JPH08274983A - Pixel interpolation method - Google Patents

Pixel interpolation method

Info

Publication number
JPH08274983A
JPH08274983A JP7298995A JP7298995A JPH08274983A JP H08274983 A JPH08274983 A JP H08274983A JP 7298995 A JP7298995 A JP 7298995A JP 7298995 A JP7298995 A JP 7298995A JP H08274983 A JPH08274983 A JP H08274983A
Authority
JP
Japan
Prior art keywords
interpolation
pixel
correlation
calculated
value
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
JP7298995A
Other languages
Japanese (ja)
Inventor
Kaoru Higuchi
馨 樋口
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sharp Corp
Original Assignee
Sharp Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sharp Corp filed Critical Sharp Corp
Priority to JP7298995A priority Critical patent/JPH08274983A/en
Publication of JPH08274983A publication Critical patent/JPH08274983A/en
Pending legal-status Critical Current

Links

Abstract

PURPOSE: To relieve visual deterioration by calculating correlation data between plural interpolation reference picture elements and applying weighting to picture element data of each interpolation reference picture element accordingly. CONSTITUTION: Interpolation reference inter-picture-element object pixels (x, y), (x+1, y), (x+1, y+1) and (x, y+1) surrounding an interpolation object pixel (x', y') are selected and let a distance from the pixel (x', y') to the pixel (x, y) on the X and Y axes be respectively α, β. Number of combinations of the interpolation reference pixels opposite to each other with the pixel (x', y') inbetween is 6. However, the combinations between the pixels (x, y) and (x+1, y), between the pixels (x+1, y) and (x+1, y+1) and the pixels (x+1,y+1) and (x, y+1) provide a distance 0 on the X and Y axes and receive a high effect of the distance, then the interpolation calculation on the distance is used and no weighting is conducted. In the other combinations, weighting is calculated depending on the correlation.

Description

【発明の詳細な説明】Detailed Description of the Invention

【0001】[0001]

【産業上の利用分野】本発明は、例えば、画像を拡大す
るときに行われる画素補間方法に関する。
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a pixel interpolation method which is performed when enlarging an image, for example.

【0002】[0002]

【従来の技術】従来から、多値画像(例えば8bit/
画素、256階調)を拡大するときに用いられる画素補
間方法として、線形補間法(Bilinear法)と選択型補間
法とがある。線形補間法は、例えば、図6に示すように、
X−Y座標系で距離1の等間隔に格子状に配列した画像
を拡大する際に行われる補間方法であって、白丸で示さ
れる補間対象画素(補間によって作成される画素)を囲ん
でその四隅に位置する黒丸で示される4個の補間参照画
素(補間対象画素を作成するために参照する画素)の濃
度差を用いて補間している。すなわち、補間対象画素の
座標を(x',y')とする一方、各補間参照画素の座標を
それぞれ(x,y),(x+1,y),(x+1,y+
1),(x,y+1)とし、さらには、各補間参照画素
の濃度値をそれぞれ、f(x,y),f(x+1,
y),f(x+1,y+1),f(x,y+1)とし、
さらには、補間対象画素(x’,y’)から補間参照画
素(x,y)までのX座標系の距離をα、Y座標系の距
離をβとした場合、補間対象画素の濃度値f(x’,
y’)は次の(1)式で算出することができる。
2. Description of the Related Art Conventionally, multi-valued images (for example, 8 bit /
There are a linear interpolation method (Bilinear method) and a selective interpolation method as pixel interpolation methods used when enlarging pixels (256 gradations). The linear interpolation method is, for example, as shown in FIG.
This is an interpolation method performed when enlarging an image arranged in a grid pattern at an equal distance of 1 in the XY coordinate system, and enclosing an interpolation target pixel (pixel created by interpolation) indicated by a white circle. Interpolation is performed using the density differences of four interpolation reference pixels (pixels to be referred to when creating the interpolation target pixel) indicated by black circles located at the four corners. That is, the coordinates of the pixel to be interpolated are (x ′, y ′), while the coordinates of each interpolation reference pixel are (x, y), (x + 1, y), (x + 1, y +).
1) and (x, y + 1), and the density values of the respective interpolation reference pixels are f (x, y) and f (x + 1, respectively).
y), f (x + 1, y + 1), f (x, y + 1),
Further, when the distance in the X coordinate system from the interpolation target pixel (x ′, y ′) to the interpolation reference pixel (x, y) is α and the distance in the Y coordinate system is β, the density value f of the interpolation target pixel is f. (X ',
y ′) can be calculated by the following equation (1).

【0003】 f(x’,y’)=(1−α)(1−β)×f(x,y)+(1−α)β ×f(x,y+1)+α(1−β)×f(x+1,y) +αβ×f(x+1,y+1)…(1) 線形補間法は、実際には、距離α、βは定数であること
が多く、そのために計算を積算のみで行うことができ、
簡単な処理により、補間を行うことができるという利点
がある。
F (x ′, y ′) = (1-α) (1-β) × f (x, y) + (1-α) β × f (x, y + 1) + α (1-β) × f (x + 1, y) + αβ × f (x + 1, y + 1) ... (1) In the linear interpolation method, in practice, the distances α and β are often constants, and therefore calculation can be performed only by integration. ,
There is an advantage that interpolation can be performed by a simple process.

【0004】一方、選択型補間法は、補間参照画素2点
間の相関を基に相関の強い方向を選択して補間を行う補
間法であって、例えば、テレビのNTSC信号のように
フィールド画像の拡大を行う場合に行われる。具体的に
は、図4に示すように、図中、水平方向に沿って配列さ
れた画素列の間隔が”2”で、各画素列の画素間隔が”
1”である画像において、画素列の間に、各画素列と等
間隔をあけて補間対象画素を設ける場合において、各補
間参照画素の座標を(x−1,y),(x,y),(x
+1,y),(x−1,y+2),(x,y+2),
(X+1,y+2)とし、さらには、各補間参照画素の
濃度値をそれぞれ、f(x−1,y),f(x,y),
f(x+1,y),f(x−1,y+2),f(x,y
+2),f(X+1,y+2)とすると、選択型補間法
は図7に示すフローチャートに基づいて補間が行われ
る。
On the other hand, the selective interpolation method is an interpolation method for performing interpolation by selecting a direction having a strong correlation based on the correlation between two points of interpolated reference pixels, and for example, a field image like an NTSC signal of a television. It is performed when expanding. Specifically, as shown in FIG. 4, in the drawing, the interval between the pixel columns arranged in the horizontal direction is “2” and the pixel interval between the pixel columns is “2”.
In the image of 1 ″, when the interpolation target pixels are provided at equal intervals between the pixel columns in the image, the coordinates of the interpolation reference pixels are (x−1, y), (x, y). , (X
+1, y), (x-1, y + 2), (x, y + 2),
(X + 1, y + 2), and the density values of the respective interpolation reference pixels are f (x-1, y), f (x, y),
f (x + 1, y), f (x-1, y + 2), f (x, y
+2), f (X + 1, y + 2), the selective interpolation method performs interpolation based on the flowchart shown in FIG.

【0005】すなわち、ステップS1で、補間対象画素
(x’,y’)を挟んで斜め方向に沿って対向する補間
対象画素(x−1,y),(x+1,y+2)間の相関
値(相関具合を示す値)をdA、補間対象画素(x’,
y’)を挟んで真上と真下に位置する補間参照画素
(x,y),(x,y+2)間の相関値をdB、同じく
斜め方向に沿って対向する補間参照画素(x+1,
y),(x−1,y+2)間の相関値をdCとし、各相
関値dA,dB,dCを次の(2),(3),(4)の
各式によって算出する。
That is, in step S1, the correlation value ((x−1, y + 2)) between the interpolation target pixels (x−1, y) and (x + 1, y + 2) that are opposed to each other in the diagonal direction with the interpolation target pixel (x ′, y ′) sandwiched therebetween. The value indicating the degree of correlation is dA, and the interpolation target pixel (x ′,
y '), the correlation value between the interpolated reference pixels (x, y), (x, y + 2) located directly above and below the interleaved interpolated reference pixel (x + 1,
The correlation value between y) and (x-1, y + 2) is defined as dC, and the correlation values dA, dB, and dC are calculated by the following equations (2), (3), and (4).

【0006】 dA=|f(x−1,y)−f(x+1,y+2)|…(2) dB=|f(x,y)−f(x,y+2)| …(3) dC=|f(x+1,y)−f(x−1,y+2)|…(4) 相関値dA,dB,dCを算出したのち、ステップS
2,S3,S4で、各相関値dA,dB,dCを比較
し、最も相関の強い(相関値の小さい)補間参照画素の
組み合わせを選択する。
DA = | f (x-1, y) -f (x + 1, y + 2) | ... (2) dB = | f (x, y) -f (x, y + 2) | ... (3) dC = | f (x + 1, y) -f (x-1, y + 2) | ... (4) After calculating the correlation values dA, dB, dC, step S
In S2, S3, and S4, the correlation values dA, dB, and dC are compared, and the combination of interpolation reference pixels having the strongest correlation (smallest correlation value) is selected.

【0007】そして、選択した補間参照画素の組み合わ
せの濃度値の平均値を算出して、その平均値を補間対象
画素(x’,y’)の補間値f(x’,y’)とする。
Then, the average value of the density values of the combination of the selected interpolation reference pixels is calculated, and the average value is set as the interpolation value f (x ', y') of the interpolation target pixel (x ', y'). .

【0008】[0008]

【発明が解決しようとする課題】しかしながら、上記従
来の補間方法には、次のような問題があった。
However, the above-mentioned conventional interpolation method has the following problems.

【0009】すなわち、線形補間法には、補間処理前の
補間参照画素を斜線または境界線(この線を挟んで濃度
値が大幅に異なる線)が横切る場合、一つないし複数の
補間参照画素が補間参照に適さない、つまり、他の補間
参照画素との間で大幅な濃度値の開きが発生することが
ある。しかしながら、従来の線形補間法では、このよう
な補間の参照に適さない画素を排除することができず、
そのために、斜線、境界線のぼやけ、ジャギー(ギザギ
ザ)といった視覚的な画像の乱れを引き起こしていた。
That is, in the linear interpolation method, when an oblique reference line or a boundary line (a line having a greatly different density value sandwiching this line) crosses the interpolation reference pixel before the interpolation processing, one or a plurality of interpolation reference pixels are provided. It may not be suitable for interpolation reference, that is, a large difference in density value may occur with other interpolation reference pixels. However, the conventional linear interpolation method cannot exclude pixels that are not suitable for the reference of such interpolation,
Therefore, it causes visual distortion such as diagonal lines, blurring of boundary lines, and jaggies.

【0010】例えば、図3に示すように、距離”1”の
等間隔にある補間参照画素において、補間参照画素
(x,y),(x+1,y),(x+1,y+1)と補
間参照画素(x,y+1)とを分ける境界線Kが存在
し、したがって、各補間参照画素の濃度値がそれぞれ、
f(x,y)=74,f(x+1,y)=96,f(x
+1,y+1)=82,f(x,y+1)=176とな
った画像において、補間参照画素(x,y)からX座標
系、Y座標系それぞれの離間距離が0.5である位置に
補間対象画素(x’,y’)がある場合、補間対象画素
(x’,y’)の補間値f(x’,y’)は上記した
(1)式によって算出すると107となる。そのため、
画素(x,y)と画素(x+1,y+1)との間の濃度
値が非常に高くなって、ジャギーが発生してしまう。
For example, as shown in FIG. 3, interpolation reference pixels (x, y), (x + 1, y), (x + 1, y + 1) and interpolation reference pixels in the interpolation reference pixels at equal intervals of "1". There is a boundary line K that separates (x, y + 1), so that the density value of each interpolation reference pixel is
f (x, y) = 74, f (x + 1, y) = 96, f (x
In the image in which +1, y + 1) = 82 and f (x, y + 1) = 176, interpolation is performed from the interpolation reference pixel (x, y) to a position where the distance between the X coordinate system and the Y coordinate system is 0.5. When the target pixel (x ', y') is present, the interpolated value f (x ', y') of the interpolation target pixel (x ', y') is 107 when calculated by the above equation (1). for that reason,
The density value between the pixel (x, y) and the pixel (x + 1, y + 1) becomes extremely high, which causes jaggies.

【0011】また、選択型補間法には、斜線が補間対象
画素の幅より狭く(細く)、かつ斜線内に位置する補間
参照画素の組み合わせの相関値がそれ以外の補間参照画
素の組み合わせの相関値より少しでも大きい(相関が弱
い)場合に、斜線が途切れてしまうといった画像の乱れ
を起こす可能性があった。
In the selective interpolation method, the diagonal line is narrower (thinner) than the width of the pixel to be interpolated, and the correlation value of the combination of interpolation reference pixels located in the diagonal line is the correlation value of the combination of other interpolation reference pixels. When the value is a little larger than the value (the correlation is weak), there is a possibility that the image is disturbed such that the diagonal line is broken.

【0012】すなわち、図5に示すように、画素列の間
隔が”2”で、各画素列の画素間隔が”1”である補間
参照画素(x−2,y)(x−1,y),(x,y),
(x+1,y),(x−2,y+2),(x−1,y+
2),(x,y+2),(X+1,y+2)(x+2,
y+2)において、画素(x−1,y),(x,y),
(x,y+2),(X+1,y+2)を通る斜線Sが存
在し、そのために各画素の濃度値がf(x−2,y)=
82、f(x−1,y)=160、f(x,y)=16
4、f(x+1,y)=56、f(x+2,y)=4
8、f(x−2,y+2)=60、f(x−1,y+
2)=64、f(x,y+2)=176、f(X+1,
y+2)=170、f(x+2,y+2)=56である
画像について考える。このような画像において、補間対
象画素(x’−1,y’),((x’,y’),(x’
+1,y’)を補間すると、次のようになる。
That is, as shown in FIG. 5, interpolation reference pixels (x-2, y) (x-1, y) in which the pixel row interval is "2" and the pixel interval between each pixel row is "1". ), (X, y),
(X + 1, y), (x-2, y + 2), (x-1, y +
2), (x, y + 2), (X + 1, y + 2) (x + 2,
y + 2), pixels (x-1, y), (x, y),
There is a diagonal line S passing through (x, y + 2) and (X + 1, y + 2), and therefore the density value of each pixel is f (x-2, y) =
82, f (x-1, y) = 160, f (x, y) = 16
4, f (x + 1, y) = 56, f (x + 2, y) = 4
8, f (x-2, y + 2) = 60, f (x-1, y +)
2) = 64, f (x, y + 2) = 176, f (X + 1,
Consider an image where y + 2) = 170 and f (x + 2, y + 2) = 56. In such an image, interpolation target pixels (x'-1, y '), ((x', y '), (x'
Interpolating + 1, y ') gives the following:

【0013】すなわち、補間対象画素(x’−1,
y’)では、補間対象画素(x−2,y),(x,y+
2)の組み合わせの相関値が94(=|82−176
|)と最も小さく(相関が強く)なる。そのため、補間
値f(x’−1,y’)=(82+176)/2=12
9となる。
That is, the pixel to be interpolated (x'-1,
y ′), interpolation target pixels (x−2, y), (x, y +)
The correlation value of the combination of 2) is 94 (= | 82-176
|) And the smallest (strong correlation). Therefore, the interpolation value f (x'-1, y ') = (82 + 176) / 2 = 12
It becomes 9.

【0014】補間対象画素(x’,y’)では、補間対
象画素(x+1,y),(x−1,y+2)の組み合わ
せの相関値が8(=|56−64|)と最も小さく(相
関が強く)なる。そのため、補間値f(x’,y’)=
(56+64)/2=60となる。
In the interpolation target pixel (x ', y'), the correlation value of the combination of the interpolation target pixels (x + 1, y), (x-1, y + 2) is 8 (= | 56-64 |), which is the smallest ( Strong correlation). Therefore, the interpolation value f (x ', y') =
(56 + 64) / 2 = 60.

【0015】補間対象画素(x’+1,y’)では、補
間対象画素(x,y),(x+2,y+2)の組み合わ
せの相関値が108(=|56−164|)と最も小さ
く(相関が強く)なる。そのため、補間値f(x’+
1,y’)=(56+164)/2=110となる。
In the pixel to be interpolated (x '+ 1, y'), the correlation value of the combination of the pixels to be interpolated (x, y) and (x + 2, y + 2) is 108 (= | 56-164 |), which is the smallest (correlation). Becomes stronger). Therefore, the interpolation value f (x '+
1, y ′) = (56 + 164) / 2 = 110.

【0016】したがって、各補間対象画素(x’−1,
y’),(x’,y’),(x’+1,y’)の補間値
の中で、補間値f(x’,y’)だけが60と非常に小
さな値になり、そのために斜線が途切れてしまうといっ
た画像の乱れを生じてしまうことがあった。
Therefore, each pixel to be interpolated (x'-1,
Among the interpolated values of y '), (x', y '), (x' + 1, y '), only the interpolated value f (x', y ') becomes 60, which is a very small value. There was a case where the image was disturbed such that the diagonal lines were interrupted.

【0017】さらには、選択型補間法には、補間計算に
比較演算を多用するために処理が複雑化して、演算処理
に時間がかかるうえ、演算に多くの記憶容量が必要とな
るという問題もあった。そのうえ、整数倍の拡大以外へ
の対応が困難であるという問題もあった。
Further, the selection type interpolation method has a problem that the processing is complicated because a lot of comparison calculations are used for the interpolation calculation, the calculation process takes a long time, and a large storage capacity is required for the calculation. there were. In addition, there is a problem that it is difficult to deal with anything other than integer multiple expansion.

【0018】したがって、本発明においては、画像の乱
れを生じさせることなく、かつ簡単な回路で演算するこ
とのできる補間方法を提供することを目的としている。
Therefore, it is an object of the present invention to provide an interpolation method which can be operated by a simple circuit without causing image distortion.

【0019】[0019]

【課題を解決するための手段】このような目的を達成す
るために、本発明においては、補間対象画素を囲んで位
置する複数の補間参照画素間の画素データ相関具合を示
す相関データを算出し、算出した相関データに対応して
各補間参照画素の画素データに重み付けを行うことに特
徴を有している。
In order to achieve such an object, in the present invention, correlation data indicating the degree of pixel data correlation between a plurality of interpolation reference pixels located surrounding an interpolation target pixel is calculated. The feature is that the pixel data of each interpolation reference pixel is weighted corresponding to the calculated correlation data.

【0020】[0020]

【作用】上記構成によれば、補間参照画素の画素データ
に相関具合に応じた重み付けを行うので、極端に画素デ
ータの値の異なる補間参照画素の影響を排除することが
できるようになる。
According to the above structure, since the pixel data of the interpolation reference pixel is weighted according to the degree of correlation, it is possible to eliminate the influence of the interpolation reference pixel having extremely different pixel data values.

【0021】[0021]

【実施例】以下、本発明の実施例を図面を参照して詳細
に説明する。
Embodiments of the present invention will now be described in detail with reference to the drawings.

【0022】第1実施例 この実施例では、図1に示すように、X−Y座標系で距
離1の等間隔に格子状に配列した画像を拡大する際に行
われる補間方法において、本発明を実施した例を示して
いる。
First Embodiment In this embodiment, as shown in FIG. 1, the present invention is applied to an interpolation method performed when enlarging an image arranged in a grid pattern at an equal distance of 1 in an XY coordinate system. The example which carried out is shown.

【0023】補間対象画素(x’,y’)を囲んでその
四隅に位置する4個の補間参照画素間対象画素(x,
y),(x+1,y),(x+1,y+1),(x,y
+1)とし、さらには、補間対象画素や各補間参照画素
の濃度値をそれぞれ、f(x’,y’),f(x,y)
f(x+1,y),f(x+1,y+1),f(x,y
+1)とし、さらには、補間対象画素(x’,y’)か
ら補間参照画素(x,y)までのX座標系の距離をα、
Y座標系の距離をβとする。
Interpolation reference pixel target pixel (x, x) located at four corners of the interpolation target pixel (x ', y')
y), (x + 1, y), (x + 1, y + 1), (x, y
+1), and the density values of the interpolation target pixel and each interpolation reference pixel are f (x ′, y ′) and f (x, y), respectively.
f (x + 1, y), f (x + 1, y + 1), f (x, y
+1), and further, the distance in the X coordinate system from the interpolation target pixel (x ′, y ′) to the interpolation reference pixel (x, y) is α,
Let β be the distance in the Y coordinate system.

【0024】補間対象画素(x’,y’)を挟んで対向
する補間参照画素の組み合わせは、(x,y)と(x+
1,y)、(x,y)と(x+1,y+1)、(x,
y)と(x,y+1)、(x+1,y)と(x+1,y
+1)、(x+1,y)と(x,y+1)、および(x
+1,y+1)と(x,y+1)の6つの組み合わせに
なる。しかしながら、(x,y)と(x+1,y)、お
よび(x+1,y+1)と(x,y+1)の各組み合わ
せでは、Y方向の距離が”0”になる。一方、(x,
y)と(x,y+1)および(x+1,y)と(x+
1,y+1)の各組み合わせではX座標系の距離が”
0”になる。そのため、これらの補間参照画素の組み合
わせでは、距離による影響が強く、距離を基にした補間
演算を用い、本発明のように相関具合に基づいた重み付
け付加した補間計算を行わない。したがって、補間対象
画素(x’,y’)を挟んで対向する補間参照画素の組
み合わせ(x,y)と(x+1,y+1)、および(x
+1,y)と(x,y+1)の組み合わせだけにおいて
相関具合に基づいた重み付け計算を行う。
The combination of interpolation reference pixels facing each other with the pixel to be interpolated (x ', y') interposed therebetween is (x, y) and (x +).
1, y), (x, y) and (x + 1, y + 1), (x,
y) and (x, y + 1), (x + 1, y) and (x + 1, y
+1), (x + 1, y) and (x, y + 1), and (x
There are six combinations of +1, y + 1) and (x, y + 1). However, in each combination of (x, y) and (x + 1, y) and (x + 1, y + 1) and (x, y + 1), the distance in the Y direction is “0”. On the other hand, (x,
y) and (x, y + 1) and (x + 1, y) and (x +
1, y + 1), the distance in the X coordinate system is "
Therefore, in the combination of these interpolation reference pixels, the influence of the distance is strong, and the interpolation calculation based on the distance is used, and the interpolation calculation with weighting addition based on the degree of correlation is not performed unlike the present invention. Therefore, the combinations (x, y) and (x + 1, y + 1) of the interpolation reference pixels that face each other across the interpolation target pixel (x ′, y ′), and (x
Only in the combination of (+1, y) and (x, y + 1), weighting calculation based on the degree of correlation is performed.

【0025】以下、その計算手順を説明する。The calculation procedure will be described below.

【0026】補間参照画素(x,y)と(x+1,y+
1)との間の相関値をdsとし、補間参照画素(x+
1,y)と(x,y+1)との間の相関値をdtとした
場合、各相関値ds,dtを次の(5),(6)式に基
づいて算出する。
Interpolation reference pixels (x, y) and (x + 1, y +)
The correlation value with 1) is ds, and the interpolation reference pixel (x +
When the correlation value between (1, y) and (x, y + 1) is dt, the correlation values ds and dt are calculated based on the following equations (5) and (6).

【0027】 ds=|f(x,y)−f(x+1,y+1)|…(5) dt=|f(x+1,y)−f(x,y+1)|…(6) そして、算出した相関値ds,dtに基づいて補間参照
画素(x,y),(x+1,y+1)間の相関具合を示
す相関比率psを、次の(7)式に基づいて、また、補
間参照画素(x+1,y),(x,y+1)間の相関具
合を示す相関比率ptを次の(8)式に基づいてそれぞ
れ算出する。
Ds = | f (x, y) -f (x + 1, y + 1) | ... (5) dt = | f (x + 1, y) -f (x, y + 1) | ... (6) Then, the calculated correlation The correlation ratio ps indicating the correlation between the interpolated reference pixels (x, y) and (x + 1, y + 1) based on the values ds and dt is calculated based on the following equation (7) and the interpolated reference pixel (x + 1, The correlation ratio pt indicating the degree of correlation between y) and (x, y + 1) is calculated based on the following equation (8).

【0028】ps=dt/(ds+dt)…(7) pt=ds/(ds+dt)…(8) ただしds=dt=0のときは、ps=pt=0.5 さらに、算出した相関比率ps,ptに基づいて、補間
対象画素(x’,y’)の補間値f(x’,y’)を次
の(9)式に基づいて算出する。
Ps = dt / (ds + dt) (7) pt = ds / (ds + dt) (8) However, when ds = dt = 0, ps = pt = 0.5 Furthermore, the calculated correlation ratio ps, Based on pt, the interpolation value f (x ', y') of the interpolation target pixel (x ', y') is calculated based on the following equation (9).

【0029】 f(x’,y’)=2×[{(1−α)(1−β)×f(x,y) +αβ×f(x+1,y+1)}×ps +{(1−α)β×f(x,y+1)+α(1−β) ×f(x+1,y)}×pt]…(9) なお、相関値ds,dtや相関比率ps,ptは、上記
した(5),(6),(7),(8)式に限らず、次に
示す(10),(11),(12),(13)式によっ
て算出してもよく、要は、補間参照画素間の相関具合を
導き出せる値を算出できる式であれば、どのような式で
あってもよい。
F (x ′, y ′) = 2 × [{(1-α) (1-β) × f (x, y) + αβ × f (x + 1, y + 1)} × ps + {(1-α ) Β × f (x, y + 1) + α (1-β) × f (x + 1, y)} × pt] (9) Note that the correlation values ds, dt and the correlation ratios ps, pt are as described in (5) above. , (6), (7), and (8) equations, but may be calculated by the following equations (10), (11), (12), and (13). Any formula may be used as long as it can calculate a value that can derive the degree of correlation of

【0030】 ds={f(x,y)−f(x+1,y+1)}2 …(10) dt={f(x+1,y)−f(x,y+1)}2 …(11) ps=dt+u/(ds+dt+2u) …(12) pt=ds+u/(ds+dt+2u) u:任意の値 …(13) なお、相関比率ps,ptを(12),(13)式のよ
うにして算出すると、その相関比率は(7),(8)式
で算出した相関比率より若干変化するが、ds=dt=
0の場合に特別な処理を施す必要がなくなる。
Ds = {f (x, y) -f (x + 1, y + 1)} 2 (10) dt = {f (x + 1, y) -f (x, y + 1)} 2 (11) ps = dt + u / (Ds + dt + 2u) (12) pt = ds + u / (ds + dt + 2u) u: arbitrary value (13) When the correlation ratios ps and pt are calculated as in the equations (12) and (13), the correlation ratios are obtained. Changes slightly from the correlation ratio calculated by the equations (7) and (8), but ds = dt =
When it is 0, it is not necessary to perform special processing.

【0031】次に、具体的な計算例を図2に基づいて説
明する。図2はX−Y座標系で距離”1”で等間隔な格
子状に配列してなる画像を各座標それぞれ2倍に拡大す
る場合の補間方法である。
Next, a specific calculation example will be described with reference to FIG. FIG. 2 shows an interpolation method in the case where an image formed by arranging in a grid pattern at equal intervals at a distance of "1" in the XY coordinate system is magnified twice for each coordinate.

【0032】補間参照画素(x,y),(x+1,y),
(x+1,y+1),(x,y+1)の各濃度値はそれ
ぞれf(x,y)=58、f(x+1,y)=121、
f(x+1,y+1)=104、f(x,y+1)=1
00である。したがって、相関比率は上記(7),
(8)式により、ps=0.386、pt=0.632
となる。
Interpolation reference pixels (x, y), (x + 1, y),
The density values of (x + 1, y + 1) and (x, y + 1) are f (x, y) = 58 and f (x + 1, y) = 121, respectively.
f (x + 1, y + 1) = 104, f (x, y + 1) = 1
00. Therefore, the correlation ratio is (7),
From equation (8), ps = 0.386, pt = 0.632
Becomes

【0033】算出した相関比率ps,ptと上記した補
間参照画素の濃度値を(9)式に代入することで補間値
は次のように算出される。
The interpolation values are calculated as follows by substituting the calculated correlation ratios ps, pt and the density values of the above-mentioned interpolated reference pixels into the equation (9).

【0034】 f(x’−0.5,y’)=2×[{(1−0)(1−0.5)×58 +0・0.5×104}×0.368+ {(1−0)0.5×100+0・(1−0.5 ) ×121}×0.632]≒84 f(x’,y’−0.5)=2×[{(1−0.5)(1−0)×58 +0・0.5×104}×0.368+ {(1−0.5)0×100+0.5(1−0) ×121}×0.632]≒97 f(x’,y’)=2×[{(1−0.5)(1−0.5)×58 +0.5・0.5×104}×0.386+ {(1−0.5)0.5×100+0.5(1−0.5) ×121}×0.632]≒100 なお、この例では、補間対象画素はこの他にも、
(x’,y’+0.5)と(x’+0.5,y’)とが
あるが、補間対象画素(x’,y’+0.5)は、補間
参照画素(x,y+1),(x+1,y+1),(x,
y+2),(x+1,y+2)によって、また、補間対
象画素(x’+0.5,y’)は補間参照画素(x+
1,y+1),(x+1,y),(x+2,y),(x
+2,y+1)によって補間計算されるため、その計算
は省略する。しかしながら、もちろん、これら補間対象
画素(x’,y’+0.5),(x’+0.5,y’)
の補間値を上述と同様の計算により求めてもよい。
F (x′-0.5, y ′) = 2 × [{(1-0) (1-0.5) × 58 + 0 · 0.5 × 104} × 0.368 + {(1- 0) 0.5 × 100 + 0 · (1-0.5) × 121} × 0.632] ≈84 f (x ′, y′−0.5) = 2 × [{(1-0.5) ( 1-0) × 58 + 0 · 0.5 × 104} × 0.368 + {(1-0.5) 0 × 100 + 0.5 (1-0) × 121} × 0.632] ≈97f (x ' , Y ′) = 2 × [{(1-0.5) (1-0.5) × 58 + 0.5 · 0.5 × 104} × 0.386 + {(1-0.5) 0.5 × 100 + 0.5 (1-0.5) × 121} × 0.632] ≈100 In this example, the interpolation target pixel is
Although there are (x ', y' + 0.5) and (x '+ 0.5, y'), the interpolation target pixel (x ', y' + 0.5) is the interpolation reference pixel (x, y + 1), (X + 1, y + 1), (x,
y + 2), (x + 1, y + 2), and the interpolation target pixel (x ′ + 0.5, y ′) is the interpolation reference pixel (x +
1, y + 1), (x + 1, y), (x + 2, y), (x
Since the interpolation calculation is performed by +2, y + 1), the calculation is omitted. However, of course, these interpolation target pixels (x ′, y ′ + 0.5), (x ′ + 0.5, y ′)
The interpolation value of may be obtained by the same calculation as described above.

【0035】また、補間対象画素(x’,y’+0.
5)は、補間参照画素(x,y),(x+1,y),
(x,y+1),(x+1,y+1),(x,y+
2),(x+1,y+2)という6点に跨がって配置さ
れている。この補間対象画素(x’,y’+0.5)で
は、6点ある補間参照画素を用いて上述の補間計算を行
って補間値を求めてもよいし、6点ある補間対象画素を
4点づつ、二つに分ける、すなわち、(x,y),(x
+1,y),(x,y+1),(x+1,y+1)のグ
ループと、(x,y+1),(x+1,y+1),
(x,y+2),(x+1,y+2)のグループとに分
け、それぞれの補間参照画素グループ毎に上述した補間
計算を行い、その平均値を補間値としてもよい。なお、
同様の位置にある補間対象画素はこの他、(x’−0.
5,y’),(x’,y’−0.5),(x’+0.
5,y’)などがある。
Further, the interpolation target pixel (x ', y' + 0.
5) is the interpolation reference pixel (x, y), (x + 1, y),
(X, y + 1), (x + 1, y + 1), (x, y +
2) and (x + 1, y + 2) are arranged across 6 points. In this interpolation target pixel (x ′, y ′ + 0.5), the interpolation value may be obtained by performing the above-described interpolation calculation using the interpolation reference pixel having 6 points, or the interpolation target pixel having 6 points may be 4 points. Each of them is divided into two, that is, (x, y), (x
+1, y), (x, y + 1), (x + 1, y + 1) groups and (x, y + 1), (x + 1, y + 1),
It is also possible to divide into groups (x, y + 2) and (x + 1, y + 2), perform the above-described interpolation calculation for each interpolation reference pixel group, and use the average value as the interpolation value. In addition,
The pixels to be interpolated at the same position are (x'-0.
5, y '), (x', y'-0.5), (x '+ 0.
5, y ').

【0036】さらには、補間対象画素(x’+0.5,
y’)や(x’,y’−0.5)のように、X軸ないし
Y軸に沿って隣接する補間参照画素に挟まれた位置にあ
る補間対象画素は、補間参照画素の相関が非常に強いも
のとみなして、上記した(9)式から、相関比率を省い
た計算式によって補間値を計算してもよい。
Further, the pixel to be interpolated (x '+ 0.5,
y ') or (x', y'-0.5), the interpolation target pixel located between the adjacent interpolation reference pixels along the X axis or the Y axis has a correlation of the interpolation reference pixel. The interpolation value may be calculated from the above formula (9) by omitting the correlation ratio, assuming that it is very strong.

【0037】次に上記した補間算出方法を、図3に示し
た境界線Kが存在する画像において、実施した際の計算
を説明する。補間参照画素(x,y),(x+1,
y),(x,y+1),(x+1,y+1)の各濃度値
は、f(x,y)=74,f(x+1,y)=96,f
(x,y+1)=176,f(x+1,y+1)=82
である。したがって、相関値は(5),(6)式によ
り、ds=8(=|74−82|)、dt=80(|9
6−176|)となる。そして、これら相関値を
(7),(8)に代入することにより、相関比率は、p
s=0.91(=80/88)、pt=0.09(=8
/88)となる。さらに、相関比率ps,ptを(9)
に代入することで補間対象画素(x’,y’)の補間値
f(x’,y’)は算出される。すなわち、 f(x’,y’)=2×[{(1−0.5)(1−0.5)×74 +0.5・0.5×82}×0.91 +{(1−0.5)0.5×176+0.5 ・(1−0.5)×96}×0.09] ≒83 このように、補間対象画素(x’,y’)の補間値f
(x’,y’)が同一領域内の補間参照画素(x,
y),(x+1,y),(x,y+1)の濃度値から掛
け離れた値を取ることがなくなり、したがって、境界線
K付近でジャギーが発生することがなくなった。
Next, the calculation when the above-described interpolation calculation method is performed on the image in which the boundary line K shown in FIG. 3 exists will be described. Interpolation reference pixels (x, y), (x + 1,
y), (x, y + 1), and (x + 1, y + 1) density values are f (x, y) = 74, f (x + 1, y) = 96, f
(X, y + 1) = 176, f (x + 1, y + 1) = 82
Is. Therefore, the correlation value is calculated from the equations (5) and (6) by ds = 8 (= | 74−82 |) and dt = 80 (| 9
6-176 |). Then, by substituting these correlation values into (7) and (8), the correlation ratio becomes p
s = 0.91 (= 80/88), pt = 0.09 (= 8)
/ 88). Furthermore, the correlation ratios ps and pt are set to (9)
The interpolation value f (x ′, y ′) of the interpolation target pixel (x ′, y ′) is calculated by substituting into That is, f (x ′, y ′) = 2 × [{(1-0.5) (1-0.5) × 74 + 0.5 · 0.5 × 82} × 0.91 + {(1- 0.5) 0.5 × 176 + 0.5 · (1−0.5) × 96} × 0.09] ≈83 Thus, the interpolation value f of the interpolation target pixel (x ′, y ′)
Interpolation reference pixel (x, y ') in the same region (x,
y), (x + 1, y), and (x, y + 1) do not take values that are far from the density values, so that jaggies do not occur near the boundary line K.

【0038】第2実施例 この実施例では、補間参照画素の濃度値平均値を算出す
るとともに、算出した濃度値平均値と各補間参照画素の
濃度値との間の差から相関値を算出している点に特徴が
ある。その算出方法を図1を基にして説明する。
Second Embodiment In this embodiment, an average density value of interpolation reference pixels is calculated, and a correlation value is calculated from the difference between the calculated average density value and the density value of each interpolation reference pixel. There is a feature in that. The calculation method will be described with reference to FIG.

【0039】まず、補間参照画素(x,y)(x+1,
y)(x,y+1)(x+1,y+1)の濃度値の平均
値avを次の(14)式によって算出する。
First, the interpolation reference pixel (x, y) (x + 1,
y) The average value av of the density values of (x, y + 1) (x + 1, y + 1) is calculated by the following equation (14).

【0040】 av={f(x,y)+f(x+1,y)+f(x,y+1) +f(x+1,y+1)}/補間参照画素数(この場合は4)…(14) そして、次に示す(15),(16),(17)(1
8)式に示すように、算出された平均値avと各濃度値
との差の絶対値によって相関値d1,d2,d3,d4
を算出する。
Av = {f (x, y) + f (x + 1, y) + f (x, y + 1) + f (x + 1, y + 1)} / interpolation reference pixel number (4 in this case) (14) Then, Show (15), (16), (17) (1
As shown in the equation 8), the correlation values d1, d2, d3, d4 are calculated according to the absolute value of the difference between the calculated average value av and each density value.
To calculate.

【0041】 d1=|av−f(x,y)| …(15) d2=|av−f(x+1,y)| …(16) d3=|av−f(x+1,y+1)| …(17) d4=|av−f(x,y+1)| …(18) さらに、算出した相関値d1〜d4を基にして、次の
(19),(20)(21)(22)式により相関比率
p1,p2,p3,p4を算出する。
D1 = | av-f (x, y) | (15) d2 = | av-f (x + 1, y) | (16) d3 = | av-f (x + 1, y + 1) | (17) ) D4 = | av-f (x, y + 1) | (18) Further, based on the calculated correlation values d1 to d4, the correlation ratio is calculated by the following equations (19), (20), (21) and (22). Calculate p1, p2, p3 and p4.

【0042】 p1=(d2+d3+d4)/(d1+d2+d3+d4)/3…(19) p2=(d1+d3+d4)/(d1+d2+d3+d4)/3…(20) p3=(d1+d2+d4)/(d1+d2+d3+d4)/3…(21) p4=(d1+d2+d3)/(d1+d2+d3+d4)/3…(22) :ただし、d1=d2=d3=d4=0のときは、p1
=p2=p3=p4=1/4 そして、算出した相関比率p1,p2,p3,p4を基
にして次の(23)式により補間対象画素(x’,
y’)の補間値f(x’,y’)を算出する。
P1 = (d2 + d3 + d4) / (d1 + d2 + d3 + d4) / 3 ... (19) p2 = (d1 + d3 + d4) / (d1 + d2 + d3 + d4) / 3 ... (20) p3 = (d1 + d2 + d4) / (d1 + d2 + d3 + d3 + d3 + d3 + d4 + d3 + d3 + d4 + d3 + d4 + 3 (D1 + d2 + d3) / (d1 + d2 + d3 + d4) / 3 (22): However, when d1 = d2 = d3 = d4 = 0, p1
= P2 = p3 = p4 = 1/4 Then, based on the calculated correlation ratios p1, p2, p3 and p4, the interpolation target pixel (x ′,
The interpolation value f (x ', y') of y ') is calculated.

【0043】 f(x’,y’)=4×{(1−α)(1−β)×f(x,y)×p1 +αβ×f(x+1,y+1)}×p3 +(1−α)β×f(x,y+1)×p4+α(1−β) ×f(x+1,y)×p2} …(23) 次に上記した補間算出方法を、図3に示した境界線Kが
存在する画像において、実施した際の計算を説明する。
補間参照画素(x,y),(x+1,y),(x,y+
1),(x+1,y+1)の各濃度値は、f(x,y)
=74,f(x+1,y)=96,f(x,y+1)=
176,f(x+1,y+1)=82である。したがっ
て、各補間参照画素の濃度値平均値avは(14)式に
よって求められ、その値はav=107である。そし
て、求めた平均値avを(15)〜(18)式に代入す
ることにより相関値d1〜d4は、d1=33,d2=
11,d3=25,d4=69となる。
F (x ′, y ′) = 4 × {(1-α) (1-β) × f (x, y) × p1 + αβ × f (x + 1, y + 1)} × p3 + (1-α ) Β × f (x, y + 1) × p4 + α (1-β) × f (x + 1, y) × p2} (23) Next, the boundary line K shown in FIG. In the image, the calculation when performed will be described.
Interpolation reference pixels (x, y), (x + 1, y), (x, y +)
1), each density value of (x + 1, y + 1) is f (x, y)
= 74, f (x + 1, y) = 96, f (x, y + 1) =
176, f (x + 1, y + 1) = 82. Therefore, the density value average value av of each interpolation reference pixel is obtained by the equation (14), and the value is av = 107. Then, by substituting the obtained average value av into the equations (15) to (18), the correlation values d1 to d4 are d1 = 33, d2 =
11, d3 = 25, d4 = 69.

【0044】そして、これら相関値を(19)〜(2
2)式に代入することにより、相関比率は、p1=(1
1+25+69)/(33+11+69+25)/3≒
0.25、p2=(33+25+69)/(33+11
+69+25)/3≒0.3、p3=(33+11+6
9)/(33++11+69+25)/3≒0.27、
p4=(33+11+25)/(33+11+69+2
5)/3≒0.17、となる。
Then, these correlation values are calculated from (19) to (2
By substituting into the equation (2), the correlation ratio is p1 = (1
1 + 25 + 69) / (33 + 11 + 69 + 25) / 3≈
0.25, p2 = (33 + 25 + 69) / (33 + 11)
+ 69 + 25) /3≈0.3, p3 = (33 + 11 + 6)
9) / (33 ++ 11 + 69 + 25) /3≈0.27,
p4 = (33 + 11 + 25) / (33 + 11 + 69 + 2
5) /3≈0.17.

【0045】さらに、求めた相関比率p1〜p4を(2
3)式に代入することで補間対象画素(x’,y’)の
補間値f(x’,y’)は求まる。すなわち、 f(x’,y’)=4×{(1−0.5)(1−0.5)×74×0.25 +0.5・0.5×82×0.27 +(1−0.5)0.5×176×0.17 +0.5(1−0.5)×96×0.3} ≒99 このように、補間対象画素(x’,y’)の補間値f
(x’,y’)が同一領域内の補間参照画素(x,
y),(x+1,y),(x,y+1)の濃度値から掛
け離れた値を取ることがなくなり、したがって、境界線
K付近でジャギーが発生することがない。
Furthermore, the calculated correlation ratios p1 to p4 are (2
By substituting into the equation 3), the interpolation value f (x ', y') of the interpolation target pixel (x ', y') can be obtained. That is, f (x ', y') = 4 * {(1-0.5) (1-0.5) * 74 * 0.25 + 0.5 * 0.5 * 82 * 0.27 + (1 −0.5) 0.5 × 176 × 0.17 +0.5 (1-0.5) × 96 × 0.3} ≈99 Thus, the interpolation value of the interpolation target pixel (x ′, y ′) f
Interpolation reference pixel (x, y ') in the same region (x,
y), (x + 1, y), and (x, y + 1) do not take values that are far from the density values, so that jaggies do not occur near the boundary line K.

【0046】第3実施例 次に、テレビのNTSC信号のようなフィールド画像の
拡大を行う場合の補間値算出例を説明する。すなわち、
図4に示すように、水平方向に沿って配列された画素列
の間隔が”2”で、各画素列の画素間隔が”1”である
画像において、画素列の間に、各画素列と等間隔をあけ
て補間対象画素を設ける場合において、各補間参照画素
の座標を(x−1,y),(x,y),(x+1,
y),(x−1,y+2),(x,y+2),(X+
1,y+2)とし、さらには、各補間参照画素の濃度値
をそれぞれ、f(x−1,y),f(x,y),f(x
+1,y),f(x−1,y+2),f(x,y+
2),f(X+1,y+2)とする。
Third Embodiment Next, an example of calculation of an interpolated value when a field image such as a television NTSC signal is enlarged will be described. That is,
As shown in FIG. 4, in an image in which the pixel rows arranged in the horizontal direction have an interval of “2” and the pixel interval of each pixel row is “1”, each pixel row is When the interpolation target pixels are provided at equal intervals, the coordinates of each interpolation reference pixel are (x-1, y), (x, y), (x + 1,
y), (x-1, y + 2), (x, y + 2), (X +
1, y + 2), and further, the density value of each interpolation reference pixel is f (x-1, y), f (x, y), f (x
+1, y), f (x-1, y + 2), f (x, y +)
2) and f (X + 1, y + 2).

【0047】そして、補間参照画素(x−1,y)と
(x+1,y+2)との組み合わせの相関値dα、保管
参照画素(x,y)と(x,y+2)との組み合わせの
相関値dβ、保管参照画素(x+1,y)と(x−1,
y+2)との組み合わせの相関値dγを次の(24),
(25),(26)式によって算出する。
Then, the correlation value dα of the combination of the interpolation reference pixels (x-1, y) and (x + 1, y + 2) and the correlation value dβ of the combination of the stored reference pixels (x, y) and (x, y + 2). , Stored reference pixels (x + 1, y) and (x-1,
The correlation value dγ in combination with y + 2) is given by the following (24),
It is calculated by the equations (25) and (26).

【0048】 dα=|f(x−1,y)−(x+1,y+2)| …(24) dβ=|f(x,y)−f(x,y+2)| …(25) dγ=|f(x+1,y)−(x−1,y+2)| …(26) 算出した相関値dα、dβ,dγを基にして次の(2
7),(28)(29)式によって相関比率pα、p
β、pγを算出する。
Dα = | f (x-1, y)-(x + 1, y + 2) | (24) dβ = | f (x, y) -f (x, y + 2) | (25) dγ = | f (X + 1, y)-(x-1, y + 2) | (26) Based on the calculated correlation values dα, dβ, dγ, the following (2
7), (28) and (29), the correlation ratios pα, p
Calculate β and pγ.

【0049】 pα=(dβ+dγ)/(dα+dβ+dγ)/2 …(27) pβ=(dα+dγ)/(dα+dβ+dγ)/2 …(28) pγ=(dα+dβ)/(dα+dβ+dγ)/2 …(29) そして、算出した相関比率pα、pβ、pγを基にして
次の(30)式によって補間対象画素(x’,y’)の
補間値f(x’,y’)を算出する。
Pα = (dβ + dγ) / (dα + dβ + dγ) / 2 (27) pβ = (dα + dγ) / (dα + dβ + dγ) / 2 (28) pγ = (dα + dβ) / (dα + dβ + dγ) / 2 (29) Then, Based on the calculated correlation ratios pα, pβ, pγ, the interpolation value f (x ′, y ′) of the interpolation target pixel (x ′, y ′) is calculated by the following equation (30).

【0050】 f(x’,y’)=1/2×[{f(x−1,y)+f(x+1,y+2)} ×pα+{f(x,y)+f(x,y+2)}×pβ +{f(x+1,y)+f(x−1,y+2)} ×pγ]…(30) なお、この例では、水平方向に沿って配列された画素列
の間隔が”2”で、各画素列の画素間隔が”1”である
画像の画素補間であり、かつ、補間参照画素を6点とし
ている。そのため、各補間参照画素の間で、補間対象画
素との間の距離にそれほど違いがないものとなってお
り、各補間参照画素と補間対象画素との間で離間距離の
影響にほとんどばらつきがなくなっている。したがっ
て、(30)式では、距離の影響に応じた係数を省略し
ている。
F (x ′, y ′) = ½ × [{f (x−1, y) + f (x + 1, y + 2)} × pα + {f (x, y) + f (x, y + 2)} × pβ + {f (x + 1, y) + f (x-1, y + 2)} × pγ] (30) In this example, the pixel rows arranged in the horizontal direction have an interval of “2”, This is pixel interpolation of an image in which the pixel interval of the pixel row is "1", and the interpolation reference pixel is set to 6 points. Therefore, the distances between the interpolation reference pixels and the interpolation target pixels are not so different, and there is almost no variation in the influence of the separation distance between the interpolation reference pixels and the interpolation target pixels. ing. Therefore, in the equation (30), the coefficient according to the influence of the distance is omitted.

【0051】しかしながら、(30)式において、上述
した(9),(23)式のように、距離の影響に応じた
係数を付加してもよい。特に、補間参照画素が10点以
上になる場合では、補間参照画素と補間対象画素との離
間距離の影響が、各補間参照画素でばらついてくるの
で、補間値の計算式に距離の影響に応じた係数を付加す
る方が好ましい。
However, in the equation (30), a coefficient depending on the influence of the distance may be added as in the above equations (9) and (23). In particular, when the number of interpolation reference pixels is 10 or more, the influence of the distance between the interpolation reference pixel and the interpolation target pixel varies among the interpolation reference pixels. It is preferable to add a coefficient.

【0052】次に上記した補間算出方法を、図5に示し
た斜線Sが存在する画像において、実施した際の計算を
説明する。
Next, a description will be given of the calculation when the above-described interpolation calculation method is performed on the image having the slanted line S shown in FIG.

【0053】図5の画像は、従来例で説明したごとく、
画素列の間隔が”2”で、各画素列の画素間隔が”1”
である画像において、補間参照画素を(x−2,y)
(x−1,y),(x,y),(x+1,y),(x−
2,y+2),(x−1,y+2),(x,y+2),
および(X+1,y+2)(x+2,y+2)の6点と
している。そして、画素(x−1,y),(x,y),
(x,y+2),(X+1,y+2)を通る斜線Sが存
在し、そのために各補間参照画素の濃度値がf(x−
2,y)=82、f(x−1,y)=160、f(x,
y)=164、f(x+1,y)=56、f(x+2,
y)=48、f(x−2,y+2)=60、f(x−
1,y+2)=64、f(x,y+2)=176、f
(X+1,y+2)=170、f(x+2,y+2)=
56となっている。
The image shown in FIG. 5 is, as described in the conventional example,
Pixel row spacing is "2", and pixel row spacing is "1"
In the image, the interpolation reference pixel is (x-2, y)
(X-1, y), (x, y), (x + 1, y), (x-
2, y + 2), (x-1, y + 2), (x, y + 2),
And (X + 1, y + 2) (x + 2, y + 2). Then, the pixels (x-1, y), (x, y),
There is a diagonal line S passing through (x, y + 2) and (X + 1, y + 2), and therefore the density value of each interpolation reference pixel is f (x-
2, y) = 82, f (x-1, y) = 160, f (x,
y) = 164, f (x + 1, y) = 56, f (x + 2,2
y) = 48, f (x−2, y + 2) = 60, f (x−)
1, y + 2) = 64, f (x, y + 2) = 176, f
(X + 1, y + 2) = 170, f (x + 2, y + 2) =
It is 56.

【0054】このような画像において、補間対象画素
(x’−1,y’),(x’,y’),(x’+1,
y’)を補間すると、次のようになる。
In such an image, pixels to be interpolated (x'-1, y '), (x', y '), (x' + 1,
Interpolating y ′) gives:

【0055】まず、補間対象画素(x’−1,y’)の
補間値を算出する。
First, the interpolation value of the interpolation target pixel (x'-1, y ') is calculated.

【0056】(24),(25),(26)の各式に補
間参照画素それぞれの濃度値を代入することで相関値d
α,dβ,dγを算出する。すなわち、dα=|82−
176|=94、dβ=|160−64|=96、dγ
=|164−60|=104となる。
The correlation value d is obtained by substituting the density values of the interpolation reference pixels into the equations (24), (25) and (26).
Calculate α, dβ, and dγ. That is, dα = | 82-
176 | = 94, dβ = | 160−64 | = 96, dγ
= | 164-60 | = 104.

【0057】算出した相関値dα,dβ,dγを(2
7),(28),(29)式に代入することで相関比率
pα,pβ,pγを算出する。すなわち、pα=(96
+104)/(94+96+104)/2=0.34
0、pβ=(94+104)/(94+96+104)
/2=0.337、pγ=(94+96)/(94+9
6+104)/2=0.323となる。
The calculated correlation values dα, dβ and dγ are (2
Correlation ratios pα, pβ, and pγ are calculated by substituting the equations 7), (28), and (29). That is, pα = (96
+104) / (94 + 96 + 104) /2=0.34
0, pβ = (94 + 104) / (94 + 96 + 104)
/2=0.337, pγ = (94 + 96) / (94 + 9)
6 + 104) /2=0.323.

【0058】算出した相関比率pα,pβ,pγを(3
0)式に代入することで、補間対象画素(x’−1,
y’)の補間値f(x’−1,y’)は算出される。す
なわち、f(x’−1,y’)=1/2×{(82+1
76)×0.340+(160+64)×0.337+
(164+60)×0.323}≒118となる。
The calculated correlation ratios pα, pβ, pγ are (3
By substituting it into the equation (0), the interpolation target pixel (x′−1,
The interpolated value f (x'-1, y ') of y') is calculated. That is, f (x'-1, y ') = 1 / 2x {(82 + 1
76) x 0.340 + (160 + 64) x 0.337 +
(164 + 60) × 0.323} ≈118.

【0059】次に補間対象画素(x’,y’)の補間値
を求める。
Next, the interpolation value of the interpolation target pixel (x ', y') is obtained.

【0060】(24),(25),(26)の各式に補
間参照画素それぞれの濃度値を代入することで相関値d
α,dβ,dγを算出する。すなわち、dα=|160
−170|=10、dβ=|164−176|=12、
dγ=|56−64|=8となる。
The correlation value d is obtained by substituting the density values of the interpolation reference pixels into the equations (24), (25) and (26).
Calculate α, dβ, and dγ. That is, dα = | 160
-170 | = 10, dβ = | 164-176 | = 12,
dγ = | 56−64 | = 8.

【0061】算出した相関値dα,dβ,dγを(2
7),(28),(29)式に代入することで相関比率
pα,pβ,pγを算出する。すなわち、pα=(12
+8)/(10+12+8)/2=0.333、pβ=
(10+8)/(10+12+8)/2=0.300、
pγ=(10+12)/(10+12+8)/2=0.
367となる。
The calculated correlation values dα, dβ and dγ are (2
Correlation ratios pα, pβ, and pγ are calculated by substituting the equations 7), (28), and (29). That is, pα = (12
+8) / (10 + 12 + 8) /2=0.333, pβ =
(10 + 8) / (10 + 12 + 8) /2=0.300,
pγ = (10 + 12) / (10 + 12 + 8) / 2 = 0.
367.

【0062】算出した相関比率pα,pβ,pγを(3
0)式に代入することで、補間対象画素(x’,y’)
の補間値f(x’,y’)は算出される。すなわち、f
(x’,y’)=1/2×{(160+170)×0.
333+(164+176)×0.300+(56+6
4)×0.367≒128となる。
The calculated correlation ratios pα, pβ and pγ are (3
By substituting it into the equation (0), the interpolation target pixel (x ', y')
The interpolated value f (x ', y') of is calculated. That is, f
(X ′, y ′) = ½ × {(160 + 170) × 0.
333+ (164 + 176) × 0.300 + (56 + 6
4) × 0.367≈128.

【0063】次に、補間対象画素(x’+1,y’)の
補間値を求める。
Next, the interpolation value of the interpolation target pixel (x '+ 1, y') is obtained.

【0064】(24),(25),(26)の各式に補
間参照画素それぞれの濃度値を代入することで相関値d
α,dβ,dγを算出する。すなわち、dα=|164
−56|=108、dβ=|56−170|=114、
dγ=|48−176|=128となる。
By substituting the density values of the interpolation reference pixels into the equations (24), (25), and (26), the correlation value d
Calculate α, dβ, and dγ. That is, dα = | 164
−56 | = 108, dβ = | 56−170 | = 114,
dγ = | 48-176 | = 128.

【0065】算出した相関値dα,dβ,dγを(2
7),(28),(29)式に代入することで相関比率
pα,pβ,pγを算出する。すなわち、pα=(11
4+128)/(108+114+128)/2=0.
345、pβ=(108+128)/(108+114
+128)/2=0.337、pγ=(108+11
4)/(108+114+128)/2=0.317と
なる。
The calculated correlation values dα, dβ and dγ are (2
Correlation ratios pα, pβ, and pγ are calculated by substituting the equations 7), (28), and (29). That is, pα = (11
4 + 128) / (108 + 114 + 128) / 2 = 0.
345, pβ = (108 + 128) / (108 + 114)
+128) /2=0.337, pγ = (108 + 11)
4) / (108 + 114 + 128) /2=0.317.

【0066】算出した相関比率pα,pβ,pγを(3
0)式に代入することで、補間対象画素(x’+1,
y’)の補間値f(x’+1,y’)は算出される。す
なわち、f(x’+1,y’)=1/2×{(164+
56)×0.345+(56+170)×0.337+
(48+176)×0.317≒112となる。
The calculated correlation ratios pα, pβ and pγ are (3
By substituting into the equation (0), the interpolation target pixel (x ′ + 1,
The interpolation value f (x '+ 1, y') of y ') is calculated. That is, f (x ′ + 1, y ′) = 1/2 × {(164+
56) x 0.345 + (56 + 170) x 0.337 +
(48 + 176) × 0.317≈112.

【0067】このように、斜線部S内の補間画素の補間
値が周囲と比較して同等の値となるので、斜線Sを途切
れさせることなく、画素の補間、ひいては画像の拡大が
できる。
In this way, since the interpolated value of the interpolated pixel in the shaded area S becomes the same value as that of the surrounding area, it is possible to interpolate the pixel without interruption of the shaded area S and to enlarge the image.

【0068】なお、上記した各実施例では、補間データ
は各画素の濃度値としていたが、画像の輝度、色相など
画像データとして扱えるものならば、どのようなデータ
であっても適用できるのはいうまでもない。
In each of the above-described embodiments, the interpolation data is the density value of each pixel, but any data that can be handled as image data such as image brightness and hue can be applied. Needless to say.

【0069】[0069]

【発明の効果】以上のように本発明によれば、画像の補
間において、斜線のぼやけやジャギー(ギザギザ)とい
った視覚的な劣化を軽減することができた。
As described above, according to the present invention, it is possible to reduce visual deterioration such as blurring of diagonal lines and jaggies (jaggies) in image interpolation.

【0070】また、斜線や境界線が補間対象画素の幅よ
り狭い(細く)、かつ斜線内に位置する補間参照画素の
組み合わせの相関値がその以外の補間参照画素の組み合
わせの相関値より少しでも大きい(相関が弱い)場合で
あっても、斜線や境界線が途切れてしまうといった画像
の乱れを軽減することができた。
Further, the diagonal line or the boundary line is narrower (thinner) than the width of the pixel to be interpolated, and the correlation value of the combination of interpolation reference pixels located within the diagonal line is a little smaller than the correlation value of the combination of other interpolation reference pixels. Even if it is large (the correlation is weak), it is possible to reduce the image disturbance such as the broken lines and the boundary lines being interrupted.

【0071】さらには、比較演算といった複雑な演算を
行う必要がないので、補間演算の回路を簡単かつ小型化
することができた。
Furthermore, since it is not necessary to perform complicated calculation such as comparison calculation, the circuit for interpolation calculation can be simplified and downsized.

【図面の簡単な説明】[Brief description of drawings]

【図1】本発明の第1、第2の実施例、および第1の従
来例に係る画像データの構成図である。
FIG. 1 is a configuration diagram of image data according to first and second embodiments of the present invention and a first conventional example.

【図2】本発明の第1実施例に係る画像データの構成図
である。
FIG. 2 is a configuration diagram of image data according to the first embodiment of the present invention.

【図3】境界線を含んだ画像データの構成図である。FIG. 3 is a configuration diagram of image data including a boundary line.

【図4】第3実施例およぞ第2の従来例に係る画像デー
タの構成図である。
FIG. 4 is a configuration diagram of image data according to a third embodiment and a second conventional example.

【図5】斜線を含んだ画像データの構成図である。FIG. 5 is a configuration diagram of image data including diagonal lines.

【図6】第1の従来例を示した模式図である。FIG. 6 is a schematic diagram showing a first conventional example.

【図7】第2の従来例のフローチャートである。FIG. 7 is a flowchart of a second conventional example.

【符号の説明】[Explanation of symbols]

(x’,y’) 補間対象画素 f(x’,y’) 補間対象画素の補間値 (X ', y') interpolation target pixel f (x ', y') interpolation value of interpolation target pixel

Claims (3)

【特許請求の範囲】[Claims] 【請求項1】 補間対象画素を囲んで位置する複数の補
間参照画素を基にして前記補間対象画素の画素データを
補間する画素補間方法であって、 前記補間参照画素間の画素データ相関具合を示す相関デ
ータを算出するとともに、算出した相関データに対応し
て各補間参照画素の画素データに重み付けを行うことを
特徴とする画素補間方法。
1. A pixel interpolation method for interpolating the pixel data of the interpolation target pixel based on a plurality of interpolation reference pixels located surrounding the interpolation target pixel, wherein the pixel data correlation between the interpolation reference pixels is determined. A pixel interpolation method characterized by calculating the correlation data shown and weighting the pixel data of each interpolation reference pixel corresponding to the calculated correlation data.
【請求項2】 補間対象画素を挟んで対向する補間参照
画素間の画素データの相関具合を示す相関データを、対
向補間参照画素の組み合わせ毎に算出するとともに、算
出した各相関データを比較して、相関関係が強い対向補
間参照画素の組み合わせほどその画素データに重み付け
を行うことを特徴とする請求項1記載の画素補間方法。
2. Correlation data indicating the degree of correlation of pixel data between interpolation reference pixels facing each other with an interpolation target pixel sandwiched therebetween is calculated for each combination of opposed interpolation reference pixels, and the calculated correlation data are compared with each other. 2. The pixel interpolation method according to claim 1, wherein the pixel data is weighted for a combination of counter interpolation reference pixels having a strong correlation.
【請求項3】 各補間参照画素の画素データ平均値を算
出するとともに、算出した画素データ平均値と各補間参
照画素の画素データとの間の差の絶対値からなる相関デ
ータを作成するとともに、算出した相関データが些少で
ある補間参照画素ほどその画素データに重み付けを行う
ことを特徴とする請求項1記載の画素補間方法。
3. An average value of pixel data of each interpolation reference pixel is calculated, and correlation data composed of an absolute value of a difference between the calculated average value of pixel data and pixel data of each interpolation reference pixel is created. 2. The pixel interpolation method according to claim 1, wherein the interpolation reference pixel whose calculated correlation data is insignificant is weighted to the pixel data.
JP7298995A 1995-03-30 1995-03-30 Pixel interpolation method Pending JPH08274983A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP7298995A JPH08274983A (en) 1995-03-30 1995-03-30 Pixel interpolation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP7298995A JPH08274983A (en) 1995-03-30 1995-03-30 Pixel interpolation method

Publications (1)

Publication Number Publication Date
JPH08274983A true JPH08274983A (en) 1996-10-18

Family

ID=13505328

Family Applications (1)

Application Number Title Priority Date Filing Date
JP7298995A Pending JPH08274983A (en) 1995-03-30 1995-03-30 Pixel interpolation method

Country Status (1)

Country Link
JP (1) JPH08274983A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012114574A1 (en) * 2011-02-21 2012-08-30 三菱電機株式会社 Image magnification device and method
JP2015138357A (en) * 2014-01-21 2015-07-30 住友重機械工業株式会社 Image generation device and operation support system

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2012114574A1 (en) * 2011-02-21 2012-08-30 三菱電機株式会社 Image magnification device and method
CN103380615A (en) * 2011-02-21 2013-10-30 三菱电机株式会社 Image magnification device and method
JP5701374B2 (en) * 2011-02-21 2015-04-15 三菱電機株式会社 Image enlargement apparatus and method
US9020303B2 (en) 2011-02-21 2015-04-28 Mitsubishi Electric Corporation Image magnification device and method
JP2015138357A (en) * 2014-01-21 2015-07-30 住友重機械工業株式会社 Image generation device and operation support system
WO2015111262A1 (en) * 2014-01-21 2015-07-30 住友重機械工業株式会社 Image generation device and operation assistance system
CN106415656A (en) * 2014-01-21 2017-02-15 住友重机械工业株式会社 Image generation device and operation assistance system
US10134119B2 (en) 2014-01-21 2018-11-20 Sumitomo Heavy Industries, Ltd. Image generation device and operation support system
CN106415656B (en) * 2014-01-21 2019-07-30 住友重机械工业株式会社 Video generation device and operations support systems

Similar Documents

Publication Publication Date Title
US5991463A (en) Source data interpolation method and apparatus
US20080285889A1 (en) Image transform method for obtaining expanded image data, image processing apparatus and image display device therefore
US5793379A (en) Method and apparatus for scaling images having a plurality of scan lines of pixel data
JP3747523B2 (en) Image processing apparatus and processing method
JPH05334427A (en) Method for expanding/contracting image
EP0264966B1 (en) Interpolator for television special effects system
US5930407A (en) System and method for efficiently generating cubic coefficients in a computer graphics system
EP0860080B1 (en) Method and apparatus for de-interlacing video fields to progressive scan video frames
JP3746311B2 (en) Input video image data filtering method
US20030048278A1 (en) Motion adaptive de-interlacing circuit and method
WO2001097171A2 (en) Magnifying digital image using edge mapping
JPH08274983A (en) Pixel interpolation method
US8368809B2 (en) Frame rate conversion with motion estimation in a plurality of resolution levels
WO2000057631A1 (en) Image processing device and processing method
US8264601B2 (en) Video picture format conversion method and corresponding device
US20040066467A1 (en) Gradient-aided apparatus and method for pixel interpolation
US20020146180A1 (en) Image processing method and image processing device
US7697817B2 (en) Image processing apparatus and method, and recorded medium
US7949204B2 (en) Image scaling interpolation for reduced pictures
JP3327961B2 (en) Image processing device
JPH0738805A (en) Image enlarging device
JP2910986B2 (en) Image data extraction method
KR0178241B1 (en) Method for preventing the blocking-effect in image expansion
JPH06113120A (en) Resolution conversion device for picture processor
KR100328585B1 (en) Method for Digital Image Enlargement Using Delta-Shape Pseudomedian Filter