synthesized from two adjacent field data are encoded. Frame-based encoding (see FIG. 1).

  In the field data, since the scanning lines are skipped, the pixel correlation in the vertical direction is weaker than the frame data in which the scanning lines are continuous. Therefore, in terms of compression efficiency, it can be said that frame-based encoding is more advantageous than field-based encoding, but this is not necessarily the case, for example, when the amount of motion of the subject is large. When the subject moves between the two fields that make up the frame, the edge of the subject in the frame data becomes a comb-like shape (comb shape) in units of lines, and the pixel correlation in the vertical direction of that portion is extremely deteriorated and is difficult to compress. (For example, refer to Patent Document 1).

  In view of these points, an image encoding method has been proposed as an interlace video signal encoding method in which field-based encoding is performed in a portion with a large amount of motion and frame-based encoding is performed in a portion with a small amount of motion. (For example, refer to Patent Document 2).

  In addition, in the field-based subband coding for the interlaced video signal, regarding the luminance signal in the video signal, the magnitude relationship of the quantization step width for each of the LH, HL, and HH subband signals is set to LH <HL <HH, and Regarding the color difference signal in the interlaced video signal in the 4: 2: 1 or 4: 1: 1-1 format, the magnitude relationship of the quantization step width with respect to the LH, HL, and HH subband signals of the layer including the highest frequency component An encoding method that satisfies HL <LH <HH has been proposed (see Patent Document 3).

JP 2002-64830 A Japanese Patent No. 2507199 Japanese Patent No. 3534465

  The inventor has noted that field-based encoding and frame-based encoding differ greatly in image quality characteristics as well as the above-described differences in encoding characteristics. This will be described below.

In general, the encoding of a signal called subband encoding is
The procedure is as follows: frequency conversion of the original signal into subbands (subband division) → quantization of “frequency domain coefficients” constituting the subbands → entropy coding of the quantized coefficients. Here, the subband is a set of “frequency domain coefficients” classified for each frequency band. “Frequency domain coefficients (hereinafter referred to as frequency coefficients)” means that the frequency transform is DCT. If the frequency transform is a wavelet transform, it is a wavelet coefficient. The base length of the frequency transformation (the number of pixels in the horizontal direction and the vertical direction of the block to be subjected to DCT in the case of DCT, the tap length of the high pass coefficient and the low pass coefficient in the case of wavelet transformation) and the image quality of the decoded image are closely related. Have

  A base length of x means that a low frequency coefficient or a high frequency coefficient is calculated based on pixels in the range x during encoding, and one frequency coefficient for x pixels during decoding. Means influence. In general, in transform coding, it is normal to quantize high-frequency coefficients that have little visual impact to increase the compression rate. During decoding, images are mainly decoded based on low-frequency coefficients that were not quantized. You can think of it. Then, normally, the smoothness of the latter is higher than the former in the image decoded based on the low frequency coefficient of the base length x and the image decoded based on the low frequency coefficient of the double base length 2x. The decoded image is high. At the time of decoding, the latter one low frequency coefficient affects the pixel value in the range twice as large as the former coefficient. This is because the influence of one coefficient value is not localized at the time of inverse frequency conversion and becomes universal. This is because the smoothing effect is produced as a result.

  Now, because an interlaced image means an image that is the target of interlaced display, when both a frame-based encoded image and a field-based encoded image at half the height are displayed Note that the images are interlaced with the same size (height). In other words, each field-based (= odd / even field) code is displayed in a vertically stretched form (actually with a scan line), Visually, there is an effect that the base length of the frequency conversion of each code is doubled in the direction perpendicular to the scanning line. This is because the range of pixel values affected by one coefficient in the vertical direction is doubled on the CRT display surface. As a result, field-based coding results in a decoded image with higher smoothness, which may be subjectively undesirable.

  Therefore, in field-based encoding, it is necessary to control the smoothness in the direction orthogonal to the scanning line (vertical direction), and a typical method is control of the base length. In order to suppress vertical smoothing, it is effective to maintain the sharpness of the edge itself extending in the scanning line direction (horizontal direction), and the typical method reflects the amount of edge extending in the horizontal direction. In this method, the quantization of the subband coefficient (LH coefficient) is refrained from the quantization of the subband coefficient (HL coefficient) reflecting the vertical edge amount.

  The encoding method of Patent Document 3 defines such a quantization relationship between subbands. However, the magnitude relationship of the quantization step width “LH <HL <HH” related to the luminance signal in this encoding method and the magnitude relationship of the quantization step width “HL <LH <HH” related to the color difference signal are both It is based on experimental knowledge (see paragraphs (0018) to (0028) of Patent Document 3) and is not derived from a theoretical point of view. It has been found.

  In view of this, the present invention more accurately reflects visual frequency characteristics based on theoretical knowledge in subband coding of interlaced video signals (in a broader sense, signals having different frequency bands in the horizontal and vertical directions). More specifically, the number of quantization steps of subband coefficients, the number of truncations for truncation of subband coefficients or their codes, or the distortion slope for Lagrangian rate control can be obtained from theoretical knowledge. Based on this, the visual frequency characteristics are more appropriately reflected to improve the decoded image quality.

  In the present invention, in order to achieve the above object, an optimal visual frequency characteristic value (a value corresponding to Visual Weight (visual weight) defined in JPEG2000) for a subband is theoretically determined, and this is subtracted. Used for band coding. In the following description, Visual Weight is used as an example of the visual frequency characteristic value.

  First, the contrast sensitivity function (CSF) will be described. This is because the Visual Weight is obtained as an integral value or a product of the contrast sensitivity function (CSF) in a predetermined frequency band, as will be described later.

  Contrast Sensitivity Function (CSF) is the reciprocal of the limit value of the difference / sum (= contrast) that can be discriminated by the observer when a stripe pattern with a predetermined spatial frequency is observed at a predetermined observation distance. Yes, it is required by experiment. The dotted line in FIG. 2 (the dotted line connecting the points where the stripes can only be recognized as gray) corresponds to this, and is specifically represented by the graph of FIG. 3 (note that the reciprocal is taken can be determined. This is because the smaller the contrast, the higher the Sensitivity).

  When this contrast is considered as “the difference between the original image and the image after compression / expansion”, this CSF has a limit that “at each frequency, the difference between the original image and the image after compression / expansion cannot be discriminated below”. Is shown. Since the subband coefficients generally reflect the contrast of the original image, the CSF is “if each frequency (= each subband) and the difference between the subband coefficients of the original image and the compressed and expanded image is less than this, It is interpreted as an index indicating a limit that the difference between the original image reflected by the coefficient and the image after expansion cannot be determined. Therefore, if the subband coefficients are quantized with “twice or less of the reciprocal of CSF”, the quantization error that occurs in each coefficient will be less than or equal to the reciprocal of the CSF, so it is a common theory that image quality degradation should not be recognized. is there.

  In addition, although CSF is also called Visual Transfer Function (VTF), in the present application, visual sensitivity characteristics in the frequency direction represented by CSF and VTF are collectively referred to as “visual frequency characteristics”.

  As described above, since the CSF is “allowable quantization amount of subband coefficient at each frequency”, for example, let us consider an allowable quantization amount of 1H subband obtained by decomposing once in the vertical direction. . The 1H subband is considered to include a wide range of frequencies higher than a certain frequency that have passed through a high-pass filter. However, how to consider the frequency and which point of the CSF graph should be read is a problem. A typical solution for this is as follows.

  First, when the resolution of the original image is d [dpi] and this is observed from the distance v [inch], the spatial frequency f is approximated by f = vd * tan (π / 360), which can be included in the original image. It becomes the maximum frequency. In normal (= maximum decimation) subband division, 1/2 sub-sampling is performed for each decomposition, so the maximum value of the spatial frequency is halved each time the decomposition level is increased by one level. Therefore, the filter used for subband division is regarded as an ideal high-pass filter and low-pass filter, and the band of 0 ≦ f ≦ fmax is 0 ≦ f <fmax / 2 and fmax / 2 in one decomposition. If it is assumed that it is divided into ≦ f ≦ fmax, the band of 1H subband coefficient can be regarded as fmax / 2 ≦ f ≦ fmax, and the band of 1L subband coefficient can be regarded as 0 ≦ f <fmax / 2.

  Therefore, the integrated value in the frequency interval fmax / 2 ≦ f ≦ fmax of CSF or the representative value in the same interval can be used as the Visual Weight for the 1H subband coefficient. Similarly, as the Visual Weight for the 1L subband coefficient, an integral value in the CSF frequency section 0 ≦ f <fmax / 2 or a representative value in the same section can be used. Here, as a representative value of the reference frequency section of CSF, an average value of integral values in the same section, a central value in the same section, a value at one end of the same section, or the like can be used. The average integrated value is a value obtained by dividing the integrated value in the reference frequency section of the CSF by the width of the same section. In addition, it is most important how to select the reference frequency section as described later, rather than using the integral value of the reference frequency section, its representative value, or what kind of representative value is used. . In the following description, it is assumed that an integral value in the reference frequency section of CSF is used.

  Then, based on the fact that the outputs of the sequentially connected filters are expressed by the product of the outputs of the individual filters, the Visual Weight for the 1LH coefficient is the integral value in the CSF frequency section fmax / 2 ≦ f ≦ fmax, and the CSF The product of the integral value in the frequency interval 0 ≦ f <fmax / 2 can be used.

  In FIG. 4, 100 indicates the integral value of the CSF for the 2L subband in the horizontal direction, 101 indicates the integral value of the CSF for the 2H subband in the vertical direction, and the product of these integral values is the Visual Weight of the 2LH subband. is there.

  As described above, in order to obtain Visual Weight (visual frequency characteristic value), it is important how to set a reference frequency section of CSF (visual frequency characteristic). However, conventionally, there has been no example in which the reference frequency section of the CSF is different between the vertical direction and the horizontal direction, whether frame-based encoding or field-based encoding.

  Here, the first gist of the present application is that in the case of field-based coding, it is necessary to make the reference frequency interval of the visual frequency characteristic different in the vertical direction and the horizontal direction (in the vertical direction and the horizontal direction). It is necessary that the visual frequency characteristic value calculated by referring to the frequency sections having different visual frequency characteristics be reflected in the encoding). Further, the second gist of the present application is that even in frame-based coding, it is vertical for signal components (such as 422 format color difference components described later) that are subsampled at different rates in the vertical and horizontal directions. It is necessary to make the reference frequency interval of the visual frequency characteristic different in the direction and the horizontal direction (the visual frequency characteristic value calculated by referring to the frequency interval in which the visual frequency characteristic is different in the vertical direction and the horizontal direction is encoded) It is necessary to reflect on Further, the third purpose of the present application is that it is necessary to appropriately set the reference frequency section of the visual frequency characteristic even in the case of “simple” wavelet transform described later (the reference frequency section is set appropriately). It is necessary to reflect the visual frequency characteristic value calculated in this way in the encoding).

  The field is a frame divided into two. If the maximum value of the spatial frequency in the vertical direction of the frame at a certain observation distance is fmax, that of the field is fmax / 2. Therefore, with respect to the vertical direction, the reference frequency section of the CSF differs between the case of frame-based coding and the case of field-based coding as shown in Table 1.

よって、垂直方向の各参照周波数区間のＣＳＦの積分値を例えば図５のように定義すれば、表１に示した各参照周波数区間についてのＣＳＦの積分値は表２のように表すことができる。

Therefore, if the integrated value of the CSF in each reference frequency section in the vertical direction is defined as shown in FIG. 5, for example, the integrated value of the CSF in each reference frequency section shown in Table 1 can be expressed as in Table 2. .

よってフレームの水平方向の空間周波数の最大値が垂直方向と同じfmaxであるとすれば、垂直方向と水平方向のＣＳＦの積分値の積であるVisual Weightは表３のように表すことができる。

Therefore, if the maximum value of the spatial frequency in the horizontal direction of the frame is the same fmax as in the vertical direction, the Visual Weight, which is the product of the integrated values of the vertical and horizontal CSFs, can be expressed as shown in Table 3.

なお、図５は輝度信号のＣＳＦについて示した。輝度信号の場合、図３に示すようにＣＳＦはピークを持ち、その低域部分では該ピークよりも小さな値をとる。しかし、Visual Weightの計算にあたっては、上記低域部分でも上記ピーク値を持ち続けるものと補正した上で（すなわち、輝度のＣＳＦも図３に示す色差Ｃｂ，ＣｒのＣＳＦと同様に単調減少の形状をとるものに補正した上で）積分値を算出する。これは低域成分が主観画質に及ぼす大きさ等を考慮した上での例外処理である。また最低解像度のＬＬサブバンドに関しても、同様な観点から、Visual Weightを最大値に補正する（通常、ＣＳＦは最大値を１に正規化して用いるのでVisual Weightの最大値は１になる）。

FIG. 5 shows the CSF of the luminance signal. In the case of a luminance signal, the CSF has a peak as shown in FIG. 3, and takes a value smaller than the peak in the low frequency part. However, in calculating Visual Weight, it is corrected that the peak value is maintained even in the low frequency part (that is, the luminance CSF is also monotonically decreasing like the CSFs of the color differences Cb and Cr shown in FIG. The integrated value is calculated (after correcting for). This is an exceptional process in consideration of the size of the low frequency component on the subjective image quality. For the lowest resolution LL subband, from the same viewpoint, the Visual Weight is corrected to the maximum value (usually, the maximum value of the Visual Weight is 1 because the CSF is used with the maximum value normalized to 1).

On the other hand, the color difference signal is different from the luminance signal. In a video signal called 4: 2: 2 format, the color difference signal is thinned out in half in the horizontal direction (Note that the format that does not thin out the color difference signal is called 4: 4: 4. 4: 2: 2 may be described as 422 and 4: 4: 4 as 444). That is, if the maximum value of the spatial frequency in the horizontal direction of the luminance signal of the frame is fmax, the maximum value of the spatial frequency in the horizontal direction of the color difference signal is fmax / 2 in the video signal of 422 format. Therefore, with respect to the horizontal direction, the reference frequency section of the CSF for luminance and color difference is expressed as shown in Table 4.

よって、４２２フォーマットの場合、水平方向の各参照周波数区間のＣＳＦの積分値は表５のように表すことができる。

Thus, if the 422 format, the integral value of the CSF of each reference frequency interval in the horizontal direction can be expressed as shown in Table 5.

したがって、４２２フォーマットの場合、フレームの垂直方向の空間周波数の最大値が水平方向と同じfmaxであるとすれば、垂直方向と水平方向のＣＳＦの積分値の積であるVisual Weightは表６のように表すことができる。

Therefore, in the case of the 422 format, assuming that the maximum value of the spatial frequency in the vertical direction of the frame is the same fmax as in the horizontal direction, the Visual Weight, which is the product of the integrated values of the vertical and horizontal CSFs, is as shown in Table 6. Can be expressed as

この表から、理論的には、フィールドベース符号化の色差の１ＨＬサブバンドのVisual
Weightと１ＬＨサブバンドのVisual Weightとは同一であるべきことが分かる。前記特許文献３における４２２フォーマットの色差信号に対する量子化ステップ幅の「HL＜LH＜HH」という関係は、上の理論的知見と一致せず、同文献には示されてないが何らかの追加条件を前提にした場合にのみ成立するものと考えざるを得ない。

From this table, theoretically, the Visual Difference of 1HL subband of color difference of field-based coding.
It can be seen that the Weight and the Visual Weight of the 1LH subband should be the same. The relationship of “HL <LH <HH” of the quantization step width with respect to the 422 format color difference signal in Patent Document 3 does not agree with the above theoretical findings, and is not shown in the same document, but has some additional conditions. It must be considered that this is only true if the assumption is made.

  Further, in the sub-sampling format of the color difference signal generally called the 4: 1: 1 or 4: 2: 0 format, the color difference signal is thinned out to 1/2 in the horizontal direction and 1/2 in the vertical direction (hereinafter simply referred to as 411). Or 420). That is, if the maximum value of the horizontal and vertical spatial frequencies of the luminance signal of the frame is fmax, the maximum value of the horizontal and vertical spatial frequencies of the color difference signal is both fmax / 2. Therefore, in the case of the 411 format, Visual Weight can be expressed as shown in Table 7 from the same discussion as above.

ここで、発明者の計算では、ＤＥ＞ＣＦである。したがって、表７におけるフィールドベース符号化における色差信号の１ＬＨサブバンド係数に対する量子化ステップ幅よりも１ＨＬサブバンド係数に対する量子化ステップ幅の方を大きくすべきという結果になり、これは前記特許文献３に示された４１１フォーマットの場合の色差信号の量子化ステップ幅の関係「HL＜LH＜HH」と一致しない。この点でも、前記特許文献３における量子化ステップ幅の関係は限定された条件でのみ成立するものと考えざるを得ない。

Here, DE> CF in the inventor's calculation. Accordingly, the result is that the quantization step width for the 1HL subband coefficient should be larger than the quantization step width for the 1LH subband coefficient of the color difference signal in the field-based encoding in Table 7. In the case of the 411 format shown in FIG. 4, the relationship between the quantization step widths of the color difference signal “HL <LH <HH” is not matched. In this respect as well, the relationship between the quantization step widths in Patent Document 3 must be considered to be established only under limited conditions.

  In addition, when the color difference signal is subsampled, there are the following two types of frequency conversion for subband division of the color difference signal (here, description will be made assuming that wavelet conversion is used as frequency conversion). One of them is a method of performing wavelet transform (subband division) on the chrominance signal the same number of times in the horizontal and vertical directions as in the luminance signal, as shown in FIGS. This is referred to herein as “normal” wavelet transform (generally “normal” frequency transform).

  The other is a method of omitting the wavelet transform for the direction in which the color difference signal is thinned out, as shown in FIGS. This is called “simple” wavelet transformation (generally “simple” frequency transformation). In the “simple” wavelet transform, in the case of the 422 format, the wavelet transform is omitted in the horizontal direction of the color difference signal only in the decomposition level 1. Therefore, at decomposition level 1, there are only 1H and 1L subbands in the vertical direction. If 1L is further decomposed, only 1H remains after all. Although not shown in FIGS. 8 and 9, in the case of the 411 format, wavelet transform (subband division) is omitted only in the decomposition level 1 in both the horizontal and vertical directions of the color difference signal.

  The reference frequency section of the visual frequency characteristic for referring to the Visual Weight needs to be appropriately selected in the case of the “normal” wavelet transform and the case of the “simple” wavelet transform. However, conventionally, there is no known example of selecting such a reference frequency section for the “simple” wavelet transform.

  Based on the above theoretical findings, the inventor calculated the Visual Weight for the luminance Y, the color difference Cb, and Cr for the typical observation distances 1000, 1700, and 3000 (note that the visual distance is the resolution) The observation distance when the image of x (dpi) is observed y inches apart is defined as xy). Examples of Visual Weight calculated for the observation distance 1700 are shown in FIGS. 10 to 12 are Visual Weights in the case of “normal” wavelet transformation, and FIGS. 13 and 14 are Visual Weights in the case of “simple” wavelet transformation. 13 and 14, “0” indicates that the subband does not exist, and 1LH is actually 1H in the vertical direction as described above.

As described above, the inverse of Visual Weight is proportional to the number of quantization steps. Visual Weight is inversely proportional to the amount of bit plane discard (number of truncations) in bit plane encoding described later. As will be described later, the number of quantization steps generally used in JPEG2000, which is a representative example of subband coding, is HL = LH <HH at the same decomposition level (see FIG. 30). Therefore, if the number of quantization steps and the amount of bit plane discard (and the combined effect when linear quantization and bit plane discard are used together) are collectively expressed as "degree of quantization" according to theoretical findings of the present invention, the relationship between the degree of quantization in intersubband should be as Figure 15.

  On the other hand, the relationship between the number of quantization steps described in the claims of Patent Document 3 is shown in FIG. In Patent Document 3, there is no description regarding the bit plane discard amount or the simple wavelet in the first place, but FIG. 15 and FIG. 16 agree only with respect to the quantization of the luminance signal in the field-based coding. .

  The theoretical knowledge described above can be widely applied not only to encoding interlaced video signals but also to encoding of signals having different frequency bands in the horizontal and vertical directions.

Therefore, the invention according to claim 1 is an encoding device that performs subband encoding on an interlace video signal on a field basis,
Subband dividing means for dividing the luminance signal of the interlaced video signal into subbands the same number of times in the horizontal and vertical directions;
Coding processing means for performing coding processing of each subband of the luminance signal obtained by subband division by the subband division means;
Means for reflecting the luminance visual frequency characteristic value corresponding to the subband calculated based on the luminance visual frequency characteristic in the encoding process of each subband of the luminance signal;
The frequency interval of the luminance visual frequency characteristic referred to for calculating the luminance visual frequency characteristic value is one of the subband divisions with respect to the frequency interval in which the frequency interval referred to in the vertical direction is referred to in the horizontal direction. The encoding apparatus is characterized in that it is set as a section shifted to the low frequency side by the decompression level.

The invention according to claim 2 is an encoding device that performs subband encoding on a field basis for an interlaced video signal in 4: 4: 4 format or 4: 1: 1 format,
Subband dividing means for dividing the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband dividing means;
Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal;
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. The encoding apparatus is characterized in that it is set as a section shifted to the low frequency side by one decomposition level.

The invention according to claim 3 is an encoding device that performs sub-band encoding on a field basis for interlaced video signals in 4: 2: 2 format,
Subband dividing means for dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the luminance signal and the color difference signal obtained by subband division by the subband dividing means;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. The coding apparatus is characterized in that the characteristic frequency section is set as a section shifted to the low frequency side by one subcomposition level of the subband division.

The invention according to claim 4 is an encoding device that performs sub-band encoding of interlaced video signals in 4: 2: 2 format on a frame basis,
Subband dividing means for dividing the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband dividing means;
Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal;
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. The encoding apparatus is set as a section shifted to a high frequency side by one decomposition level.

The invention according to claim 5 is an encoding apparatus that performs subband encoding on a frame basis for an interlaced video signal in a 4: 1: 1 format.
Subband dividing means for dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the luminance signal and the color difference signal obtained by subband division by the subband dividing means;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. The coding apparatus is characterized in that the characteristic frequency section is set as a section shifted to the low frequency side by one subcomposition level of the subband division.

The invention according to claim 6 is an encoding device that performs sub-band encoding of interlaced video signals in 4: 2: 2 format on a field basis,
Subband division means for dividing the color difference signal in the interlaced video signal into subbands only in the vertical direction only at the composition level 1, and subbanding in the horizontal direction and the vertical direction at the decomposition level 2 or higher;
Encoding processing means for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband dividing means;
Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal;
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. The encoding apparatus is characterized in that it is set as a section shifted to the low frequency side by one decomposition level.

The invention according to claim 7 is an encoding device that performs subband encoding on a field basis for an interlaced video signal in a 4: 1: 1 format.
Subband division means for dividing a luminance signal and a color difference signal in the interlaced video signal into subbands in a horizontal direction and a vertical direction, but not subband division only for the decomposition level 1 with respect to the color difference signal;
Encoding processing means for performing encoding processing of each subband of the luminance signal and the color difference signal obtained by subband division by the subband dividing means;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. The encoding apparatus is characterized in that it is set as a section shifted to the low frequency side by one decomposition level.

The inventions according to claims 8 to 14 below enable the inventions according to claims 1 to 7 to be implemented using a computer.

That is, the invention according to claim 8 performs interband video signal field-based subband coding, so
A subband dividing step of dividing the luminance signal in the interlaced video signal into the same number of times in the horizontal and vertical directions.
An encoding processing step for performing encoding processing of each subband of the luminance signal obtained by subband division by the subband division step;
A program for executing a step of reflecting the luminance visual frequency characteristic value corresponding to the subband calculated based on the luminance visual frequency characteristic in the encoding processing of each subband of the luminance signal,
The frequency interval of the luminance visual frequency characteristic referred to for calculating the luminance visual frequency characteristic value is one of the subband divisions with respect to the frequency interval in which the frequency interval referred to in the vertical direction is referred to in the horizontal direction. It is a program characterized in that it is set as a section shifted to the low frequency side by the decompression level.

According to the ninth aspect of the present invention, there is provided a computer for field-based subband encoding of interlaced video signals in 4: 4: 4 format or 4: 1: 1 format.
A subband dividing step of dividing the color difference signal in the interlaced video signal into the same number of times in the horizontal and vertical directions.
An encoding processing step for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband division step;
A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the previous color difference signal,
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. It is a program characterized in that it is set as a section shifted to the low frequency side by one decomposition level.

According to the tenth aspect of the present invention, there is provided a computer in order to perform field-based subband encoding of a 4: 2: 2 format interlaced video signal.
A subband dividing step of dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal direction and the vertical direction.
An encoding process step for performing an encoding process on each subband of the luminance signal and the color difference signal obtained by subband division in the subband division step;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. The program is characterized in that it is set as a section shifted to the low frequency side by one decomposition level of the subband division with respect to the characteristic frequency section.

According to the eleventh aspect of the present invention, in order to subframe-encode an interlace video signal in 4: 2: 2 format on a frame basis,
A subband dividing step of dividing the color difference signal in the interlaced video signal by the same number of times in the horizontal direction and the vertical direction.
An encoding processing step for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband division step;
A program for executing a process of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal,
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. The program is set as a section shifted to the high frequency side by one decompression level.

In the invention according to claim 12 , in order to perform subband coding on a frame basis for an interlaced video signal in a 4: 1: 1 format,
A subband dividing step of dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal direction and the vertical direction.
An encoding process step for performing an encoding process on each subband of the luminance signal and the color difference signal obtained by subband division in the subband division step;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. The program is characterized in that it is set as a section shifted to the low frequency side by one decomposition level of the subband division with respect to the characteristic frequency section.

According to the thirteenth aspect of the present invention, an interlace video signal in 4: 2: 2 format is subband encoded on a field basis.
A subband division step of dividing the color difference signal in the interlaced video signal into subbands only in the vertical direction only at the composition level 1, and subbanding in the horizontal direction and the vertical direction at the decomposition level 2 or higher;
An encoding processing step for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband division step;
A program for executing a process of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal,
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. This is a program characterized in that it is set as a section shifted to the low frequency side by one decompression level.

According to the fourteenth aspect of the present invention, in order to perform field-based subband encoding of a 4: 1: 1 format interlaced video signal,
A subband dividing step of dividing the luminance signal and the color difference signal in the interlaced video signal in the horizontal direction and the vertical direction, but not performing the subband division on the color difference signal only at the decomposition level;
An encoding process step for performing an encoding process on each subband of the luminance signal and the color difference signal obtained by subband division in the subband division step;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. This is a program characterized in that it is set as a section shifted to the low frequency side by one decompression level.

A fifteenth aspect of the present invention is a computer-readable information recording medium on which the program according to any one of the eighth to fourteenth aspects is recorded.

According to the first aspect of the present invention, field-based subband encoding that appropriately reflects visual frequency characteristics can be performed on an interlaced video signal.

According to the second aspect of the invention, it is possible to perform field-based subband encoding that appropriately reflects the visual frequency characteristics of an interlace video signal in 4: 4: 4 format or 4: 1: 1 format.

According to the third aspect of the present invention, field-based subband encoding that appropriately reflects visual frequency characteristics is possible for 4: 2: 2 format interlaced video signals.

According to the invention of claim 4, 4: 2: For 2 format interlaced video signal are possible adequately reflects frame-based subband coding and the visual frequency characteristic.

According to the fifth aspect of the invention, it is possible to perform frame-based subband coding that appropriately reflects the visual frequency characteristics for an interlaced video signal in the 4: 1: 1 format.

According to the sixth aspect of the present invention, a field-based subband code that appropriately reflects the visual frequency characteristics by using a “simple” wavelet transform or the like for subband division for a 4: 2: 2 format interlaced video signal. Is possible.

According to the seventh aspect of the present invention, a field-based subband code that appropriately reflects visual frequency characteristics by using a “simple” wavelet transform or the like for subband division with respect to an interlaced video signal of 4: 1: 1 format. Is possible.

According to the eighth aspect of the present invention, it is possible to execute field-based subband encoding that appropriately reflects the visual frequency characteristics of an interlaced video signal using a computer.

According to the ninth aspect of the present invention, field-based subband encoding that appropriately reflects visual frequency characteristics is applied to an interlace video signal in 4: 4: 4 format or 4: 1: 1 format using a computer. Can be executed.

According to the tenth aspect of the present invention, it is possible to perform field-based subband encoding that appropriately reflects the visual frequency characteristics of a 4: 2: 2 format interlaced video signal using a computer.

According to the eleventh aspect of the present invention, it is possible to execute frame-based subband encoding that appropriately reflects the visual frequency characteristics of the color difference signal in the interlaced video signal of 4: 2: 2 format using a computer. It is.

According to the twelfth aspect of the present invention, it is possible to execute frame-based subband encoding that appropriately reflects the visual frequency characteristics of an interlaced video signal of 4: 1: 1 format using a computer.

According to the thirteenth aspect of the present invention, a field in which visual frequency characteristics are appropriately reflected using a “simple” wavelet transform or the like for subband division for a 4: 2: 2 format interlaced video signal using a computer. Base subband coding can be performed.

According to the fourteenth aspect of the present invention, a field in which visual frequency characteristics are appropriately reflected using a “simple” wavelet transform or the like for subband division of a 4: 1: 1 format interlaced video signal using a computer. Base subband coding can be performed.

  FIG. 17 is a block diagram of an encoding apparatus according to an embodiment of the present invention. The encoding apparatus according to the present embodiment executes encoding processing using a JPEG2000 encoding algorithm, and includes a DC level shift / color conversion block 200, a wavelet conversion block 201, a quantization block 202, and an entropy encoding block 203. , A packet generation block 204, a code formation block 205, and a control block 206 for controlling the operation of each of the above blocks.

The DC level shift / color conversion block 200 is an unnecessary element depending on the color system of the input video signal. The wavelet transform block 201 is an element constituting the subband dividing means of the inventions according to claims 1 to 16. Quantization block 202, entropy coding block 2
03, the packet generation block 204 and the code forming block 205 are elements constituting the encoding processing means of the inventions according to claims 1 to 8 .

  The control block 206 also includes means for performing the control specific to the present invention described above. The control specific to the present invention is the control of the number of quantization steps used for linear quantization in the quantization block 202, the entropy coding block 203 or the packet generation, as will be described in detail with reference to the following embodiments. Control of truncation number (number of discarded bit planes or sub bit planes) when coefficient or code truncation is performed in block 204 or distortion slope when Lagrange rate control is performed in packet generation block 204 Control.

  In order to execute such control peculiar to the present invention, the control block 206 according to the present embodiment includes a VisualWeight table 207 storing Visual Weight as shown in FIGS. 10 to 14 and a JPEG2000 encoding device. 30, a reference quantization step number table 208 storing quantization step numbers (referred to as reference quantization step numbers) normally used for linear quantization, a quantization step number control means 209, and a truncate Number control means 210 and slope control means 211 are included.

The quantization step number control means 209 sets the reference quantization step number read from the reference quantization step number table 208 in the quantization block 202 as it is, or uses the Visual Weight read from the VisualWeight table 207 as a reference quantization. This is means for correcting the number of steps and setting the corrected number of quantization steps in the quantization block 202. The truncation number control means 210 is a means for converting the Visual Weight read from the VisualWeight table 207 into a truncation number in bit plane units or sub bit plane units, and setting the truncation number in the entropy encoding block 203 or the packet generation block 204. is there. The slope control unit 211 is a unit that corrects the distortion slope calculated in the entropy encoding block 203 using the Visual Weight read from the VisualWeight table 207 and sets the corrected distortion slope in the packet generation block 204.

Such means 207, 208, 209, 210, 211 are the visual frequency characteristic values of luminance or color difference corresponding to subbands in the subband coding processing of luminance signal or color difference signal in the inventions according to claims 1 to 7. "Means for reflecting".

  Note that the video signal to be encoded may be given as an RGB value, or may be given as a YCbCr value that is a luminance / color difference. For example, when “444 or 411 format video signal” is specified, it is assumed that the video signal is given as a YCbCr value. Otherwise, both the RGB value and the YCbCr value are targeted. In JPEG2000, when the input image is RGB data, processing is usually performed for each component after DC level shift and color conversion to luminance Y / color differences Cb and Cr. When the input image is YCbCr data, the DC level is used. Shift is also performed without color conversion.

[Outline of JPEG2000]
Here, an outline of JPEG2000 will be described. For example, when a color image composed of three RGB components is compressed, the DC level / color conversion block 200 performs a DC level shift, and then color conversion (components) of luminance Y and color differences Cb and Cr into components. Conversion). Each of these components is subjected to subband division by wavelet transform in the wavelet transform block 201, and each subband coefficient is subjected to linear quantization (scalar quantization) in the quantization block 202 as necessary, and then entropy. The encoding block 203 performs entropy encoding in units of bit planes (more precisely, the bit plane is subdivided into three sub-bit planes and encoded). In JPEG 2000, arithmetic coding called MQ coding is used for this entropy coding. The packet generation block 204 generates a packet in which necessary entropy codes (MQ codes) are collected, and the code formation block 205 arranges the packets in a predetermined order and adds a necessary tag or tag information. A code stream of the format is formed.

  The code stream decompression process is exactly the reverse of the compression process. The code of each component in the code stream is entropy decoded and returned to wavelet coefficients. The wavelet coefficients are inversely quantized as necessary and then subjected to inverse wavelet transform to be returned to the image data of each component. The image data of each component is returned to the original color system image data by reverse color conversion and reverse DC level shift when DC level shift and color conversion are performed at the time of compression.

Note that the DC level shift (forward conversion) and the reverse DC level shift (reverse conversion) in JPEG 2000 are as follows.
Forward conversion I (x, y) ← I (x, y) -2 ^ Ssiz (i)
Inverse transformation I (x, y) ← I (x, y) + 2 ^ Ssiz (i)
However, Ssiz (i) is obtained by subtracting 1 from the bit depth of each component i of the original image (i = 0, 1, 2 for RGB images). 2 ^ Ssiz (i) means 2 to the power of Ssiz (i), and I (x, y) is an original signal value (pixel value) at coordinates (x, y).

  The DC level shift is applied to a signal that takes a positive number such as an RGB value, and is an operation of subtracting half of the dynamic range of the signal from each signal value. Is the operation of adding half of the dynamic range of the signal. DC level shift is not applied to signed integers such as Cb and Cr signals of YCbCr signals.

  JPEG2000 defines reversible conversion (RCT) and irreversible conversion (ICT) as component conversion (color conversion).

RCT forward and inverse transforms are expressed by the following equations.
Forward conversion Y0 (x, y) = floor ((I0 (x, y) + 2 * (I1 (x, y) + I2 (x, y)) / 4)
Y1 (x, y) = I2 (x, y) -I1 (x, y)
Y2 (x, y) = I0 (x, y) -I1 (x, y)
Inverse transformation I1 (x, y) = Y0 (x, y) -floor ((Y2 (x, y) + Y1 (x, y)) / 4)
I0 (x, y) = Y2 (x, y) + I1 (x, y)
I2 (x, y) = Y1 (x, y) + I1 (x, y)
In the equation, I represents an original signal, Y represents a signal after conversion, and 0 to 2 following I and Y are suffixes. For RGB signals, I0 = R, I1 = G, I2 = B in the I signal, and Y0 = Y, Y1 = Cb, Y2 = Cr in the Y signal. Floor (X) is a function that replaces the real number X with an integer that does not exceed X and is closest to X.

The forward conversion and the reverse conversion of ICT are expressed by the following equations.
Forward conversion Y0 (x, y) = 0.299 * I0 (x, y) + 0.587 * I1 (x, y) + 0.144 * I2 (x, y)
Y1 (x, y) =-0.16875 * I0 (x, y) -0.33126 * I1 (x, y) + 0.5 * I2 (x, y)
Y2 (x, y) = 0.5 * I0 (x, y) -0.41869 * I1 (x, y) -0.08131 * I2 (x, y)
Inverse transformation I0 (x, y) = Y0 (x, y) + 1.402 * Y2 (x, y)
I1 (x, y) = Y0 (x, y) -0.34413 * Y1 (x, y) -0.71414 * Y2 (x, y)
I2 (x, y) = Y0 (x, y) + 1.772 * Y1 (x, y)
In the equation, I represents an original signal, and Y represents a signal after conversion. For RGB signals, I0 = R, I1 = G, I2 = B in the I signal, and I0 = Y, I1 = Cb, I2 = Cr in the Y signal.

  In JPEG2000, quantization and entropy coding are closely related, and the configuration for encoding after quantization is the configuration for discarding unnecessary bitplane or subbitplane codes after encoding (or necessary bits). There is also a configuration in which only a plane or a sub-bit plane is encoded.

  In the case of quantization by linear quantization, publicly known linear quantization is performed on the wavelet coefficients, and a bit plane composed of the quantized coefficients is entropy encoded.

  Without linear quantization, discard (truncate) the code of unnecessary bitplanes or subbitplanes, or encode only the necessary bitplanes or subbitplanes (the remaining lower bitplanes or subbits) The configuration is such that the plane is discarded. In the present application, truncation includes a case where only necessary (sub) bitplanes except unnecessary (sub) bitplanes are encoded and a case where unnecessary (sub) bitplane codes are discarded after encoding. The number of bit planes (or the number of sub bit planes) called, discarded, or made unnecessary is called the number of truncations.

  In JPEG2000, when 5 × 3 wavelet transform is used, linear quantization cannot be applied, and only truncation is performed. In this application, both linear quantization and truncation may be referred to as quantization. In addition, in the case of an encoding method in which nonlinear quantization of the coefficient itself is allowed, this nonlinear quantization is also included in the quantization referred to in the present application.

  Here, the 5 × 3 wavelet transform is a conversion in which 5 pixels are used to obtain one low-pass filter output (low-pass coefficient), and 3 pixels are used to obtain one high-pass filter output (high-pass coefficient). is there. JPEG2000 also defines 9 × 7 wavelet transform. This 9 × 7 wavelet transform is a transform that obtains the output (low-pass coefficient) of one low-pass filter using 9 pixels, and obtains the output (high-pass coefficient) of one high-pass filter using 7 pixels. The main difference between the 5 × 3 wavelet transform and the 9 × 7 wavelet transform is the difference in the filter range, and the low pass filter is applied to the center of the even position and the high pass filter is applied to the center of the odd position.

The conversion formula of the 5 × 3 wavelet transform is as follows.
(Forward conversion)
[step1] C (2i + 1) = P (2i + 1) −floor ((P (2i) + P (2i + 2)) / 2)
[step2] C (2i) = P (2i) + floor (((C (2i-1) + C (2i + 1) +2) / 4)
(Inverse transformation)
[step1] P (2i) = C (2i) −floor ((C (2i-1) + C (2i + 1) +2) / 4)
[step2] P (2i + 1) = C (2i + 1) + floor ((P (2i) + P (2i + 2)) / 2).

The conversion formula of 9 × 7 wavelet transform is as follows.
(Forward conversion)
[step1] C (2n + 1) = P (2n + 1) + α * (P (2n) + P (2n + 2))
[step2] C (2n) = P (2n) + β * (C (2n-1) + C (2n + 1))
[step3] C (2n + 1) = C (2n + 1) + γ * (C (2n) + C (2n + 2))
[step4] C (2n) = C (2n) + δ * (C (2n-1) + C (2n + 1))
[step5] C (2n + 1) = K * C (2n + 1)
[step6] C (2n) = (1 / K) * C (2n)
(Inverse transformation)
[step1] P (2n) = K * C (2n)
[step2] P (2n + 1) = (1 / K) * C (2n + 1)
[step3] P (2n) = X (2n) -δ * (P (2n-1) + P (2n + 1))
[step4] P (2n + 1) = P (2n + 1) -γ * (P (2n) + P (2n + 2))
[step5] P (2n) = P (2n) -β * (P (2n-1) + P (2n + 2))
[step6] P (2n) = P (2n + 1) -α * (P (2n) + P (2n + 2))
However, α = -1.586134342059924
β = -0.052980118572961
γ = 0.882911075530934
δ = 0.443506852043971
K = 1.230174104914001
Next, the wavelet transform procedure and the definition of the composition level, resolution level, and subband will be described.

  18 to 21 are diagrams for explaining a process of performing 5 × 3 wavelet transform in two dimensions (vertical direction and horizontal direction) on a 16 × 16 monochrome image.

  As shown in FIG. 18, XY coordinates are taken, and a pixel value of a pixel having a Y coordinate y is expressed as P (y) (0 ≦ y ≦ 15) with respect to a certain X coordinate x. In JPEG2000, a coefficient C (2i + 1) is obtained by first applying a high-pass filter in the vertical direction (Y-coordinate direction) around an odd-numbered pixel (y = 2i + 1). Next, a coefficient C (2i) is obtained by applying a low-pass filter around the pixel having an even Y coordinate (y = 2i) (this is performed for all X coordinates). Here, the high-pass filter is represented by the above-described forward conversion [step 1], and the low-pass filter is represented by the forward conversion [step 2]. Note that there may be no adjacent pixel at the edge of the image with respect to the central pixel. In this case, the pixel value is appropriately compensated according to a predetermined rule.

  For simplicity, if the coefficient obtained by the high-pass filter is denoted by H and the coefficient obtained by the low-pass filter is denoted by L, the image in FIG. 18 is converted into an array of L and H coefficients as shown in FIG. Is converted.

  Subsequently, a high-pass filter is applied to the coefficient array in FIG. 19 around the coefficient whose X coordinate is odd (y = 2i + 1) in the horizontal direction, and then the coefficient whose X coordinate is even (x = 2i). (This is performed for all the Y coordinates. In this case, P (2i) and the like in [step 1] and [step 2] of the forward transformation are read as representing coefficient values).

For simplicity,
The coefficient obtained by applying a low pass filter around the L coefficient is LL,
The coefficient obtained by applying a high-pass filter around the L coefficient is HL,
The coefficient obtained by applying a low-pass filter around the H coefficient is LH,
The coefficient obtained by applying a high-pass filter around the H coefficient is HH,
19 is converted into a coefficient array as shown in FIG. Here, the coefficient group to which the same symbol is attached is called a subband. The example of FIG. 20 includes four subbands.

  This completes one wavelet transform (one decomposition (decomposition)) for each of the vertical and horizontal directions. When only the LL coefficients are collected (collected for each subband as shown in FIG. 21). , Subtracting only the LL subband), an “image” having a resolution of 1/2 of the original image is obtained (in this way, classifying each subband is called deinterleaving, and the state shown in FIG. 20 is obtained. Placement is called interleaving).

  The second wavelet transform may be performed in the same manner as described above with the LL subband as the original image. In this case, when deinterleaving is performed, a schematic FIG. 22 is obtained. The prefix numbers of the coefficients shown in FIG. 21 and FIG. 22 indicate how many times the wavelet transform has been obtained with respect to the horizontal and vertical directions, and are called decomposition levels. Further, FIG. 23 shows resolution levels that are substantially opposite to the decomposition levels.

  In the above discussion, when one-dimensional wavelet transformation is desired, processing in only one direction may be performed, and the number of times wavelet transformation has been performed in any direction becomes the decomposition level.

  On the other hand, in the wavelet inverse transform, an inverse low-pass filter is first applied to the array of interleaved coefficients as shown in FIG. 20 in the horizontal direction, centering on coefficients whose X coordinates are even (x = 2i), and then X An inverse high-pass filter is applied around the coefficient whose coordinates are odd (x = 2i + 1) (this is performed for all y). Here, the inverse low-pass filter is represented by the inverse transformation [step 1] equation, and the inverse high-pass filter is represented by the inverse transformation [step 2] equation. In the end portion of the image, there may be no adjacent coefficient for the central coefficient. In this case, the coefficient value is appropriately supplemented by a predetermined rule.

  By the above horizontal processing, the coefficient array in FIG. 20 is converted (inversely converted) into a coefficient array as shown in FIG. Subsequently, in the vertical direction, an inverse low-pass filter is applied centering on a coefficient whose Y coordinate is an even number (y = 2i), and then an inverse high-pass filter is applied centering on a coefficient whose Y coordinate is an odd number (y = 2i + 1). . By performing this process for all the X coordinates, one-time two-dimensional wavelet inverse transformation is completed, and the image of FIG. 19 is restored (reconstructed). If the wavelet transform is performed a plurality of times, FIG. 18 is regarded as an LL subband, and the same inverse transform is repeated using coefficients of other subbands such as HL of the same composition level. .

  As described above, when the 5 × 3 wavelet transform is used, linear quantization of the coefficients constituting the subband is not performed. On the other hand, in the case of 9 × 7 wavelet transform, wavelet coefficients can be linearly (scalar) quantized for each subband, and in this case, the same number of quantization steps is used in the same subband.

  Next, entropy coding of wavelet coefficients will be described. Each subband is divided into rectangles called code blocks, and coefficient bit-plane coding is performed in units of code blocks. In JPEG2000, it is possible to encode the coefficient from the upper bit (MSB) to the lower bit (LSB) in units of subbit planes for each subband.

  Assume that the coefficients of the 2LL subband in FIG. 22 have the values shown in FIG. These values are expressed in binary numbers, and each bit is divided into bit planes. The coefficients shown in FIG. 24 can be divided into four bit planes shown in FIG. Since the binary representation of 15 is “1111”, “1” stands at the position corresponding to the coefficient of the value “15” of all the bit planes.

  In JPEG 2000, one bit plane is classified into three sub bit planes (also referred to as coding passes), and encoding is performed for each sub bit plane. Subbit planes (coding passes) include: significance propagation pass (significant coefficients around, pass coding non-significant coefficients), magnitude refinement pass (pass coding significant coefficients), cleanup pass (above There is a path for encoding the remaining bits not corresponding to the two paths). However, as a result of classification, there are cases where there is no bit belonging to a specific sub-bit plane (coding path) in one bit plane, and in this case, an empty sub-bit plane is generated. The top bit plane is always cleanup pass only. Therefore, the four bit planes in FIG. 25 are decomposed into sub bit planes as shown in FIG.

  Since the sub bit plane is encoded from the upper level to the lower level, it is possible to generate a code having a configuration as shown in FIG. FIG. 27 shows that the code starts with the 2LL subband and ends with the 1HH subband. Further, the example of FIG. 27 is a case where all the sub-bit planes are encoded. For example, when it is determined that the code of the colored sub-bit plane in the drawing is unnecessary, As described above, the entropy coding itself can be omitted, or the code of the sub-bit plane can be discarded after entropy coding. The minimum unit for such omission or discard is the sub bit plane. However, when it is desired to perform this easily, omission or discard in bit plane units can be selected.

  This completes the description of the outline of JPEG 2000, and the description returns to the description of the present embodiment.

[Continuation of description of embodiment]
As described above, in order to calculate an appropriate visual frequency characteristic value corresponding to each subband, it is important to appropriately set the frequency section of the visual frequency characteristic referred to for the calculation. . Examples of appropriate reference frequency sections (sometimes abbreviated as reference sections) used in the present embodiment are shown in FIGS. The visual frequency characteristic is the CSF, and the visual frequency characteristic value is the Visual Weight. The reference interval shown in FIG. 28 is appropriate when the “normal” wavelet transform is used, and the reference interval shown in FIG. 29 is appropriate when the “simple” wavelet transform is used.

  In FIG. 28 and FIG. 29, when the maximum frequency in the one-dimensional direction (vertical direction or horizontal direction) is fmax, 2L to 1H subbands generated by the one-dimensional wavelet transform applied to the direction. Is shown in relation to fmax. The expression of the reference section is the same as in Table 1 and Table 4. The “left end” in the figure is the left end of the reference section, and the “right end” in the figure is the right end of the reference section. For example, in (1) of FIG. 28, when frame-based encoding is performed on an interlaced video signal in 444 format, the reference section of the 2L subband is “the left end is 0 and the right end is 0.25”, that is, 0 ≦ f <0.25 fmax. It shows that it is the section. Note that the inequality signs at both ends of the reference section can be handled as 0 ≦ f ≦ 0.25 fmax depending on the assumptions and interpretations provided for the band division characteristics of the wavelet transform filter.

  The Visual Weight stored in the VisualWeight table 207, that is, the Visual Weight shown in FIGS. 10 to 14 was calculated as an integral value of CSF or a product thereof for the reference section as shown in FIGS. Is.

  The reference quantization step number shown in FIG. 30 is a commonly used quantization step number calculated from the subband gain and the component conversion gain in the case of the 9 × 7 wavelet transform of JPEG2000. The present invention is applicable to both field-based encoding and frame-based encoding using normal wavelet transform (however, the case of decomposition level 2 is shown in FIG. 30). Such reference quantization step numbers are stored in the reference quantization step number table 208.

The quantization step number control means 209 reads the necessary Visual Weight from the VisualWeight table 207, and calculates the corrected quantization step number by dividing the reference quantization step number. That is,
Number of quantization steps after correction = reference number of quantization steps / Visual Weight formula (S)
By performing linear quantization using such a modified number of quantization steps, visual frequency characteristics are favorably reflected in linear quantization.

In addition, linear quantization is performed using the reference quantization step number, and by performing the truncation of the coefficient or sign by the number of truncations corresponding to the Visual Weight read from the Visual Weight table 207, the Visual Weight is set to the degree of quantization. It can also be reflected. This is a method of calculating the number of truncations.
Since it is equivalent to “linear quantization of the coefficient by (2 to the power of n)”,
Number of truncations per bit plane = log2 (reciprocal of Visual Weight) Formula (BP)
Thus, the number of truncations per bit plane can be obtained (where 2 means the bottom of the log). In addition, “n subbit planes of truncation” is equivalent to “linearly quantizing a coefficient by (the power of 2 1/3) to the nth power”.
Number of truncations per subbit plane = 3 x log2 (reciprocal of Visual Weight)
Formula (SBP)
Can obtain the number of truncations in units of sub-bit planes (where 2 means the bottom of the log).

The truncation number control means 210 is a Visual Weight required from the VisualWeight table 207.
ight is read, and the truncation number is calculated by using the equation (BP) or the equation (XBP) using the ight, and the truncation number is set in the entropy coding block 203 (when truncation is performed at the coefficient stage), or The number of truncations is set in the packet generation block 204 (when truncation is performed at the code stage). This also allows visual frequency characteristics to be appropriately reflected in the degree of quantization in the encoding process.

  When Lagrangian rate control is performed in the packet generation block 204, the slope control unit 211 performs slope correction for reflecting Visual Weight on the distortion slope, which will be described later.

  Hereinafter, some examples of the present embodiment will be described.

  In this embodiment, a 444 format interlaced video signal (RGB value) is encoded on a field basis.

  FIG. 31 shows a schematic flow of processing in the present embodiment. First, in the control block 206, field-based encoding is selected (step 1). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for field-based encoding. Note that the field-based / frame-based encoding may be explicitly specified by the user in the control block 206, or the motion amount of the video may be specified in the control block 206 by a method such as that disclosed in Patent Document 2, for example. The field-based / frame-based encoding may be automatically selected according to the size of. The same applies to other examples described later.

  In the control block 206, the number of quantization steps for linear quantization is set. In this embodiment, the number of quantization steps is set on the assumption that “normal” wavelet transform (9 × 7 wavelet transform) is used for subband division. That is, in the quantization step number control means 209, the Visual Weight calculated by the CSF integration for the reference section shown in (2) of FIG. The reference section shown in (2) of FIG. 28 is shown again in FIG. As can be seen from FIG. 32, the reference interval for the vertical subband of the luminance is shifted to the low frequency side by one decomposition level with respect to the reference interval for the horizontal subband of the luminance. You should be careful. Therefore, if 1700 is selected as the observation distance, the “444 field” Visual Weight in FIG. 10 is read from the VisualWeight table 207. Then, the quantization step number control means 209 calculates each component and each subband from the Visual Weight read from the VisualWeight table 207 and the reference quantization step number read from the reference quantization step number table 208 according to the equation (S). The modified quantization step number for is obtained and set for the quantization block 202. The corrected number of quantization steps is as shown in FIG. The above is the process of step2.

  The following steps 3 to 7 are processing steps that are repeatedly executed for each frame of the input video signal.

The DC level / color conversion block 200 performs DC level shift and color conversion (ICT) on the video signal (RGB value) (step 3). Then, for each field, the “normal” wavelet transform is performed in the wavelet transform block 201 for each component of Y, Cb, and Cr after the transformation (step 4). Here, it is assumed that 9 × 7 transformation is used as the wavelet transformation. It is assumed that the wavelet transform is executed up to the decomposition level 2.

  Linear quantization using the set number of quantization steps is executed in the quantization block 202 for the coefficient of each subband of each component (step 5). Linear quantization in JPEG 2000 is performed by dividing a coefficient by the number of quantization steps and converting it to an integer using a floor function.

  Next, in the entropy coding block 203, MQ coding is performed on all bit planes of the coefficients of each component after linear quantization (step 6). The generated MQ codes are combined into packets by the packet generation block 204, and a code stream is formed by using the packets by the code formation block 205 (step 7).

  As described above, according to the present embodiment, encoding is performed in which the visual frequency characteristic is appropriately reflected on the number of quantization steps of linear quantization (in other words, the degree of quantization) based on theoretical knowledge. be able to.

  As is apparent from FIG. 33, it should be noted that the magnitude relationship of the number of quantization steps is “LH <HL <HH” for all components and all decomposition levels. Since neither truncation of the coefficient nor truncation of the code is performed, this magnitude relation is the magnitude relation of the degree of quantization as it is.

  It is also possible to prepare a table storing the corrected number of quantization steps shown in FIG. 33 in advance in the control block 206 and read the number of quantization steps from this table and use it for linear quantization. It is.

As is apparent from the description so far, the present embodiment is Ru embodiment der of the invention according to claim 1. Step 4 corresponds to subband division processing by the subband division means, and steps 5 to 7 correspond to encoding processing by the encoding processing means. Then, Visual Weight which is a visual frequency characteristic value of luminance corresponding to the subband is reflected in the number of quantization steps set in step 2, and this quantization step number is used for linear quantization in step 5. The luminance visual frequency characteristic value corresponding to the subband is reflected in the encoding process of each subband of the luminance signal.

  In this embodiment, a 411 format interlaced video signal (YCbCr value) is encoded on a field basis.

  FIG. 34 shows a schematic flow of processing in the present embodiment. First, in the control block 206, field-based encoding is selected (step 11). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for field-based encoding. However, since the input video signal is a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  In the control block 206, the number of quantization steps for linear quantization is set. In this embodiment, the number of quantization steps is set on the assumption that “normal” wavelet transform (9 × 7 wavelet transform) is used for subband division. That is, the quantization step number control means 209 reads the Visual Weight calculated by the CSF integration for the reference section shown in (6) of FIG. The reference section shown in (6) of FIG. 28 is shown again in FIG. As can be seen from FIG. 35, the reference interval for the vertical subband of the color difference is shifted to the low frequency side by one decomposition level with respect to the reference interval for the horizontal subband of the color difference. It should be noted. Therefore, if 1700 is selected as the observation distance, the “411 field” Visual Weight in FIG. 12 is read from the VisualWeight table 207. Then, the quantization step number control means 209 calculates each component and each subband from the Visual Weight read from the VisualWeight table 207 and the reference quantization step number read from the reference quantization step number table 208 according to the equation (S). The modified quantization step number for is obtained and set for the quantization block 202. The corrected number of quantization steps is as shown in FIG. The above is the processing of step 12.

  The following steps 13 to 16 are processing steps repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “normal” wavelet transform is performed for each of the Y, Cb, and Cr components for each field (step 13). As the wavelet transform, 9 × 7 wavelet transform is used. It is assumed that the wavelet transform is performed up to decomposition level 2.

  Linear quantization using the set number of quantization steps is executed in the quantization block 202 for the coefficients of each component and each subband (step 14). Next, in the bit plane coding block 203, MQ coding of all the bit planes of the quantized coefficients is performed (step 15). The generated MQ codes are collected into packets by the packet generation block 204, and a code stream is formed by using the packets by the code formation block 204 (step 16).

  As is apparent from FIG. 36, it should be noted that the magnitude relationship of the number of quantization steps is “LH <HL <HH” for all components and all decomposition levels. Since neither truncation of coefficients nor truncation of codes is performed, this magnitude relationship is a magnitude relationship of the degree of quantization in the encoding process as it is.

  It is also possible to prepare a table storing the corrected number of quantization steps shown in FIG. 36 in advance in the control block 206 and read the number of quantization steps from this table and use it for linear quantization. It is.

As is apparent from the description so far, the present embodiment is an embodiment of the invention according to claim 2.
Step 13 corresponds to subband division processing by the subband division means, and step14
˜16 corresponds to the encoding processing by the encoding processing means. The number of quantization steps set in step 12 includes Visual Weight, which is the visual frequency characteristic value of the color difference corresponding to the subband.
Is reflected, and the visual frequency characteristic value of the color difference corresponding to the subband is reflected in the encoding process of each subband of the color difference signal. Further, the magnitude relationship of the number of quantization steps for the LH, HL, and HH subbands of the luminance signal decomposition level 1 that reflects the visual frequency characteristic value is “LH <HL <
Since “HH” and truncation is not performed, the magnitude relationship of the degree of quantization by the encoding process for these subbands is “LH <HL <HH”.

  In this embodiment, a 422 format interlaced video signal (YCbCr value) is encoded on a field basis.

  FIG. 37 shows a schematic flow of processing in the present embodiment. First, in the control block 206, field-based encoding is selected (step 21). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for field-based encoding. However, since the input video signal has a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, the control block 206 sets the number of quantization steps for linear quantization (step 22). In this embodiment, the reference quantization step number read from the reference quantization step number table 208 by the quantization step number control means 209 is set for the quantization block 202.

  Next, the control block 206 sets the number of truncations for the entropy coding block 203. In this embodiment, the number of truncations is set on the assumption that “normal” wavelet transform (9 × 7 wavelet transform) is used for subband division. That is, the truncation number control means 210 reads the Visual Weight calculated by the CSF integration for the reference section shown in (4) of FIG. The reference section shown in (4) of FIG. 28 is shown again in FIG. As can be seen from FIG. 38, the reference interval for the horizontal subband of the color difference is shifted to the low frequency side by one decomposition level with respect to the reference interval for the horizontal subband of the luminance. You should be careful. Therefore, if 1700 is selected as the observation distance, the “422 field” Visual Weight in FIG. 11 is read from the VisualWeight table 207. Then, the truncation number control means 210 converts the Visual Weight read from the VisualWeight table 207 into the number of truncations in sub-bit plane units using the above equation (SBP). However, since the calculated value by the formula (SBP) is a real number, it is rounded to an integer. The number of truncations after conversion to an integer is a value as shown in FIG. This truncation number is set in the entropy coding block 203. The above is the processing of step 23.

  The following steps 24 to 27 are processing steps repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “normal” wavelet transform is performed for each of the Y, Cb, and Cr components for each field (step 24). As the wavelet transform, 9 × 7 wavelet transform is used. It is assumed that the wavelet transform is performed up to decomposition level 2.

  Linear quantization using the reference quantization step number is performed in the quantization block 202 for the coefficient of each subband of each component (step 25). In the entropy coding block 203, MQ coding is performed on the quantized coefficients of each subband of each component, excluding lower subbitplanes by the set number of truncations (step 26). The generated MQ codes are collected into packets by the packet generation block 204, and a code stream is formed by using the packets by the code formation block 205 (step 27).

  Here, it should be noted that the number of truncations used for truncation in the entropy coding block 203 (FIG. 39) has a magnitude relationship of “HL = LH <HH”. In this embodiment, linear quantization using the reference quantization step number (FIG. 30) is also performed, but the reference quantization step number also has a magnitude relationship of “HL = LH <HH”. Therefore, it should be noted that the “degree of quantization” in the encoding process, which combines the effects of linear quantization and truncation, has a magnitude relationship of “HL = LH <HH”.

  It is also possible to prepare a table storing the number of truncations shown in FIG. 39 in advance in the control block 206 and set the number of truncations read from this table in the entropy encoding block 203.

  In this embodiment, truncation is performed in units of sub-bit planes, but truncation in units of bit planes may be performed, and in this case, the above formula (BP) is used for conversion from Visual Weight to the number of truncations. Become. Further, the entropy coding block 203 may not perform truncation, but may discard lower subbit plane codes corresponding to the number of truncations designated by the packet generation block 204.

As apparent from the above description, the present embodiment is an embodiment of the invention according to claim 3.

  In this embodiment, a 422 format interlaced video signal (YCbCr value) is encoded on a frame basis.

  FIG. 40 shows a schematic flow of processing in the present embodiment. First, in the control block 206, frame-based encoding is selected (step 31). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for frame-based encoding. However, since the input video signal has a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, in the control block 206, the number of quantization steps for linear quantization is set. In this embodiment, the number of quantization steps is set on the assumption that “normal” wavelet transform (9 × 7 wavelet transform) is used for subband division. That is, the quantization step number control means 209 reads the Visual Weight calculated by the CSF integration for the reference section shown in (3) of FIG. The reference section shown in (3) of FIG. 28 is shown again in FIG. As can be seen from FIG. 41, the reference interval for the vertical subband of the color difference is shifted to the high frequency side by one decomposition level with respect to the reference interval for the horizontal subband of the color difference. You should be careful. Accordingly, if 1700 is selected as the observation distance, the “422 frame” Visual Weight in FIG. 11 is read from the VisualWeight table 207. Then, the quantization step number control means 209 calculates each component from the Visual Weight read from the VisualWeight table 207 and the reference quantization step number (FIG. 30) read from the reference quantization step number table 208 according to the equation (S). Then, the corrected number of quantization steps for each subband is obtained and set in the quantization block 202. The number of quantization steps after the correction is a value as shown in FIG.

  The following steps 33 to 36 are processing steps repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “normal” wavelet transform is performed for each of the Y, Cb, and Cr components for each frame (step 33). A 9 × 7 wavelet transform is used as the wavelet transform. It is assumed that the wavelet transform is performed up to decomposition level 2.

  Linear quantization for the coefficients of each component and each subband is performed in the quantization block 202 (step 34). Next, the entropy coding block 203 performs MQ coding of all the bit planes of the quantized coefficients (step 35). The generated MQ code is packetized by the packet generation block 204, and a code stream is formed by using the packet by the code formation block 205 (step 36).

  As is apparent from FIG. 42, it should be noted that the magnitude relationship between the number of quantization steps for the Cb and Cr components is “HL <LH <HH”. Since truncation is not performed, this magnitude relationship is directly a magnitude relationship of the degree of quantization for the Cb and Cr components.

  It is also possible to prepare a table storing the corrected quantization step number shown in FIG. 42 in the control block 206 in advance, and read the quantization step number from this table and use it for linear quantization. It is.

As apparent from the above description, the present embodiment is an embodiment of the invention according to claim 4.

  In this embodiment, a 411 format interlaced video signal (YCbCr value) is encoded on a frame basis.

  FIG. 43 shows a schematic flow of processing in the present embodiment. First, in the control block 206, frame-based encoding is selected (step 41). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for frame-based encoding. However, since the input video signal is a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, the control block 206 sets the number of quantization steps for linear quantization. In this embodiment, the number of quantization steps is set on the assumption that “normal” wavelet transform (9 × 7 wavelet transform) is used for subband division. That is, the quantization step number control means 209 reads the Visual Weight calculated by the CSF integration for the reference section shown in (5) of FIG. The reference section shown in (5) of FIG. 28 is shown again in FIG. As can be seen from FIG. 44, it should be noted that the reference interval for the color difference subband is shifted to the lower frequency side by one decomposition level with respect to the reference interval for the luminance subband. Therefore, assuming that 1700 is selected as the observation distance, the Visual Weight of “411 frame” in FIG. 12 is read from the VisualWeight table 207. Then, the quantization step number control means 209 calculates each component from the Visual Weight read from the VisualWeight table 207 and the reference quantization step number (FIG. 30) read from the reference quantization step number table 208 according to the equation (S). Determine the modified number of quantization steps for each subband. The corrected number of quantization steps is a value as shown in FIG.

  The following steps 43 to 46 are processing steps that are repeatedly executed for each frame of the input video signal. In the wavelet transform block 201, “normal” wavelet transform is performed for each component of Y, Cb, and Cr for each frame (step 43). A 9 × 7 wavelet transform is used as the wavelet transform. It is assumed that the wavelet transform is performed up to decomposition level 2. Linear quantization for the coefficients of each component and each subband is performed in the quantization block 202 (step 44). Next, the entropy coding block 203 performs MQ coding of all the bit planes of the quantized coefficients (step 45). The generated MQ code is packetized by the packet generation block 204, and a code stream is formed by using the packet by the code formation block 205 (step 46).

  It is also possible to prepare a table storing the corrected quantization step number shown in FIG. 45 in the control block 206 in advance, and read the quantization step number from this table and use it for linear quantization. It is.

As is apparent from the above description, this embodiment is an embodiment of the invention according to claim 5 .

  In this embodiment, a 422 format interlaced video signal (YCbCr value) is encoded on a frame basis.

  FIG. 46 shows a schematic flow of processing in the present embodiment. First, in the control block 206, frame-based encoding is selected (step 51). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for frame-based encoding. However, since the input video signal is a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, in the control block 206, the quantization step number control means 209 reads the reference quantization step number (FIG. 30) from the reference quantization step number table 208, and sets it as it is to the quantization block 202 (step 52).

  Next, the control block 206 sets the number of truncations for the entropy code block 203 (step 53). In this embodiment, “simple” wavelet transform (9 × 7 wavelet transform) is used for subband division, and truncation is set on the assumption that truncation is performed in units of subbit planes. That is, the truncation number control means 210 of the control block 206 reads the Visual Weight calculated by the CSF integration for the reference section shown in (7) of FIG. The reference section shown in (7) of FIG. 29 is shown again in FIG. As can be seen from FIG. 47, it should be noted that the reference section for the color difference vertical subband is the same as the reference section for the color difference horizontal subband in decomposition 2 (or higher). is there. Therefore, if 1700 is selected as the observation distance, the “422 frame” Visual Weight in FIG. 13 is read from the VisualWeight table 207. Then, the truncation number control means 210 converts the Visual Weight read from the VisualWeight table 207 into a truncation number in bit plane units using the equation (BP). However, since the calculated value by the formula (BP) is a real number, it is rounded to an integer. The number of truncations after being converted to an integer is a value as shown in FIG. This truncation number is set in the entropy coding block 203.

  The following steps 54 to 57 are processing steps that are repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “simple” wavelet transform is performed for each of the Y, Cb, and Cr components for each frame (step 54). As the wavelet transform, 9 × 7 wavelet transform is used. It is assumed that the wavelet transform is performed up to decomposition level 2.

  Linear quantization for the coefficients of each subband of each component is performed in quantization block 202 (step 55). In the entropy coding block 203, MQ coding is performed on the coefficients of each subband of each component, excluding lower bit planes corresponding to the set number of truncations (step 56). The generated codes are collected into packets by the packet generation block 204, and a code stream is formed by using the packets by the code formation block 205 (step 57).

  Here, as can be seen from FIG. 48, it should be noted that the number of truncations for the subband of the decomposition level 1 with respect to luminance has a magnitude relationship of “HL = LH <HH”. In this embodiment, linear quantization using the reference quantization step number (FIG. 30) is also performed, but the reference quantization step number also has a magnitude relationship of “HL = LH <HH”. Therefore, it should be noted that the “degree of quantization” by the encoding process, which combines the effects of linear quantization and truncation, also has a magnitude relationship of “HL = LH <HH”.

  It is also possible to prepare a table storing the number of truncations shown in FIG. 48 in advance in the control block 206 and supply the number of truncations read from this table to the bit plane encoding block 203. In this embodiment, truncation is performed in units of bit planes, but truncation in units of sub bit planes may be performed. In this case, the above formula (SBP) is used for conversion from Visual Weight to the number of truncations. Become. Further, the entropy coding block 203 may not perform truncation, but may discard lower bit plane codes for the number of truncations set in the packet generation block 204.

  Here, a supplementary explanation will be given regarding the number of quantization steps used for linear quantization. When the simple wavelet transform is performed as in the present embodiment, whether or not to use the reference quantization step number for linear quantization depends on how the RGB value of the color difference is calculated after decoding. In this embodiment, it is assumed that “simple” wavelet is used for encoding, but “normal” wavelet inverse transform is used for decoding (this is possible by flag operation in the code). That is, at the time of decoding, it is assumed that interpolation of the subsampled color difference is performed by inverse wavelet transform. In this way, when performing “normal” wavelet inverse transformation at the time of decoding, the subband gain and component transformation gain are the same as in the general case where the reference quantization step number shown in FIG. 30 is used. The reference quantization step number of FIG. 30 can be used for linear quantization at the time of encoding.

  In this embodiment, a 422 format interlaced video signal (YCbCr value) is encoded on a field basis.

  FIG. 49 shows a schematic flow of processing in the present embodiment. First, in the control block 206, field-based encoding is selected (step 61). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for field-based encoding. However, since the input video signal has a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, in the control block 206, the quantization step number control means 209 reads the reference quantization step number (FIG. 30) from the reference quantization step number table 208, and sets it in the quantization block 202 (step 62). As described in connection with the sixth embodiment, the present embodiment also assumes that “normal” wavelet inverse transformation is used at the time of decoding.

  Next, the control block 206 sets the number of truncations for the entropy code block 203 (step 63). In this embodiment, “simple” wavelet transform (9 × 7 wavelet transform) is used for subband division, and truncation is set on the assumption that truncation is performed in units of subbit planes. That is, the truncation number control means 210 of the control block 206 reads the Visual Weight calculated by the CSF integration for the reference section shown in (8) of FIG. The reference section shown in (8) of FIG. 29 is shown again in FIG. As can be seen from FIG. 50, the reference interval for the vertical subband of the color difference is shifted to the low frequency side by one decomposition level with respect to the reference interval for the horizontal subband of the color difference. Should be noted. Accordingly, if 1700 is selected as the observation distance, the “422 field” Visual Weight in FIG. 13 is read from the VisualWeight table 207. Then, the truncation number control means 210 converts the Visual Weight read from the VisualWeight table 207 into the number of truncations in sub-bit plane units using the above equation (SBP). However, since the calculated value by the formula (SBP) is a real number, it is rounded to an integer. The number of truncations after the integerization is as shown in FIG. This truncation number is set for the entropy coding block 203.

  The following steps 64 to 67 are processing steps that are repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “simple” wavelet transform is performed on each of the Y, Cb, and Cr components for each field (step 64). As the wavelet transform, 9 × 7 wavelet transform is used. It is assumed that wavelet transform is performed up to composition level 2.

  Linear quantization is performed in the quantization block 202 for each subband coefficient of each component using the reference quantization step number (step 65). In the entropy coding block 203, MQ coding is performed on the quantized coefficients of each subband of each component, excluding lower subbitplanes corresponding to the set number of truncations (step 66). The generated MQ codes are collected into packets by the packet generation block 204, and a code stream is formed using the packets by the code formation block 205 (step 67).

  It is also possible to prepare a table storing the number of truncations shown in FIG. 51 in the control block 206 in advance and set the number of truncations read from this table in the entropy encoding block 203. In this embodiment, truncation is performed in units of sub-bit planes, but truncation in units of bit planes may be performed, and in this case, the above formula (BP) is used for conversion from Visual Weight to the number of truncations. Become. Furthermore, the entropy encoding block 203 may not perform truncation, and the packet generation block 204 may discard the codes of the lower subbit planes corresponding to the set truncation number.

As is apparent from the above description, this embodiment is an embodiment of the invention according to claim 6 .

  In this embodiment, a 411 format interlaced video signal (YCbCr value) is encoded on a field basis.

  FIG. 52 shows a schematic flow of processing in the present embodiment. First, in the control block 206, field-based encoding is selected (step 71). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for field-based encoding. However, since the input video signal has a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, in the control block 206, the quantization step number control means 209 reads the reference quantization step number (FIG. 305) from the reference quantization step number table 208, and sets it for the quantization block 202 (step 72). As described in connection with the sixth embodiment, it is assumed that the “normal” wavelet inverse transform is used in decoding in this embodiment as well.

  Next, the control block 206 sets the number of truncations (step 73). In this embodiment, “simple” wavelet transform (9 × 7 wavelet transform) is used for subband division, and truncation is set on the assumption that truncation is performed in units of subbit planes. That is, the truncation number control means 210 reads the Visual Weight calculated by the CSF integration for the reference section shown in (10) of FIG. The reference section shown in (10) of FIG. 29 is shown again in FIG. As can be seen from FIG. 53, the reference section for the vertical subband of the color difference is shifted to the low frequency side by one decomposition level with respect to the reference section for the horizontal subband of the color difference. Should be noted. Therefore, assuming that 1700 is selected as the observation distance, the Visual Weight of “411 field” in FIG. 14 is read from the VisualWeight table 207. Then, the truncation number control means 210 converts the Visual Weight read from the VisualWeight table 207 into the number of truncations in sub-bit plane units using the above equation (SBP). However, since the calculated value by the formula (SBP) is a real number, it is rounded to an integer. The number of truncations after conversion to an integer is a value as shown in FIG. This truncation number is set in the entropy coding block 203.

  The following steps 74 to 77 are processing steps repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “simple” wavelet transform is performed on each of the Y, Cb, and Cr components for each field (step 74). As the wavelet transform, 9 × 7 wavelet transform is used. It is assumed that the wavelet transform is performed up to decomposition level 2. It should be noted that in this embodiment, wavelet transform, that is, subband division is not performed for the Cb and Cr components only at the composition level 1.

  Linear quantization is performed in the quantization block 202 for each subband coefficient of each component using the reference quantization step number (step 75). In the entropy coding block 203, MQ coding is performed on the quantized coefficients of each subband of each component, excluding lower subbitplanes corresponding to the set number of truncations (step 76). The generated MQ codes are collected into packets by the packet generation block 204, and a code stream is formed using the packets by the code formation block 205 (step 77).

  As can be seen from FIG. 30, the reference quantization step number for luminance used for linear quantization has a magnitude relationship of “HL = LH <HH”. On the other hand, as shown in FIG. 54, the number of truncations for the luminance at decomposition level 1 has a magnitude relationship of “LH <HL = HH”. Therefore, the magnitude relationship of “degree of quantization” with respect to the luminance at decomposition level 1 that combines the effects of linear quantization and truncation is “LH <HL <HH”.

  It is also possible to prepare a table storing the number of truncations shown in FIG. 54 in advance in the control block 206 and set the number of truncations read from this table for the entropy coding block 203. In this embodiment, truncation is performed in units of sub-bit planes, but truncation in units of bit planes may be performed, and in this case, the above formula (BP) is used for conversion from Visual Weight to the number of truncations. Become. Furthermore, the entropy encoding block 203 may not perform truncation, and the packet generation block 204 may discard the codes of lower subbit planes corresponding to the number of truncations.

As is apparent from the above description, this embodiment corresponds to an embodiment of the invention according to claim 7 .

  In this embodiment, a 411 format interlaced video signal (YCbCr value) is encoded on a frame basis.

  FIG. 55 shows a schematic flow of processing in the present embodiment. First, in the control block 206, frame-based encoding is selected (step 81). Accordingly, each block 200 to 205 for compression processing is controlled to perform an operation for frame-based encoding. However, since the input video signal is a YCbCr value, the DC level shift / color conversion block 200 performs neither DC level shift nor color conversion.

  Next, the quantization step number control means 209 of the control block 206 reads the reference quantization step number (FIG. 30) from the reference quantization step number table 208 and sets it for the quantization block 202 (step 82). In this embodiment, “simple” wavelet transform is used. However, as described in connection with the sixth embodiment, it is assumed that “normal” wavelet inverse transform is used at the time of decoding.

  Next, the control block 206 uses the “simple” wavelet transform (9 × 7 wavelet transform) for subband division, and sets the number of truncations on the assumption that truncation is performed in units of sub-bit planes (step 83). That is, the truncation number control means 210 reads the Visual Weight calculated by the CSF integration for the reference section shown in (9) of FIG. The reference section shown in (9) of FIG. 29 is shown again in FIG. As can be seen from FIG. 56, it should be noted that the reference interval for the color difference subband and the reference interval for the luminance subband are the same for the decomposition level 2 (or more). Therefore, assuming that 1700 is selected as the observation distance, the Visual Weight of “411 frame” in FIG. 14 is read from the VisualWeight table 207. Then, the truncation number control means 210 converts the Visual Weight read from the VisualWeight table 207 into the number of truncations in sub-bit plane units using the above equation (SBP). However, since the calculated value by the formula (SBP) is a real number, it is rounded to an integer. The number of truncations after conversion to an integer is a value as shown in FIG. This truncation number is set in the entropy coding block 203.

  The following steps 84 to 87 are processing steps repeatedly executed for each frame of the input video signal.

  In the wavelet transform block 201, “simple” wavelet transform is performed for each of the Y, Cb, and Cr components for each frame (step 84). As the wavelet transform, 9 × 7 wavelet transform is used. It is assumed that the wavelet transform is performed up to decomposition level 2. It should be noted that in this embodiment, wavelet transform, that is, subband division is not performed for the Cb and Cr components only at the composition level 1.

  Linear quantization is performed in the quantization block 202 for each subband coefficient of each component using the reference quantization step number (step 85). In the entropy coding block 203, MQ coding is performed on the quantized coefficients of each subband of each component, excluding lower subbitplanes corresponding to the set number of truncations (step 86). The generated MQ codes are collected into packets by the packet generation block 204, and a code stream is formed using the packets by the code formation block 205 (step 87).

  As can be seen from FIG. 305, the reference quantization step number for luminance used for linear quantization has a magnitude relationship of “HL = LH <HH”. Also, as shown in FIG. 57, the number of truncations for the luminance at decomposition level 1 also has a magnitude relationship of “HL = LH <HH”. Therefore, it should be noted that the magnitude relationship of “degree of quantization” for the luminance subband of decomposition level 1 that combines the effects of linear quantization and truncation is “HL = LH <HH”.

  It is also possible to prepare a table storing the number of truncations shown in FIG. 57 in advance in the control block 206 and supply the number of truncations read from this table to the entropy coding block 203. In this embodiment, truncation is performed in units of sub-bit planes, but truncation in units of bit planes may be performed, and in this case, the above formula (BP) is used for conversion from Visual Weight to the number of truncations. Become. Furthermore, the entropy encoding block 203 may not perform truncation, and the packet generation block 204 may discard the codes of the lower sub-bit planes corresponding to the number of locates.

(Other examples)
Each of the above embodiments has a configuration in which the visual frequency characteristics are accurately reflected on the number of quantization steps of linear quantization and / or the number of truncations of truncation of a coefficient or code based on theoretical knowledge.

  The present invention can also be applied to an encoding device that performs Lagrange rate control. This will be described with reference to FIG.

  Lagrangian rate control is executed by a packet generation block 204 included in the encoding processing means. A distortion slope for Lagrangian rate control is calculated by the entropy coding block 203. In the slope control means 211 of the control block 206, a new distortion slope is calculated by multiplying the calculated distortion slope by the square of the Visual Weight read from the VisualWeight table 207, and this is calculated as the Lagrangian rate control in the packet generation block 204. By using this, Lagrange rate control that accurately reflects the visual frequency characteristics can be performed based on the theoretical knowledge of the present invention. The Visual Weight to be read from the VisualWeight table 207 is selected as described in the above embodiments.

  To describe further, rate control is not only to adjust the total code amount to the target value, but also to code that is performed to obtain the maximum PSNR (Peak Signal-to-Noise Ratio) or the highest image quality at that code amount. For this purpose, for example, in JPEG 2000, the code values are ordered for each code block and each bit plane (more precisely, a sub-bit plane). The code block is a subband divided into rectangles.

Now, when aiming to obtain the maximum PSNR, the PSNR value of the code of each code block is “increment of quantization error when the code is discarded / its code amount ≡ distortion slope”.
Can be expressed as If the error does not increase even if the code is discarded, it is an unnecessary code. If the code is maintained and the error is reduced and the total code amount does not increase, it is a useful code to maintain. The code length which is the denominator of the distortion slope is an amount obtained as actual data at the time of entropy coding. The increment of quantization error, which is a numerator, is an increment of error in units of code blocks, and is usually calculated as an increment of square error (sum of squares of errors) in units of code blocks.

In order to reflect visual frequency characteristics in the distortion slope,
“The square of the Visual Weight of the subband to which the code belongs × (increment of quantization error when the code is discarded / the amount of code) ≡ new distortion slope”
Is calculated. As the Visual Weight used at this time, a value obtained by integrating the CSF for the reference frequency section set according to the theoretical viewpoint of the present invention is used. If the value of the code is given with this new distortion slope and then the known Lagrange rate control is performed, the Lagrange rate control that accurately reflects the visual frequency characteristics can be performed. In Lagrange rate control, codes are ordered in value, so a code with a desired code amount can be obtained by repeating the operation of “discarding low-value codes and leaving high-value codes”. .

  Note that Lagrange rate control itself is a well-known technique, and there are many known materials related to it, such as “D.S. Taubman and M.W. Marcellin,“ JPEG2000: Image Compression Fundamentals, Standards and Practice ”(Kluwer Publishing, 2002)”. Therefore, further detailed explanation of the Lagrange rate control itself is omitted.

  The encoding apparatus according to each of the above embodiments can be realized by using a computer having a general configuration such as a CPU 301, a memory 302, a hard disk device 303, and the like as shown in FIG. . In order to realize this, a program that causes the computer to function as each block shown in FIG. 17 or a program that causes the computer to execute the process according to each block is installed in, for example, the hard disk device 303, and this program is loaded into the memory 302. Then, the CPU 301 is caused to execute.

  In this case, for example, the processing flow is as follows. One frame of the interlaced video signal recorded in the hard disk device 303 is read into the memory 302 by a command from the CPU 301. The CPU 301 executes field-based or frame-based encoding processing as described with respect to the above embodiment on the frame, and the resulting code stream is written into the memory 302. When the encoding process for one frame is completed, the code stream is recorded in the hard disk device 303 according to a command from the CPU 301. By repeating the above processing, all the frames of the original interlaced video signal on the hard disk device 303 are encoded. When field-based / frame-based selection is designated by the user, for example, an attempt is made to encode using a graphical user interface comprising a display device (not shown in FIG. 58) and a user input device such as a mouse. It is possible to specify field base / frame base when specifying a file of an interlaced video signal to be performed.

Such a computer program corresponding to each of the embodiments corresponds to an embodiment of the invention according to claims 8 to 14 .

That is, a program that causes a computer to execute the process steps shown in FIG. 31 according to the first embodiment corresponds to an embodiment according to the eighth aspect . A program for causing a computer to execute the process steps shown in FIG. 34 according to the second embodiment corresponds to an embodiment of the invention according to claim 9 . A program for causing a computer to execute the processing steps shown in FIG. 37 according to the third embodiment corresponds to an embodiment of the invention according to claim 10 . The program executed by the computer in the process steps shown in FIG. 40 according to the fourth embodiment corresponds to one embodiment of the invention according to claim 11 . A program for causing a computer to execute the processing steps shown in FIG. 43 according to the fifth embodiment corresponds to an embodiment of the invention according to claim 12 . A program for causing a computer to execute the processing steps shown in FIG. 49 according to the seventh embodiment corresponds to an embodiment of the invention according to claim 13 . A program for causing a computer to execute the processing steps shown in FIG. 52 according to the eighth embodiment corresponds to an embodiment of the invention according to claim 14 . In these embodiments, the visual frequency characteristic value can be reflected in the Lagrange rate control instead of linear quantization or truncation as described above, and this aspect is also claimed.
It is included in the invention according to 8-14 .

Various information recording (storage) media that can be read by a computer, such as a magnetic disk, an optical disk, a magneto-optical disk, and a semiconductor storage element, on which such a program is recorded are
This corresponds to an embodiment of the invention according to claim 15 .

  When implementing the present invention using a computer, functions (processing steps) of a general JPEG2000 encoding apparatus are implemented as hardware or programs, and other functions (processing steps) unique to the present invention are implemented. Of course, it is also possible to adopt a form in which is implemented as a separate program. The latter program is a function (processing step) related to selection of field-based / frame-based selection, quantization step number, truncation number, or distortion slope in the control block 206, for example. Of course, the present invention includes such a program and a computer-readable information recording (storage) medium in which the program is recorded.

  Although an embodiment using JPEG 2000 as an encoding method has been described, it is clear from the above description that the encoding method is not limited to JPEG 2000 alone.

It is a figure for demonstrating frame-based encoding and field-based encoding. It is a figure which shows the example of a CSF test chart. It is a graph which shows the example of CSF. It is a figure which shows the example of the integration area of CSF regarding a 2LH subband. It is a figure for demonstrating the integral value for every area of CSF. It is explanatory drawing of the "normal" wavelet transformation in the case of carrying out frame base encoding of the 422 format interlace video signal. It is explanatory drawing of the "normal" wavelet transform in the case of carrying out field base encoding of the 422 format interlace video signal. It is explanatory drawing of the "simple" wavelet transform in the case of carrying out frame base encoding of the 422 format interlace video signal. It is explanatory drawing of the "simple" wavelet transformation in the case of carrying out field base encoding of the 422 format interlace video signal. The Visual Weight when a “normal” wavelet transform is used to encode an interlaced video signal in 444 format is shown. The Visual Weight when the “normal” wavelet transform is used to encode the interlaced video signal in the 422 format is shown. The Visual Weight when the “normal” wavelet transform is used to encode the interlaced video signal in the 411 format is shown. The Visual Weight when the “simple” wavelet transform is used to encode the interlaced video signal in the 422 format. The Visual Weight when the “simple” wavelet transform is used to encode the interlaced video signal in the 411 format. The magnitude relationship of the degree of quantization based on the theoretical knowledge of the present invention will be collectively shown. The magnitude relationship of the number of quantization steps for field-based encoding described in Patent Document 3 is shown. It is a block diagram for demonstrating embodiment of this invention. The original image used for description of 5x3 wavelet transform and its coordinate system are shown. The coefficient arrangement | sequence after filtering to a perpendicular direction is shown. The coefficient arrangement | sequence after filtering to a horizontal direction is shown. A deinterleaved coefficient array is shown. A deinterleaved coefficient array after two wavelet transforms is shown. It is a figure which shows the relationship between a decomposition level and a resolution level. It is a figure which illustrates the specific value of a 2LL subband coefficient. FIG. 25 is a diagram showing four bit planes of 2LL subband coefficients shown in FIG. 24. FIG. 26 is a diagram illustrating a sub bit plane of each bit plane illustrated in FIG. 25. The example of the code structure produced | generated by JPEG2000 is shown. CSF reference interval for calculating Visual Weight when “normal” wavelet transform is used. CSF reference interval for calculating Visual Weight when “simple” wavelet transform is used. Indicates the reference quantization step number. It is a flowchart for demonstrating the process in Example 1 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 1. FIG. The quantization step number set in Example 1 is shown. It is a flowchart for demonstrating the process in Example 2 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 2. FIG. The quantization step number set in Example 2 is shown. It is a flowchart for demonstrating the process in Example 3 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 3. FIG. The truncation number set in Example 3 is shown. It is a flowchart for demonstrating the process in Example 4 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 4. FIG. The quantization step number set in Example 4 is shown. It is a flowchart for demonstrating the process in Example 5 of this invention. It is a figure explaining the relationship of the reference area of CSF for calculating Visual Weight used in Example 5. FIG. The quantization step number set in Example 5 is shown. It is a flowchart for demonstrating the process in Example 6 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 6. FIG. The truncation number set in Example 6 is shown. It is a flowchart for demonstrating the process in Example 7 of this invention. It is a figure explaining the relationship of the reference area of CSF for calculating Visual Weight used in Example 7. FIG. The truncation number set in Example 7 is shown. It is a flowchart for demonstrating the process in Example 8 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 8. FIG. The truncation number set in Example 8 is shown. It is a flowchart for demonstrating the process in Example 9 of this invention. It is a figure for demonstrating the relationship of the reference area of CSF for calculating Visual Weight used in Example 9. FIG. The truncation number set in Example 9 is shown. It is a block diagram which shows the structural example of a computer.

200 DC level shift / color conversion block 201 Wavelet conversion block 202 Quantization block 203 Entropy encoding block 204 Packet generation block 205 Code formation block 206 Control block 207 VisualWeight table 208 Reference quantization step number table 209 Quantization step number control means 210 Truncation number control means 211 Slope control means

Claims (15)

An encoding device for subband encoding an interlaced video signal on a field basis,
Subband dividing means for dividing the luminance signal in the interlaced video signal into subbands the same number of times in the horizontal and vertical directions;
Encoding processing means for performing encoding processing of each subband of the luminance signal obtained by subband division by the subband dividing means;
Means for reflecting the luminance visual frequency characteristic value corresponding to the subband calculated based on the luminance visual frequency characteristic in the encoding process of each subband of the luminance signal;
The frequency interval of the luminance visual frequency characteristic referred to for calculating the luminance visual frequency characteristic value is one of the subband divisions with respect to the frequency interval in which the frequency interval referred to in the vertical direction is referred to in the horizontal direction. An encoding device, characterized in that it is set as a section shifted to the low frequency side by the decompression level. An encoding device that performs sub-band encoding of interlaced video signals in 4: 4: 4 format or 4: 1: 1 format on a field basis,
Subband dividing means for dividing the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband dividing means;
Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal;
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. An encoding apparatus, characterized in that it is set as a section shifted to a low frequency side by one decomposition level. A coding device for subband coding a 4: 2: 2 format interlaced video signal on a field basis,
Subband dividing means for dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the luminance signal and the color difference signal obtained by subband division by the subband dividing means;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. An encoding apparatus, wherein the characteristic frequency section is set as a section shifted to a low frequency side by one decomposition level of the subband division. An encoding device that performs subband encoding of a 4: 2: 2 format interlaced video signal on a frame basis,
Subband dividing means for dividing the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband dividing means;
Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal;
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. An encoding device, characterized in that it is set as a section shifted to the high frequency side by one decomposition level. An encoding device that performs subband encoding of a 4: 1: 1 format interlaced video signal on a frame basis,
Subband dividing means for dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal and vertical directions.
Encoding processing means for performing encoding processing of each subband of the luminance signal and the color difference signal obtained by subband division by the subband dividing means;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. An encoding apparatus, wherein the characteristic frequency section is set as a section shifted to a low frequency side by one decomposition level of the subband division. A coding device for subband coding a 4: 2: 2 format interlaced video signal on a field basis,
Subband division means for dividing the color difference signal in the interlaced video signal into subbands only in the vertical direction only at the composition level 1, and subbanding in the horizontal direction and the vertical direction at the decomposition level 2 or higher;
Encoding processing means for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband dividing means;
Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal;
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. The coding apparatus is set as a section shifted to the low frequency side by one decomposition level. An encoding apparatus for subband encoding a 4: 1: 1 format interlaced video signal on a field basis,
Subband division means for dividing a luminance signal and a color difference signal in the interlaced video signal into subbands in a horizontal direction and a vertical direction, but not subband division only for the decomposition level 1 with respect to the color difference signal;
Encoding processing means for performing encoding processing of each subband of the luminance signal and the color difference signal obtained by subband division by the subband dividing means;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. Means for reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. The coding apparatus is set as a section shifted to the low frequency side by one decomposition level. In order to perform field-based subband coding of interlaced video signals,
A subband dividing step of dividing the luminance signal in the interlaced video signal into the same number of times in the horizontal and vertical directions.
An encoding processing step for performing encoding processing of each subband of the luminance signal obtained by subband division by the subband division step;
A program for executing a step of reflecting the luminance visual frequency characteristic value corresponding to the subband calculated based on the luminance visual frequency characteristic in the encoding processing of each subband of the luminance signal,
The frequency interval of the luminance visual frequency characteristic referred to for calculating the luminance visual frequency characteristic value is one of the subband divisions with respect to the frequency interval in which the frequency interval referred to in the vertical direction is referred to in the horizontal direction. A program characterized by being set as a section shifted to the low frequency side by the decompression level. In order to perform field-based subband encoding of interlaced video signals in 4: 4: 4 format or 4: 1: 1 format,
A subband dividing step of dividing the color difference signal in the interlaced video signal into the same number of times in the horizontal and vertical directions.
An encoding processing step for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband division step;
A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the previous color difference signal,
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. A program characterized in that it is set as a section shifted to the low frequency side by one decomposition level. In order to perform 4: 2: 2 format interlaced video signal field-based subband coding,
A subband dividing step of dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal direction and the vertical direction.
An encoding process step for performing an encoding process on each subband of the luminance signal and the color difference signal obtained by subband division in the subband division step;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. A program characterized by being set as a section shifted to a low frequency side by one decomposition level of the subband division with respect to a characteristic frequency section. In order to subframe-encode interlaced video signals in 4: 2: 2 format on a frame basis,
A subband dividing step of dividing the color difference signal in the interlaced video signal by the same number of times in the horizontal direction and the vertical direction.
An encoding processing step for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband division step;
A program for executing a process of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal,
The frequency interval of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is the frequency band referred to in the vertical direction of the subband division with respect to the frequency interval referred to in the horizontal direction. A program characterized in that it is set as a section shifted to the high frequency side by one decompression level. In order to subframe-encode interlaced video signals in 4: 1: 1 format on a frame basis,
A subband dividing step of dividing the luminance signal and the color difference signal in the interlaced video signal by the same number of times in the horizontal direction and the vertical direction.
An encoding process step for performing an encoding process on each subband of the luminance signal and the color difference signal obtained by subband division in the subband division step;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency interval of the color difference visual frequency characteristic referred to in the horizontal direction to calculate the visual frequency characteristic value of the color difference is the luminance visual frequency referred to in the horizontal direction to calculate the visual frequency characteristic value of the luminance. A program characterized by being set as a section shifted to a low frequency side by one decomposition level of the subband division with respect to a characteristic frequency section. In order to perform 4: 2: 2 format interlaced video signal field-based subband coding,
A subband division step of dividing the color difference signal in the interlaced video signal into subbands only in the vertical direction only at the composition level 1, and subbanding in the horizontal direction and the vertical direction at the decomposition level 2 or higher;
An encoding processing step for performing encoding processing of each subband of the color difference signal obtained by subband division by the subband division step;
A program for executing a process of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of the color difference in the encoding process of each subband of the color difference signal,
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. The program is set as a section shifted to the low frequency side by one decomposition level. In order to perform field-based subband encoding of 4: 1: 1 format interlaced video signals,
A subband dividing step of dividing the luminance signal and the color difference signal in the interlaced video signal in the horizontal direction and the vertical direction, but not performing the subband division on the color difference signal only at the decomposition level;
An encoding process step for performing an encoding process on each subband of the luminance signal and the color difference signal obtained by subband division in the subband division step;
The luminance frequency frequency value corresponding to the subband calculated based on the luminance visual frequency characteristic is reflected in the encoding process of each subband of the luminance signal, and the color difference is applied to the encoding process of each subband of the color difference signal. A program for executing a step of reflecting the visual frequency characteristic value of the color difference corresponding to the subband calculated based on the visual frequency characteristic of
The frequency section of the color difference visual frequency characteristic referred to in order to calculate the color difference visual frequency characteristic value is a frequency section referred to in the vertical direction, and the frequency band referred to in the horizontal direction is the subband division. The program is set as a section shifted to the low frequency side by one decomposition level. 15. A computer-readable information recording medium on which the program according to claim 8 is recorded.

Images