JP3763163B2 - Image encoding transmission apparatus and image encoding transmission method - Google Patents

Description

【００５０】
そして、最適化されたダイナミックレンジＤＲ´および最適化された最大値ＭＡＸ´は、フレーミング部１１に供給される。フレーミング部１１は、ブロック毎に量子化部７からの量子化値ｘ、最小自乗法推定部９からの最適化されたダイナミックレンジＤＲ´および最適化された最大値ＭＡＸ´をフレーミング化する。
【００５１】
図５は、量子化値とともに最適化された最大値ＭＡＸ´および最適化された最小値ＭＩＮ´を伝送する第３の実施例の画像符号化装置を示している。なお、第３の実施例を説明するにあたり、第１の実施例と同一の部分には同一符号を付し、その説明は省略する。
【００５２】
減算部５では、図１の実施例と同様に、（３×３）ブロックの９画素の８ビットからなる入力画素値ｙから最小値ＭＩＮがそれぞれ減算され、９画素の画素値が正規化される。正規化された９画素の画素値は、量子化部７に供給される。量子化部７は、（３×３）ブロックの９画素の４ビットからなる量子化値を最小自乗法推定部１０およびフレーミング部１１にそれぞれ供給する。
【００５３】
そして、最小自乗法推定部１０には、入力画像信号である（３×３）ブロックの９画素の入力画素値ｙと量子化された（３×３）ブロックの９画素の画素の量子化値ｘが供給され、この最小自乗法推定部１０は、ブロック毎に供給された入力画素値（真値）ｙおよび量子化値ｘを用いて、最小自乗法により、各ブロックの量子化値の復号値ｙ´と復号値ｙ´に対応する真値との誤差（ｙ´−ｙ）の自乗和を最小とするような最適化された最小値ＭＩＮ´および最適化された最大値ＭＡＸ´を推定する。つまり、入力画素値ｙの復号値ｙ´および量子化値ｘの間には、次式（７）の関係が成立する。
【００５４】
ｙ´＝ｘ×（ＭＡＸ−ＭＩＮ）／ｎ＋ＭＩＮ （７）
（但し、ｎは、量子化ビット数である。）
【００５５】
そして、ブロック内の画素数をｍとすると、最適化された最小値ＭＩＮ´は、上述した式（３）を用いて、上述した式（４）によって求められる。また、最適化された最大値ＭＡＸ´は、上述した式（３）を用いて、上述した式（６）によって求められる。そして、最適化された最小値ＭＩＮ´および最適化された最大値ＭＡＸ´は、フレーミング部１１に供給される。フレーミング部１１は、ブロック毎に量子化部７からの量子化値ｘ、最小自乗法推定部１０からの最適化された最小値ＭＩＮ´および最適化された最大値ＭＡＸ´をフレーミング化する。
【００５６】
次に、図６は、この発明の画像符号化装置を階層符号化装置に適応した一例を示している。入力端子１０１からの入力画像信号が、平均化部１０２および減算部１０４にそれぞれ供給される。平均化部１０２では、例えば、供給された入力画像信号を（２×２）ブロック単位で４つの画素を加算し、その加算値を１／４に平均化する。平均化された画素は、補間部１０３および高能率符号化部１０８にそれぞれ供給される。
【００５７】
補間部１０３では、例えば、クラス分類適応処理技術を用いて、平均化することにより間引かれた画素の補間処理を行い、その画素は、減算器１０４へ供給される。このように、クラス分類適応処理技術を用いた補間方法に関して、例えば、本出願人による特開平５−３２８１８５号公報において提案されている。また、この補間部１０３は、クラス分類適応処理技術を用いた方法以外のすでに公知である補間方法を用いてもよい。
【００５８】
減算部１０４では、入力画素値ｙから補間処理が行われた画素が減算され、その残差信号が、高能率符号化部１０５に供給される。この高能率符号化部１０５は、上述したこの発明の実施例と同様にダイナミックレンジＤＲ、最小値ＭＩＮおよび最大値ＭＡＸのうち少なくとも２つのパラメータを最適化するようにしたＡＤＲＣによる高能率符号化部である。
【００５９】
すなわち、高能率符号化部１０５では、量子化値ｘと付加コード（例えば、第１の実施例において、最適化されたダイナミックレンジＤＲ´、最適化された最小値ＭＩＮ´）が生成され、生成された量子化値ｘと付加コードが可変長符号化部１０６に供給される。可変長符号化部１０６において、供給された量子化値ｘと付加コードがハフマン符号化、ランレングス符号化等の可変長符号化がなされ、その可変長符号化データが出力端子１０７から下位階層の符号化データとして伝送される。
【００６０】
同様に高能率符号化部１０８では、平均化部１０２から供給された平均化された画素が、高能率符号化部１０８に供給される。この高能率符号化部１０８は、上述したこの発明の実施例と同様にダイナミックレンジＤＲ、最小値ＭＩＮおよび最大値ＭＡＸのうち少なくとも２つのパラメータを最適化するようにしたＡＤＲＣによる高能率符号化部である。
【００６１】
すなわち、高能率符号化部１０８では、量子化値ｘと付加コード（例えば、第１の実施例において、最適化されたダイナミックレンジＤＲ´、最適化された最小値ＭＩＮ´）が生成され、生成された量子化値ｘと付加コードが可変長符号化部１０９に供給される。可変長符号化部１０９において、伝送された量子化値ｘと付加コードがハフマン符号化、ランレングス符号化等の可変長符号化がなされ、その可変長符号化データが出力端子１１０から上位階層の符号化データとして伝送される。
【００６２】
この発明の第１、第２および第３の実施例では、伝送されるパラメータとして、最適化されたダイナミックレンジＤＲ´、最適化された最小値ＭＩＮ´および最適化された最大値ＭＡＸ´のうち、少なくとも２つのパラメータを伝送しているが、この発明はこれに限らず、最適化された量子化ステップ幅Δ´を少なくとも含む２つのパラメータを伝送するようにしてもよい。
【００６３】
つまり、この発明において、最小自乗法推定部において、少なくとも伝送される２つのパラメータを最適化して、その最適化された２つのパラメータを量子化値とともに伝送すればよい。
【００６４】
また、この発明の第１および第３の実施例では、減算部５において、９画素の量子化値から最小値ＭＩＮを減算しているが、この発明はこれに限らず、第２の実施例と同様に、最大値ＭＡＸから９画素の入力画素値を減算するようにしてもよい。
【００６５】
また、この発明の第２の実施例では、減算部１９において、最大値ＭＡＸから９画素の量子化値を減算しているが、この発明はこれに限らず、第１の実施例と同様に、９画素の量子化値から最小値ＭＩＮを減算するようにしてもよい。
【００６６】
また、この発明のブロック分割部は、入力画像信号を（３×３）ブロックの２次元ブロックに分割しているが、この発明はこれに限らず、（３×３）ブロック以外のブロックで構成されてもよいし、また、入力画像信号を３次元ブロックで構成してもよい。
【００６７】
さらに、この発明の実施例では、ＡＤＲＣを用いたが、この発明は、これに限らずＤＣＴ（Discrete Cosine Transform ）を用いる場合、ＤＣ成分に対して最適化を行い、ＧＢＴＣ（Grobal Block Trancation Coding）を用いる場合、平均値および標準偏差に対して最適化を行うことも可能である。
【００６８】
また、この発明の実施例は、ディジタル画像信号に対する符号化装置であるが、この発明はこれに限らず、ディジタルオーディオ信号等のデータに対して適用することも可能である。
【００６９】
なお、この発明の主旨を逸脱しない範囲において、様々な変形や応用例が考えうる。したがって、この発明の要旨は、実施例に限定されるものではない。
【００７０】
【発明の効果】
この発明は、原信号値と復元値の復号誤差が小さくなるように、符号化側において、量子化値を求めるために最初に求められたパラメータを最小自乗法を用いて最適化し、その最適化したパラメータと量子化値とを伝送している。従って、この発明に依れば、復号側において、復元値と原信号値との誤差を小さくすることができる。
【図面の簡単な説明】
【図１】この発明の画像符号化装置の第１の実施例を示すブロック図である。
【図２】この発明に係る最小自乗法推定部の構成例を示す図である。
【図３】この発明に係る最小自乗法推定部の構成例を示す図である。
【図４】この発明の画像符号化装置の第２の実施例を示すブロック図である。
【図５】この発明の画像符号化装置の第３の実施例を示すブロック図である。
【図６】この発明の画像符号化装置を階層符号化装置に用いた一例を示すブロック図である。
【図７】従来のＡＤＲＣのブロック図である。
【符号の説明】
１・・・ブロック分割部、２・・・最大値検出部、３・・・最小値検出部、４、５・・・減算部、６・・・ステップ幅算出部、７・・・量子化部、８・・・最小自乗法推定部、１１・・・フレーミング部、１２・・・エラー訂正符号付加部、１３・・・変調部、１４・・・記録部、１５・・・記録媒体、１６・・・伝送部、１７・・・伝送路[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image encoding device, an image encoding method, an image encoding transmission method, and an image recording medium for encoding a digital image signal in order to compress the data amount of the digital image signal. Encoding and transmitting apparatus for encoding and transmitting additional information together with encoded information and The present invention relates to an image encoding transmission method.
[0002]
[Prior art]
FIG. 7 shows a block diagram of an example of an image encoding apparatus that compresses a digital image signal. The image encoding apparatus shown in FIG. 7 performs ADRC (Adaptive Dynamic Renge Coding) that adaptively encodes pixels in a block in accordance with the dynamic range of the block by dividing the input image signal into predetermined blocks. It is the encoding apparatus used.
[0003]
ADRC is proposed in Japanese Patent Application Laid-Open No. 61-14498 by the present applicant. ADRC will be briefly described with reference to FIG. An input image signal input from the input terminal 120 is supplied to the block dividing unit 121. Then, in the block dividing unit 121, the input image signal is divided into blocks including nine pixels of 3 pixels × 3 lines (hereinafter referred to as (3 × 3)), for example. Then, an output signal from the block dividing unit 121 is supplied to the maximum value detecting unit 122 and the minimum value detecting unit 123 for each block.
[0004]
The maximum value detecting unit 122 detects the maximum value MAX of the pixel values included in the block, and the minimum value detecting unit 123 detects the minimum value MIN. The maximum value MAX from the maximum value detection unit 122 is supplied to the subtraction unit 124. Further, the minimum value MIN from the minimum value detection unit 123 is supplied to the subtraction unit 124, the subtraction unit 125, and the framing unit 128, respectively.
[0005]
In the subtraction unit 124, the minimum value MIN is subtracted from the maximum value MAX to generate the dynamic range DR. The dynamic range DR is supplied to the quantization step width calculation unit 126 and the framing unit 128, respectively. The quantization step width calculation unit 126 calculates the quantization step width Δ from the dynamic range DR supplied from the subtraction unit 124, and supplies the calculated quantization step width Δ to the quantization unit 127.
[0006]
The subtractor 125 is supplied with nine pixels of the (3 × 3) block from the block dividing unit 121, and subtracts the minimum value MIN from the pixel value of each of the nine pixels, thereby normalizing each pixel value. Is executed. The normalized pixel values are supplied to the quantization unit 127. In the quantization unit 127, each normalized pixel value is quantized with the quantization step width Δ and supplied to the framing unit 128 as a quantization value x.
[0007]
In the framing unit 128, for each block, the supplied dynamic range DR and the minimum value MIN are framing together with the quantized value x of 9 pixels of the block as a block parameter, and output as an output signal. . Then, the output signal is recorded on a recording medium such as a disk or transmitted through a transmission path.
[0008]
[Problems to be solved by the invention]
However, on the decoding side, when decoding the quantized value of the block using the parameter of the block to be transmitted, in the sense that the decoding error between the original signal value and the decoded restored value is minimized, There is no guarantee that it is optimal to decode the quantized value of the block using the parameter of the block obtained in (1), and in some cases, the decoding error becomes large and the decoded image deteriorates.
[0009]
Accordingly, an object of the present invention is to provide a parameter (for example, in the case of ADRC) first obtained for obtaining a quantized value on the encoding side so that a decoding error between the original signal value and the restored value becomes small. Image coding and transmission apparatus capable of reducing decoding error by optimizing maximum value MAX, minimum value MIN or dynamic range DR) and An object of the present invention is to provide an image encoding transmission method.
[0010]
[Means for Solving the Problems]
In order to achieve the above-described object, the present invention provides a plurality of input digital image signals in an image encoding transmission apparatus that encodes and transmits an input digital image signal so as to reduce the amount of generated data of the input digital image signal. A block dividing unit that divides the block into pixels, a maximum value of a plurality of pixels in the block, a first detection unit that detects a minimum value of the plurality of pixels, and a maximum value and a minimum value A second detection unit for detecting a dynamic range as a difference; Difference value obtained by subtracting the pixel values of multiple pixels from the maximum value, or difference value obtained by subtracting the minimum value from the pixel values of multiple pixels A quantization unit that generates a quantization value of a plurality of pixels, and a maximum value, a minimum value, and a dynamic value generated for each block so as to minimize a sum of squares of decoding errors of the quantization values. An optimization unit that optimizes at least two pieces of information indicating a range, and two pieces of information that indicate an optimized maximum value, an optimized minimum value, and an optimized dynamic range as parameters. An image encoding transmission apparatus comprising: a transmission unit that transmits to a transmission path together with quantized values of a plurality of pixels.
[0011]
The present invention also relates to an image encoding transmission method for encoding and transmitting an input digital image signal so as to reduce the amount of generated data of the input digital image signal, and dividing the input digital image signal into blocks composed of a plurality of pixels. For each block, detecting a maximum value of a plurality of pixels in the block, a minimum value of the plurality of pixels, and detecting a dynamic range that is a difference between the maximum value and the minimum value; Difference value obtained by subtracting the pixel values of multiple pixels from the maximum value, or difference value obtained by subtracting the minimum value from the pixel values of multiple pixels To generate a quantized value of a plurality of pixels, and the maximum value, minimum value, and dynamic range generated for each block so that the sum of squares of decoding errors of the quantized value is minimized. A plurality of pixels for each block, with two parameters: a step of optimizing at least two of the information to be shown, and an information showing the optimized maximum value, the optimized minimum value, and the optimized dynamic range. And a step of transmitting to the transmission path together with the quantized value of the image encoding transmission method.
[0013]
As described above, according to the present invention, after obtaining a block parameter, decoding is performed using the parameter to obtain an error between the restored value and the true value, and the parameter is modified so as to minimize the error. Therefore, it is possible to further reduce the restoration error between the original signal value and the restoration value.
[0014]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a first embodiment of an image encoding apparatus according to the present invention. The input image signal is supplied to the block dividing unit 1 through the input terminal IN. In the block division unit 1, the input image signal is divided into, for example, (3 × 3) blocks, and an input pixel value y consisting of 8 bits of 9 pixels of the (3 × 3) block is the maximum value detection unit 2. Value detector 3 Are supplied to the subtracting unit 5 and the least square method estimating unit 8, respectively.
[0015]
In the maximum value detection unit 2, a value at which the level of the pixel value is maximum is detected among the nine pixels included in the (3 × 3) block, and the maximum value MAX is supplied to the subtraction unit 4. In addition, the minimum value detection unit 3 detects a value at which the level of the pixel value is the minimum among the nine pixels included in the (3 × 3) block, and the minimum value MIN is sent to the subtraction unit 4 and the subtraction unit 5. Supplied respectively.
[0016]
In the subtraction unit 4, the minimum value MIN is subtracted from the maximum value MAX to generate a dynamic range DR. The dynamic range DR is supplied to the quantization step width calculation unit 6. The subtracting unit 5 subtracts the minimum value MIN from the input pixel value y of 9 pixels of the (3 × 3) block from the block dividing unit 1 and normalizes the 9 pixels of the normalized (3 × 3) block. Pixel values are generated. The pixel values of the 9 pixels of the normalized (3 × 3) block are supplied to the quantization unit 7.
[0017]
The quantization step width calculation unit 6 calculates a quantization step width Δ based on the supplied dynamic range DR, and the calculated quantization step width Δ is supplied to the quantization unit 7. The quantization unit 7 performs 4-bit quantization of the pixel values of 9 pixels of the (3 × 3) block normalized by the supplied quantization step width Δ. The quantized value x consisting of 4 bits of 9 pixels of the (3 × 3) block is supplied to the least square method estimating unit 8 and the framing unit 11, respectively.
[0018]
The least square method estimating unit 8 receives the input pixel value y of 9 pixels of the (3 × 3) block and the quantized value x of 9 pixels of the quantized (3 × 3) block, which are input image signals. The least square method estimator 8 is supplied and uses the input pixel value (true value) y and the quantized value x supplied for each block, and the decoded value y of the quantized value of each block by the least square method. The optimized dynamic range DR ′ and the optimized minimum value MIN ′ are estimated so as to minimize the sum of squares of the error (y′−y) between “′” and the true value y corresponding to the decoded value y ′. . That is, the relationship of the following expression (1) is established between the decoded value y ′ of the input pixel value y and the quantized value x.
[0019]
y ′ = x × DR / n + MIN (1)
(Where n is the number of quantization bits)
[0020]
Then, the optimized dynamic range DR ′ is obtained by the following equation (2).
[0021]
DR ′ = n · Δ ′ (2)
(However, Δ ′ is an optimized quantization step width.)
[0022]
Further, when the number of pixels in the block is m, the optimized quantization step width is obtained by the following equation (3).
[0023]
Δ ′ = (m · Σxy−Σx · Σy) / (m · Σx ² -(Σx) ² (3)
[0024]
Further, the optimized minimum value MIN ′ is obtained by the following equation (4).
[0025]
MIN ′ = (Σy−Δ ′ · Σx) / m (4)
[0026]
Then, the optimized dynamic range DR ′ and the optimized minimum value MIN ′ are supplied to the framing unit 11. The framing unit 11 frames the quantized value x from the quantizing unit 7, the optimized dynamic range DR ′ and the optimized minimum value MIN ′ from the least square method estimating unit 8 for each block, and generates a framing signal. Is supplied to the error correction code adding unit 12.
[0027]
The error correction code adding unit 12 adds an error correction code to the framing signal and supplies it to the modulation unit 13. The modulation unit 13 modulates the framing signal to which the error correction code is added using a modulation method such as an EFM modulation method. Then, the modulated signal is supplied to the recording unit 14 where it is recorded on a recording medium 15 such as a disk.
[0028]
In the case of transmitting a modulated signal via the transmission line 17, the present invention is configured by the transmission unit 16 instead of the recording unit 14. Then, the modulation unit 13 modulates the framing signal to which the error correction code is added using a modulation method optimal for transmission, and supplies the modulated framing signal to the transmission path 17 via the transmission unit 16. As for the framing technique and the modulation technique, since many techniques are already known, any technique may be used although omitted here.
[0029]
Next, details of the least square method estimating unit 8 of the first embodiment will be described with reference to FIGS. FIG. 3 is an actual configuration example for obtaining the dynamic range DR ′ optimized by the least square method and the optimized minimum value MIN ′, and FIG. 2 calculates constants required for the configuration. It is the example of a structure to do.
[0030]
As described above, the data x used in the first embodiment is a quantized value, and the data y ′ is a restored value of the data x. The true value of the data y ′ is y, and the number of pixels in the block is m. Therefore, the least square method estimating unit 8 uses the supplied pixel value y and quantized value x for each block, and uses the least square method to calculate the decoded value y ′ and the decoded value y of each block. Optimized minimum value MIN ′ and optimized dynamic range DR ′ that minimize the sum of squares of error (y′−y) from the true value corresponding to ′ are estimated.
[0031]
First, data x is supplied from the input terminal 21 shown in FIG. 2, and the data x is supplied to the register 22. Data y is supplied to the register 24 via the input terminal 23. The counter 25 counts the number of pixels for each block in accordance with the clock, and the counted number of pixels m is supplied to the terminal 28 via the registers 26 and 27.
[0032]
The multiplier 29 multiplies the 4-bit data x supplied from the register 22 by the 8-bit data y supplied from the register 24, and the 12-bit data xy is added through the register 30. 31. In the adding circuit 31, the supplied 12-bit data xy is supplied to the adder 31a. Data output from the adder 31a is supplied to the registers 31b and 31c. The data xy supplied from the register 30 and the data output from the adder 31a are added by the adder 31a via the register 31b. 18-bit data is output from the adder circuit 31 and supplied to the terminal 33 via the register 32 as data Σxy. That is, the adder circuit 31 generates Σxy in units of blocks.
[0033]
The 4-bit data x supplied from the register 22 is supplied to the adder circuit 35 via the register 34. In the adding circuit 35, the supplied 4-bit data x is supplied to the adder 35a. Data output from the adder 35a is supplied to the registers 35b and 35c. The data x supplied from the register 34 and the data output from the adder 35a are added by the adder 35a via the register 35b. 10-bit data is output from the adder circuit 35 and supplied to the terminal 37 via the register 36 as data Σx. That is, the adder circuit 35 generates Σx in units of blocks.
[0034]
The 10 bits of data Σx from the adder circuit 35 are supplied to the multiplier 38. The multiplier 38 calculates the square of the data Σx, and data (Σx) consisting of 20 bits. ² Is supplied to the terminal 40 via the register 39.
[0035]
The 8-bit data y supplied from the register 24 is supplied to the adder circuit 42 via the register 41. In the adder circuit 42, the supplied 8-bit data y is supplied to the adder 42a. Data output from the adder 42a is supplied to the registers 42b and 42c. The data y supplied from the register 41 and the data output from the adder 42a are added by the adder 42a via the register 42b. 14-bit data is output from the adder circuit 42 and supplied to the terminal 44 through the register 43 as data Σy. That is, the adder circuit 42 generates Σy in units of blocks.
[0036]
The multiplier 45 calculates the square of the data x consisting of 4 bits supplied from the register 22, and the data x consisting of 8 bits. ² Is supplied to the adder circuit 47 via the register 46. In this adder circuit 47, the supplied 8-bit data x ² Is supplied to the adder 47a. Data output from the adder 47a is supplied to the registers 47b and 47c. Data x supplied from the register 46 ² Then, the data output from the adder 47a is added by the adder 47a via the register 47b. The adder circuit 47 outputs 14-bit data, and data Σx ² Is supplied to the terminal 49 via the register 48. In other words, the adder circuit 47 generates Σx in block units. ² Is generated.
[0037]
Further, as shown in FIG. 2, the input terminal 50 is connected to the counter 25 and the clear terminals of the registers 31b, 35b, 42b and 47b, and the input terminal 51 is the clear of the registers 27, 32, 36, 39, 43 and 48. Each register connected to the terminal is controlled for each block by a signal supplied to these input terminals 50 and 51.
[0038]
FIG. 3 shows a general configuration example of an actual least square method. The number m of pixels supplied to the terminal 28 is supplied to the reciprocal circuit 61, the multiplier 64, and the multiplier 71, respectively. In the reciprocal circuit 61, the reciprocal number (1 / m) of the supplied pixel number m is calculated, and the reciprocal number (1 / m) is supplied to the terminal 63 via the register 62. In the multiplier 64, the data Σxy supplied to the terminal 33 is multiplied by the number of pixels m, and the data mΣxy is supplied to the subtractor 66 through the register 65.
[0039]
The Σx supplied to the terminal 37 is supplied to the multiplier 67 and the multiplier 74, respectively, and the data Σy supplied to the terminal 44 is supplied to the multiplier 67. In the multiplier 67, the data Σx and the data Σy are multiplied, and the data ΣxΣy is supplied to the subtractor 66 through the register 68. The subtractor 66 subtracts the data ΣxΣy from the data mΣxy, and the output data mΣxy−ΣxΣy is supplied to the multiplier 79 via the registers 69 and 70.
[0040]
In the multiplier 71, the data Σx supplied to the terminal 47 ² Is multiplied by the number of pixels m, and the data mΣx is transmitted via the register 72. ² Is supplied to the subtractor 73. The multiplier 74 calculates the square of the data Σx, and the data (Σx) ² Is supplied to the subtractor 73. In the subtractor 73, the data mΣx ² To data (Σx) ² Is subtracted and the output data mΣx ² -(Σx) ² Is supplied to the reciprocal circuit 77 through the register 76. In the reciprocal circuit 77, the data mΣx is the same as described above. ² -(Σx) ² Reciprocal of 1 / (mΣx ² -(Σx) ² )) And the reciprocal (1 / (mΣx) through the register 78. ² -(Σx) ² )) Is supplied to the multiplier 79.
[0041]
In the multiplier 79, the data mΣxy−ΣxΣy and the data 1 / (mΣx ² -(Σx) ² ) And data (mΣxy−ΣxΣy) / (mΣx ² -(Σx) ² ) Are supplied to the multipliers 81 and 89 via the register 80, respectively. In the multiplier 81, the data Σx supplied to the terminal 37 and the data (mΣxy−ΣxΣy) / (mΣx ² -(Σx) ² ) And the result is the data (mΣxΣxy− (Σx) ² Σy) / (mΣx ² -(Σx) ² ) Is supplied to the subtractor 83 via the register 82.
[0042]
In the subtractor 83, the data (mΣxΣxy− (Σx)) from the data Σy supplied to the terminal 44 from the register 82 ² Σy) / (mΣx ² -(Σx) ² ) Is subtracted, and as a result, the data m (Σx is transmitted through the register 84. ² Σy−ΣxΣxy) / (mΣx ² -(Σx) ² ) Is supplied to the multiplier 85. In the multiplier 85, 1 / m from the terminal 63 and data m (Σx ² Σy−ΣxΣxy) / (mΣx ² -(Σx) ² ) And result in data (Σx ² Σy−ΣxΣxy) / (mΣx ² -(Σx) ² ) Is the optimized minimum value MIN ′. Then, the generated minimum value MIN ′ is supplied to the output terminal 87 via the register 86 and output.
[0043]
The terminal 88 has a value 2 set in advance corresponding to the number of quantization bits used in the quantization step width calculation unit 6. ⁿ Is supplied. In the multiplier 89, 2 supplied to the terminal 88. ⁿ (N is the number of quantization bits) and data (mΣxy−ΣxΣy) / (mΣx ² -(Σx) ² ) And result in data 2 ⁿ (MΣxy−ΣxΣy) / (mΣx ² -(Σx) ² ) Is the optimized dynamic range DR ′. The generated dynamic range DR ′ is supplied to the output terminal 93 via the registers 90, 91 and 92 and output. In this way, an optimized minimum value MIN ′ and an optimized dynamic range DR ′ can be obtained.
[0044]
FIG. 4 shows an image coding apparatus according to a second embodiment that transmits the maximum value MAX ′ optimized together with the quantization value and the optimized dynamic range DR ′. In the description of the second embodiment, the same parts as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted.
[0045]
The subtracting unit 19 subtracts the 8-pixel input pixel value y of 9 pixels of the (3 × 3) block from the maximum value MAX, and normalizes the 9-pixel input pixel value. The normalized pixel value of 9 pixels is supplied to the quantization unit 7. The quantization unit 7 supplies quantized values composed of 4 bits of 9 pixels of a (3 × 3) block to the least square method estimation unit 9 and the framing unit 11, respectively.
[0046]
Then, the least square method estimating unit 9 inputs the input pixel value y of the 9 pixels of the (3 × 3) block and the quantized value of the 9 pixels of the quantized (3 × 3) block, which are input image signals. x is supplied. The least square method estimating unit 9 uses the input pixel value (true value) y and the quantized value x supplied for each block, and decodes the decoded value y ′ of the quantized value of each block by the least square method. The optimized dynamic range DR ′ and the optimized maximum value MAX ′ are estimated so as to minimize the square sum of the error (y′−y) from the true value corresponding to the value y ′. That is, the relationship of the following equation (5) is established between the decoded value y ′ of the input pixel value y and the quantized value x.
[0047]
y ′ = MAX−x × DR / n (5)
(Where n is the number of quantization bits)
[0048]
When the number of pixels in the block is m, the optimized dynamic range DR ′ is obtained by the above-described equation (2) using the above-described equation (3). Further, the optimized maximum value MAX ′ is obtained by the following equation (6) using the above-described equation (3).
[0049]

[0050]
The optimized dynamic range DR ′ and the optimized maximum value MAX ′ are supplied to the framing unit 11. The framing unit 11 frames the quantized value x from the quantizing unit 7, the optimized dynamic range DR ′ from the least square method estimating unit 9, and the optimized maximum value MAX ′ for each block.
[0051]
FIG. 5 shows an image coding apparatus according to a third embodiment that transmits the optimized maximum value MAX ′ and the optimized minimum value MIN ′ together with the quantized value. In the description of the third embodiment, the same parts as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted.
[0052]
As in the embodiment of FIG. 1, the subtracting unit 5 subtracts the minimum value MIN from the input pixel value y consisting of 8 bits of 9 pixels of the (3 × 3) block, and normalizes the pixel value of 9 pixels. The The normalized pixel value of 9 pixels is supplied to the quantization unit 7. The quantization unit 7 supplies a quantized value consisting of 4 bits of 9 pixels of a (3 × 3) block to the least square method estimation unit 10 and the framing unit 11, respectively.
[0053]
Then, the least square method estimating unit 10 inputs the input pixel value y of the 9 pixels of the (3 × 3) block and the quantized value of the 9 pixels of the quantized (3 × 3) block, which are input image signals. x is supplied, and the least square method estimation unit 10 decodes the quantized value of each block by the least square method using the input pixel value (true value) y and the quantized value x supplied for each block. An optimized minimum value MIN ′ and an optimized maximum value MAX ′ are estimated so as to minimize the sum of squares of an error (y′−y) between the value y ′ and the true value corresponding to the decoded value y ′. To do. That is, the relationship of the following equation (7) is established between the decoded value y ′ of the input pixel value y and the quantized value x.
[0054]
y ′ = xx × (MAX−MIN) / n + MIN (7)
(Where n is the number of quantization bits)
[0055]
Then, assuming that the number of pixels in the block is m, the optimized minimum value MIN ′ is obtained by the above-described equation (4) using the above-described equation (3). Further, the optimized maximum value MAX ′ is obtained by the above-described equation (6) using the above-described equation (3). Then, the optimized minimum value MIN ′ and the optimized maximum value MAX ′ are supplied to the framing unit 11. The framing unit 11 frames the quantized value x from the quantizing unit 7, the optimized minimum value MIN ′ and the optimized maximum value MAX ′ from the least square method estimating unit 10 for each block.
[0056]
Next, FIG. 6 shows an example in which the image coding apparatus of the present invention is applied to a hierarchical coding apparatus. Input image signals from the input terminal 101 are supplied to the averaging unit 102 and the subtracting unit 104, respectively. For example, the averaging unit 102 adds four pixels to the supplied input image signal in units of (2 × 2) blocks, and averages the added value to ¼. The averaged pixels are supplied to the interpolation unit 103 and the high efficiency encoding unit 108, respectively.
[0057]
In the interpolation unit 103, for example, using the class classification adaptive processing technique, the interpolation processing of the pixels thinned out by averaging is performed, and the pixels are supplied to the subtractor 104. As described above, an interpolation method using the class classification adaptive processing technique is proposed in, for example, Japanese Patent Laid-Open No. 5-328185 by the present applicant. The interpolation unit 103 may use a known interpolation method other than the method using the class classification adaptive processing technique.
[0058]
The subtracting unit 104 subtracts the pixel subjected to the interpolation process from the input pixel value y, and supplies the residual signal to the high efficiency encoding unit 105. This high-efficiency encoding unit 105 is an ADRC-based high-efficiency encoding unit that optimizes at least two parameters of the dynamic range DR, the minimum value MIN, and the maximum value MAX as in the above-described embodiment of the present invention. It is.
[0059]
That is, the high-efficiency encoding unit 105 generates and generates a quantization value x and an additional code (for example, the optimized dynamic range DR ′ and the optimized minimum value MIN ′ in the first embodiment). The quantized value x and the additional code are supplied to the variable length coding unit 106. In the variable length coding unit 106, the supplied quantization value x and the additional code are subjected to variable length coding such as Huffman coding and run length coding, and the variable length coded data is sent from the output terminal 107 to the lower layer. It is transmitted as encoded data.
[0060]
Similarly, in the high efficiency encoding unit 108, the averaged pixels supplied from the averaging unit 102 are supplied to the high efficiency encoding unit 108. The high-efficiency encoding unit 108 is an ADRC-based high-efficiency encoding unit that optimizes at least two parameters of the dynamic range DR, the minimum value MIN, and the maximum value MAX, as in the above-described embodiment of the present invention. It is.
[0061]
That is, the high-efficiency encoding unit 108 generates and generates a quantization value x and an additional code (for example, the optimized dynamic range DR ′ and the optimized minimum value MIN ′ in the first embodiment). The quantized value x and the additional code are supplied to the variable length encoding unit 109. In the variable length coding unit 109, the transmitted quantized value x and the additional code are subjected to variable length coding such as Huffman coding and run length coding, and the variable length coded data is sent from the output terminal 110 to the upper layer. It is transmitted as encoded data.
[0062]
In the first, second, and third embodiments of the present invention, as the parameters to be transmitted, among the optimized dynamic range DR ′, the optimized minimum value MIN ′, and the optimized maximum value MAX ′ However, the present invention is not limited to this, and two parameters including at least the optimized quantization step width Δ ′ may be transmitted.
[0063]
That is, in the present invention, the least square method estimation unit may optimize at least two parameters to be transmitted and transmit the two optimized parameters together with the quantized value.
[0064]
In the first and third embodiments of the present invention, the subtraction unit 5 subtracts the minimum value MIN from the quantized value of 9 pixels. However, the present invention is not limited to this, and the second embodiment. Similarly to the above, the input pixel value of 9 pixels may be subtracted from the maximum value MAX.
[0065]
In the second embodiment of the present invention, the subtractor 19 subtracts the quantized value of 9 pixels from the maximum value MAX. However, the present invention is not limited to this, and is the same as in the first embodiment. The minimum value MIN may be subtracted from the quantized value of 9 pixels.
[0066]
The block division unit of the present invention divides the input image signal into two-dimensional blocks of (3 × 3) blocks. However, the present invention is not limited to this, and is composed of blocks other than (3 × 3) blocks. Alternatively, the input image signal may be composed of a three-dimensional block.
[0067]
Further, in the embodiment of the present invention, ADRC is used. However, the present invention is not limited to this, and when DCT (Discrete Cosine Transform) is used, the DC component is optimized and GBTC (Global Block Trancation Coding). Can be used to optimize the mean and standard deviation.
[0068]
The embodiment of the present invention is a coding apparatus for a digital image signal. However, the present invention is not limited to this and can be applied to data such as a digital audio signal.
[0069]
Various modifications and application examples can be considered without departing from the gist of the present invention. Therefore, the gist of the present invention is not limited to the examples.
[0070]
【The invention's effect】
The present invention optimizes a parameter first obtained using a least square method on a coding side so as to obtain a quantized value so that a decoding error between an original signal value and a restored value becomes small. Transmit the optimized parameters and quantized values. ing. Therefore, according to the present invention, the error between the restored value and the original signal value can be reduced on the decoding side.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a first embodiment of an image coding apparatus according to the present invention.
FIG. 2 is a diagram illustrating a configuration example of a least squares method estimation unit according to the present invention.
FIG. 3 is a diagram showing a configuration example of a least squares method estimation unit according to the present invention.
FIG. 4 is a block diagram showing a second embodiment of the image coding apparatus according to the present invention.
FIG. 5 is a block diagram showing a third embodiment of the image coding apparatus according to the present invention.
FIG. 6 is a block diagram showing an example in which the image coding apparatus according to the present invention is used in a hierarchical coding apparatus.
FIG. 7 is a block diagram of a conventional ADRC.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 ... Block division part, 2 ... Maximum value detection part, 3 ... Minimum value detection part, 4, 5 ... Subtraction part, 6 ... Step width calculation part, 7 ... Quantization 8 is a least square method estimating unit, 11 is a framing unit, 12 is an error correction code adding unit, 13 is a modulating unit, 14 is a recording unit, and 15 is a recording medium. 16: Transmission unit, 17 ... Transmission path

Claims (8)

In an image encoding and transmission apparatus for encoding and transmitting the input digital image signal so as to reduce the amount of generated data of the input digital image signal,
A block dividing unit for dividing the input digital image signal into blocks composed of a plurality of pixels;
For each block, a first detection unit that detects a maximum value of the plurality of pixels in the block and a minimum value of the plurality of pixels;
A second detection unit that detects a dynamic range that is a difference between the maximum value and the minimum value;
The difference value obtained by subtracting the pixel values of the plurality of pixels from the maximum value or the difference value obtained by subtracting the minimum value from the pixel values of the plurality of pixels is quantized to generate a quantized value of the plurality of pixels. A quantization unit;
An optimization unit for optimizing at least two of the information indicating the maximum value, the minimum value, and the dynamic range generated for each block so as to minimize the square sum of the decoding errors of the quantization values;
Using the optimized maximum value, the optimized minimum value, and the information indicating the optimized dynamic range as parameters, the information is transmitted to the transmission path together with the quantized values of the plurality of pixels for each block. An image encoding transmission apparatus comprising: a transmission unit configured to perform transmission. In the optimization unit, the maximum value and the minimum value are optimized,
The image coding and transmitting apparatus according to claim 1, wherein the transmission unit transmits the optimized maximum value and the optimized minimum value as the parameter. In the optimization unit, the maximum value and the dynamic range are optimized,
The image encoding transmission apparatus according to claim 1, wherein the transmission unit transmits the optimized maximum value and the optimized dynamic range as the parameters. In the optimization unit, the minimum value and the dynamic range are optimized,
The image encoding transmission apparatus according to claim 1, wherein the transmission unit transmits the optimized minimum value and the optimized dynamic range as the parameters. In an image encoding transmission method for encoding and transmitting the input digital image signal so as to reduce the amount of generated data of the input digital image signal,
Dividing the input digital image signal into blocks of a plurality of pixels;
Detecting a maximum value of the plurality of pixels in the block and a minimum value of the plurality of pixels for each of the blocks;
Detecting a dynamic range which is a difference between the maximum value and the minimum value;
The difference value obtained by subtracting the pixel values of the plurality of pixels from the maximum value or the difference value obtained by subtracting the minimum value from the pixel values of the plurality of pixels is quantized to generate a quantized value of the plurality of pixels. Steps,
Optimizing at least two of information indicating a maximum value, a minimum value, and a dynamic range generated for each block so as to minimize a square sum of decoding errors of the quantized values;
Using the optimized maximum value, the optimized minimum value, and the information indicating the optimized dynamic range as parameters, the information is transmitted to the transmission path together with the quantized values of the plurality of pixels for each block. An image encoding transmission method characterized by comprising the steps of: In the optimizing step, the maximum value and the minimum value are optimized,
6. The image encoding transmission method according to claim 5, wherein, in the transmitting step, the optimized maximum value and the optimized minimum value are transmitted as the parameter. In the optimizing step, the maximum value and the dynamic range are optimized,
6. The image encoding transmission method according to claim 5, wherein, in the transmitting step, the optimized maximum value and the optimized dynamic range are transmitted as the parameters. In the optimizing step, the minimum value and the dynamic range are optimized,
6. The image encoding transmission method according to claim 5, wherein in the transmitting step, the optimized minimum value and the optimized dynamic range are transmitted as the parameters.

Images