JP3709598B2 - Dither matrix creation method - Google Patents

Description

【００４０】
２値化誤差和Ｅは、次式２のごとく、係数マトリックスαと周辺画素の２値化誤差ｅとに基づいて計算される。
【００４１】
【数２】

【００４２】
係数マトリックスαは、本実施の形態１では、次のようなマトリックスが用いられている。αabは係数マトリックスαの位置（ａ，ｂ）の係数を表している。またｅabは係数マトリックスαの位置（ａ，ｂ）の係数に対応する位置の周辺画素の２値化誤差を表している。
【００４３】
【数３】

【００４４】
ここで、＊は注目画素の位置を表している。
このようにして求められた補正濃度値Ｉは閾値ｔ（本実施の形態１では例えばｔ＝１２８）と比較されて、Ｉ＞ｔならば、注目画素をオン（「１」）に２値化し、Ｉ≦ｔならば、注目画素をオフ（「０」）に２値化する。
【００４５】
次いで注目画素にオンに２値化した場合、次式の計算により、注目画素の２値化誤差ｅが求められる。
【００４６】
【数４】

【００４７】
また、注目画素をオフに２値化した場合には、次式のごとく、注目画素の２値化誤差ｅにＩが設定される。
【００４８】
【数５】

【００４９】
このようにして、通常の誤差拡散法による２値化処理（Ｓ１１６）を終了し、次に、現在の注目画素が特定領域Ａに含まれているか否かが判定される（Ｓ１１８）。特定領域Ａとは、図５に示したごとく、集積結果マトリックスＭ１を形成するために、２値化値のカウント処理を行う領域であり、本実施の形態１では全ての均一濃度画素マトリックスＤ０〜Ｄ２５５において同一の位置に存在する。
【００５０】
この特定領域Ａは、誤差拡散処理における先頭画素ラインＬ０部分は含まず、最終画素ラインＬｘ部分を含んでいる。これは、先頭画素ラインＬ０部分では、その前のラインが存在しないため２値化誤差の拡散において、後方のラインと比較して２値化誤差の分配に歪みを生じており、ドットの集中や特定のパターンが生じ易いからであり、この２値化誤差拡散の歪みの影響は先頭画素ラインＬ０から後方に離れるに従って少なくなり、最終画素ラインＬｘ部分では最も影響が少ないからである。
【００５１】
現在の注目画素が特定領域Ａに含まれていると判定されると（Ｓ１１８で「ＹＥＳ」）、注目画素の２値化値がワーキングメモリ１４に保存される（Ｓ１２０）。現在の注目画素が特定領域Ａに含まれていないと判定されると（Ｓ１１８で「ＮＯ」）、注目画素の２値化値は保存されない。
【００５２】
次に濃度値ｉにおいて未処理の画素が有るか否かが判定されて（Ｓ１２２）、未処理画素が有れば（Ｓ１２２で「ＹＥＳ」）、再度ステップＳ１１２に戻り、未処理画素を注目画素として上述の処理を行う。
注目画素が所定領域（前方縁部領域ＢＦ）内であると判定されると（Ｓ１１４で「ＹＥＳ」）、次のような２値化処理が行われる（Ｓ１２４）。
【００５３】
すなわち、今回の注目画素について、前記式１，２，３，４の演算は行うが、２値化のための補正濃度値Ｉと閾値ｔとの比較は実行されず、主走査方向へｍ画素戻った位置の画素に対する２値化の結果を、そのまま、今回の注目画素の２値化結果として採用する。この２値化結果がオンであれば式３の演算が行われ、オフであれば式４の演算が行われる。すなわち、閾値ｔとの比較は行わずに、ｍ画素前の２値化結果に応じた誤差拡散法を実行する。
【００５４】
前記値ｍは、特定領域Ａの主走査方向での長さである。したがって、前方縁部領域ＢＦ内の画素については、特定領域Ａの内で主走査方向の２値化処理が外部領域Ｂから特定領域Ａへと横切る境界ＣＦに接する前方縁部領域ＡＦにおける２値化結果がそのまま採用されることになる。すなわち、図６において特定領域Ａの右側の境界ＣＲの外側に接する３つの画素列Ｑ1 ，Ｑ2 ，Ｑ3 からなる前方縁部領域ＢＦは、特定領域Ａの左側の境界ＣＦの内側に接する３つの画素列Ｐ1 ，Ｐ2 ，Ｐ3 からなる前方縁部領域ＡＦとまったく同じ２値化状態となる。例えば、画素ｂ1 の２値化は、画素ａ1 の２値化結果がそのまま用いられる。
【００５５】
このようにして、各画素の２値化がなされ、濃度値ｉの均一濃度画素マトリックスの全ての画素について２値化が終了すれば（Ｓ１２２で「ＮＯ」）、図２に示すごとく、濃度値ｉ＝２５５か否かを判定する（Ｓ１３０）。最初は濃度値ｉ＝０であるので（Ｓ１３０で「ＮＯ」）、次に濃度値ｉがインクリメントされる（Ｓ１４０）。したがって、次に全画素が濃度値ｉ＝１の均一濃度画素マトリックスに対して前述した２値化処理を実行する（Ｓ１１０）。
【００５６】
以後、順次、濃度値ｉをインクリメントしつつ（Ｓ１４０）、該当する濃度値ｉの均一濃度画素マトリックスを前述したように２値化する（Ｓ１１０）処理を繰り返す。
濃度値ｉ＝２５５の均一濃度画素マトリックスの２値化処理（Ｓ１１０）が終了すると図７に示すごとく、ワーキングメモリ１４内には、濃度値ｉ＝０〜２５５の２５６個の２値化画素マトリックスＦ０〜Ｆ２５５が形成されている。
【００５７】
次に濃度値ｉ＝２５５であるので（Ｓ１３０で「ＹＥＳ」）、全ての２値化画素マトリックスについて、同一位置の画素の内、オンに２値化されている数をカウントし、このカウント値を要素とする集積結果マトリックスＭ１を形成する（Ｓ１５０）。
【００５８】
すなわち、図７に示すごとく、左上隅を原点（０，０）として横方向をｘ軸、縦方向をｙ軸とすると、まず、全ての２値化画素マトリックスＦ０〜Ｆ２５５の（０，０）について、オンとなった画素をカウントする。そのカウント結果を、図８に示すごとくワーキングメモリ１４内に用意された集積結果マトリックスＭ１の同一位置（０，０）に格納する。このカウント処理を各画素位置について行い、集積結果マトリックスＭ１をすべて埋める。このカウント処理は、各画素がオンの場合は「１」に、オフの場合は「０」に２値化されているので、同一位置の画素における数値を合計して求めても良い。
【００５９】
次に、集積結果マトリックスＭ１のカウント値の低い画素位置から、順にディザマトリックス用の閾値を設定し（Ｓ１６０）、その閾値からなるマトリックスをディザマトリックスとして、ディザマトリックス格納メモリ１６に保存する（Ｓ１７０）。
【００６０】
この閾値の設定の方法としては、カウント値そのものを閾値として設定しても良い。この場合はステップＳ１５０のカウント処理がディザマトリックスの閾値設定の処理に該当するので、ステップＳ１６０では特に処理は行わない。そして、集積結果マトリックスＭ１そのものをディザマトリックスとして、ディザマトリックス格納メモリ１６に保存する（Ｓ１７０）。
【００６１】
ステップＳ１６０にては、単に集積結果マトリックスＭ１の各カウント値自体をそのまま閾値とする代りに、予め用意されたカウント値と閾値とのテーブルに応じて、集積結果マトリックスＭ１の各カウント値を閾値に変換しても良い。
また、ステップＳ１６０にては、カウント値の低い順番から、閾値を「０，１，２，…」と設定して行っても良いし、逆にカウント値の高い順番から、閾値を「２５５，２５４，２５３，…」と設定して行っても良い。この場合、同一のカウント値が２つ以上存在する場合は、それらはすべて同一の閾値が与えられるものとしても良いし、同一のカウント値を順番付けして、異なるカウント値と同様に閾値を設定しても良い。
【００６２】
このようにして形成されたディザマトリックスを用いて、中間調のカラー画像を２値化処理し、カラープリンタにて記録したところ、人間の目に対してノイジーとならず、かつ解像度が悪化せず、紋様の発生が抑制された疑似中間調の２値化画像を生成することができた。勿論、ディザ法であり、誤差拡散法のように多数の計算は行わないので、迅速に処理できた。
【００６３】
更に、本実施の形態１では、均一濃度画素マトリックスから、先頭画素ラインＬ０を含まず、かつ最終画素ラインＬｘを含む特定領域Ａの２値化画素マトリックスＦ０〜Ｆ２５５を求め、この２値化画素マトリックスＦ０〜Ｆ２５５から、均一濃度画素マトリックスより小さい集積結果マトリックスＭ１を求めて、この集積結果マトリックスＭ１に基づいてディザマトリックスを形成している。このため、先頭画素ラインＬ０部分で生じている誤差分配の歪みの影響が及びにくく、一層、紋様の発生が抑制された疑似中間調の２値化画像を生成することができた。
【００６４】
更に、前述のごとく、前方縁部領域ＢＦ内の画素については、特定領域Ａの前方縁部領域ＡＦの２値化結果がそのまま採用されている。したがって、誤差分配は前記係数マトリックスαの形状に応じた位置の周辺画素から注目画素に２値化誤差が分配されるが、境界ＣＲの左側に接している特定領域Ａの後方縁部領域ＡＲについても直接あるいは間接に前方縁部領域ＢＦ内の画素からの２値化誤差分配が影響する。この前方縁部領域ＢＦは、特定領域Ａの前方縁部領域ＡＦの２値化結果と同じとなっているので、特定領域Ａの前方縁部領域ＡＦと特定領域Ａの後方縁部領域ＡＲとの処理の連続性が維持される。そのため、以下の処理にて作成されたディザマトリックスを用いて、中間調画像を２値化しても、ディザマトリックスで処理された領域間に色むらや明暗むらを生じるのを抑制でき、疑似的境界の発生を抑制できる。
【００６５】
尚、ステップＳ１２２での未処理画素の判定は、特定領域Ａ内に未処理画素が有るか否かの判定でも良い。
［実施の形態２］
本実施の形態２は、前記実施の形態１に対して、ステップＳ１１０の２値化処理が図９，図１０に示すごとくである点と、ステップＳ１１６０，Ｓ２０４０，Ｓ２０５０にて行われる誤差拡散法として、注目画素を２値化した場合に２値化の誤差を未だ２値化していない周辺の画素の濃度に分配する方法による点とが異なる。
【００６６】
尚、図９においては、ステップＳ１１２０，Ｓ１１８０，Ｓ１２００，Ｓ１２２０の処理は、実施の形態１の図３にしめすステップＳ１１２，Ｓ１１８，Ｓ１２０，Ｓ１２２の処理と同じである。以下、特に実施の形態１と異なる点を中心に説明する。
【００６７】
まず、ステップＳ１１２０の次に、逆転領域ＲＶか否かを判定し（Ｓ１１２２）、逆転領域ＲＶであれば（Ｓ１１２２で「ＹＥＳ」）、逆方向２値化処理（Ｓ２０００）が行われる。
前記逆転領域ＲＶは、図１１に示すごとく、特定領域Ａの後方縁部領域ＡＲ直前の画素列Ｐm-3 と、特定領域Ａの後方縁部領域ＡＲの全部（画素列Ｐm-2 ，Ｐm-1 ，Ｐm ）と、外部領域Ｂの前方縁部領域ＢＦの全部（画素列Ｑ1 ，Ｑ2 ，Ｑ3 ）とからなる領域である。
【００６８】
したがって、主走査方向に行われる２値化処理が画素列Ｐm-3 に達しない内は（Ｓ１１２２で「ＮＯ」）、ステップＳ１１６０〜Ｓ１２２０の処理が実行され、通常の誤差拡散法により注目画素の２値化がなされるとともに、２値化誤差が計算されて未だ２値化されていない周辺画素に分配される。
【００６９】
すなわち、次式５のごとく、既に２値化された周辺画素から注目画素へ分配されている２値化誤差和Ｅにて濃度値ｉが補正されて補正濃度値Ｉが計算される。
【００７０】
【数６】

【００７１】
このようにして求められた補正濃度値Ｉは閾値ｔ（本実施の形態２では例えばｔ＝１２８）と比較されて、Ｉ＞ｔならば、注目画素をオン（「１」）に２値化し、Ｉ≦ｔならば、注目画素をオフ（「０」）に２値化する。
次いで注目画素にオンに２値化した場合、次式６の計算により、注目画素の２値化誤差ｅが求められる。
【００７２】
【数７】

【００７３】
また、注目画素をオフに２値化した場合には、次式７のごとく、注目画素の２値化誤差ｅにＩが設定される。
【００７４】
【数８】

【００７５】
次に、２値化誤差ｅが未２値化の周辺画素に、下に示す係数マトリックスβに基づいて分配される。
【００７６】
【数９】

【００７７】
ここで、＊は注目画素の位置を表し、数値は、数値に対応する位置の注目画素に対して、数値を係数としてかけた２値化誤差ｅを分配することを表している。次のステップＳ１１８０〜ステップＳ１２２０の処理は実施の形態１のステップＳ１１８〜Ｓ１２２と同じである。
【００７８】
主走査方向に行われる２値化処理が画素列Ｐm-3 に達した場合、すなわち逆転領域ＲＶに入った場合は（Ｓ１１２２で「ＹＥＳ」）、ステップＳ２０００の処理が実行される。
図１０のフローチャートに示すごとく、まず、２値化処理対象画素が設定される（Ｓ２０１０）。この設定について、図１１に示す「…，ａ，ｂ，ｃ，ｄ，ｅ，…，Ｒ，Ｓ，Ｔ，Ｕ，Ｖ，Ｗ，Ｘ，Ｙ，Ｚ，…」からなる画素行を用いて説明すると、ステップＳ１２２０からステップＳ１１２０に戻るループにて、注目画素として画素Ｓが設定されると、その設定とは別に実際の２値化処理対象の画素として画素Ｙが設定される処理が行われる。次に注目画素として画素Ｔが設定されると２値化処理対象画素としては画素Ｘが設定され、注目画素として画素Ｕが設定されると２値化処理対象画素としては画素Ｗが設定され、注目画素として画素Ｖが設定されると２値化処理対象画素としては画素Ｖが設定され、注目画素として画素Ｗが設定されると２値化処理対象画素としては画素Ｕが設定され、注目画素として画素Ｘが設定されると２値化処理対象画素としては画素Ｔが設定され、注目画素として画素Ｙが設定されると２値化処理対象画素としては画素Ｓが設定される。すなわち、逆転領域ＲＶでは、他の領域とは逆方向に２値化処理が進むことになる。
【００７９】
次に、２値化処理対象画素の濃度値と２値化誤差和Ｅとが新たに読み込まれる（Ｓ２０１５）。そして、次に、誤差拡散法による２値化処理のために、２値化処理対象画素位置に応じた係数マトリックスが指定される（Ｓ２０２０）。前述した逆転領域ＲＶ以外で使用する係数マトリックスβを用いないのは、主走査方向が逆転するため、２値化処理対象画素に対する未２値化周辺画素の配置が異なるからである。
【００８０】
逆転領域ＲＶ内の７つの画素列Ｐm-3 ，Ｐm-2 ，Ｐm-1 ，Ｐm ，Ｑ1 ，Ｑ2 ，Ｑ3 に対して、係数マトリックスが、それぞれ、未２値化周辺画素の配置に適合させた係数マトリックスβ1 ，β2 ，β3 ，β4 ，β5 から選択されて適用される。各係数マトリックスβ1 ，β2 ，β3 ，β4 ，β5 の一例を下に示す。
【００８１】
【数１０】

【００８２】
【数１１】

【００８３】
【数１２】

【００８４】
【数１３】

【００８５】
【数１４】

【００８６】
逆転領域ＲＶ内の画素列Ｐm-3 には係数マトリックスβ5 が適用され、画素列Ｐm-2 には係数マトリックスβ4 が適用され、画素列Ｐm-1 ，Ｐm ，Ｑ1 には係数マトリックスβ3 が適用され、画素列Ｑ2 には係数マトリックスβ2 が適用され、画素列Ｑ3 には係数マトリックスβ1 が適用される。
【００８７】
次に、ステップＳ２０３０の判定にて、ステップＳ２０４０あるいはステップＳ２０５０が実行される。ステップＳ２０４０の処理は、係数マトリックスとしてステップＳ２０２０にて指定された係数マトリックスを用いる以外は、前述したステップＳ１１６０の処理と同じである。
【００８８】
すなわち、今回の注目画素について、前記式５，６，７の演算および２値化誤差ｅの分配は行うが、２値化のための補正濃度値Ｉと閾値ｔとの比較は実行されず、主走査方向へｍ画素戻った位置の画素に対する２値化の結果を、そのまま、今回の注目画素の２値化結果として採用する。この２値化結果がオンであれば式６の演算が行われ、オフであれば式７の演算が行われる。すなわち、閾値ｔとの比較は行わずに、ｍ画素前の２値化結果に応じて２値化誤差を算出して、周辺画素に係数マトリックスβに基づいて分配する。
【００８９】
前記値ｍは、特定領域Ａの主走査方向での長さである。したがって、前方縁部領域ＢＦ内の画素については、特定領域Ａの内で主走査方向の２値化処理が外部領域Ｂから特定領域Ａへと横切る境界ＣＦに接する前方縁部領域ＡＦにおける２値化結果がそのまま採用されることになる。すなわち、図６において特定領域Ａの右側の境界ＣＲの外側に接する３つの画素列Ｑ1 ，Ｑ2 ，Ｑ3 からなる前方縁部領域ＢＦは、特定領域Ａの左側の境界ＣＦの内側に接する３つの画素列Ｐ1 ，Ｐ2 ，Ｐ3 からなる前方縁部領域ＡＦとまったく同じ２値化状態となる。例えば、画素Ｗの２値化は、画素ａの２値化結果がそのまま用いられる。
【００９０】
すなわち、画素Ｒの２値化処理を行った後、次に画素Ｙの２値化を行うが、画素Ｙは所定領域（前方縁部領域ＢＦ）内であるので（Ｓ２０３０で「ＹＥＳ」）、２値化自体は既に２値化結果が得られている画素ｃと同じ値を設定し、２値化誤差の分配は、係数マトリックスβ1 にて行う。次に画素Ｘの２値化となり、ここでも２値化自体は既に２値化結果が得られている画素ｂと同じ値を設定し、２値化誤差の分配は、係数マトリックスβ2 にて行う。次に画素Ｗの２値化となり、ここでも２値化自体は既に２値化結果が得られている画素ａと同じ値を設定し、２値化誤差の分配は、係数マトリックスβ3 にて行う。
次に画素Ｖの２値化となるが、画素Ｖは所定領域（前方縁部領域ＢＦ）内ではないので（Ｓ２０３０で「ＮＯ」）、閾値ｔとの比較により２値化し、２値化誤差の分配は、係数マトリックスβ3 にて行う。次の画素Ｕの処理も画素Ｖの場合と同じである。次に画素Ｔの２値化となり、ここでも閾値ｔとの比較により２値化し、２値化誤差の分配は、係数マトリックスβ4 にて行う。次に画素Ｓの２値化となり、ここでも閾値ｔとの比較により２値化し、２値化誤差の分配は、係数マトリックスβ5 にて行う。
【００９１】
そして、次に注目画素は逆転領域ＲＶ内から出て画素Ｚとなるので（Ｓ１１２２で「ＮＯ」）、画素Ｚ自身の２値化処理となり、ステップＳ１１６０の処理に任される。
本実施の形態２は上述のごとく構成されているため、実施の形態１と同じ効果を生じると共に、更に、境界ＣＲを挟んだ逆転領域ＲＶにて２値化の主走査方向を逆転させているため、前記実施の形態１に比較して、境界ＣＲの右側に接している前方縁部領域ＢＦ内の画素から境界ＣＲの左側に接している特定領域Ａの後方縁部領域ＡＲに対する２値化誤差分配の影響を十分に大きくすることができる。すなわち、前方縁部領域ＡＦから後方縁部領域ＡＲへの２値化誤差分配が十分に効果的なものとなる。このため、誤差拡散法による２値化処理において、特定領域Ａの前方縁部領域ＡＦと特定領域Ａの後方縁部領域ＡＲとの処理の連続性が十分に維持されるので、本実施の形態２の処理にて作成されたディザマトリックスを用いて、中間調画像を２値化しても、ディザマトリックスによる処理領域間での疑似的境界の発生を十分に抑えることができる。
【００９２】
［実施の形態３］
本実施の形態３は、実施の形態１，２と異なる点はディザマトリックス作成処理であり、均一濃度画素マトリックスの各画素の２値化処理が以下のように行われる点である。すなわち、２値化処理しようとする均一濃度画素マトリックスの濃度値ｉ未満で最も濃度値ｉに近い濃度値の均一濃度画素マトリックスを２値化処理した最近下方２値化画素マトリックスＬにおいてオンおよび濃度値ｉを越えていて最も濃度値ｉに近い濃度値の均一濃度画素マトリックスを２値化処理した最近上方２値化画素マトリックスＨにおいてもオンとなっている画素位置と同じ画素位置は必ずオンとし、最近下方２値化画素マトリックスにおいてオフおよび最近上方２値化画素マトリックスにおいてもオフとなっている画素位置と同じ画素位置は必ずオフとされる条件下に誤差拡散処理される。更に各均一濃度画素マトリックスの２値化処理順序が、処理対象の均一濃度画素マトリックスの濃度値ｉが、最近下方２値化画素マトリックスと最近上方２値化画素マトリックスとの各濃度値の中央の整数値に対応するようにして、順次処理されている点も異なる。
【００９３】
本実施の形態３のディザマトリックス作成処理を図１２，１３，１４に示す。
まず、全画素が濃度値ｉ＝０である均一濃度画素マトリックスと、全画素が濃度値ｉ＝２５５である均一濃度画素マトリックスとを誤差拡散法により２値化する（Ｓ３０８０）。この均一濃度画素マトリックスは、図４に示すごとく、ｉ＝０，２５５の２種類の均一濃度画素マトリックスＤ０，Ｄ２５５をワーキングメモリ１４に読み込んで用いれば良い。そして、その２値化画素マトリックスＦ０，Ｆ２５５（図７）をワーキングメモリ１４に保存する（Ｓ３０９０）。ただし、ステップＳ３０８０にて行われる濃度値ｉ＝０，２５５の均一濃度画素マトリックスの誤差拡散法による２値化結果は、ｉ＝０の場合は全画素「０」（オフ）、ｉ＝２５５の場合は全画素「１」（オン）となるので、特に誤差拡散法による計算はせずに、予め全画素がオフの２値化画素マトリックスと全画素がオンの２値化画素マトリックスとを備えておいて、これらのマトリックスを用いても良い。
【００９４】
次に１〜２５４の濃度値について濃度値配列の作成を行い（Ｓ３１００）、次にその配列から一つづつ濃度値ｉに読み込んで（Ｓ３１０５）、ステップＳ３１１０の処理を行い、配列の全ての濃度値について処理が終了すれば、ステップＳ３１３０で配列終了と判定されて、ステップＳ３１５０，Ｓ３１６０，Ｓ３１７０の処理が行われる。このステップＳ３１５０，Ｓ３１６０，Ｓ３１７０の処理は実施の形態１のステップＳ１５０，Ｓ１６０，Ｓ１７０と同じである。
【００９５】
ステップＳ３１００における濃度値の配列は、２値化処理された濃度値の中央の濃度値がｉに設定されるように作成される。すなわち、配列の先頭は「０」と「２５５」の中央の整数値である「１２８」（もう一つの中央の整数値の「１２７」でも良い。）、次が、「０」と「１２８」との中央の「６４」、「１２８」と「２５５」との中央の「１９２」（もう一つの中央の整数値の「１９１」でも良い。）、次が「０」と「６４」との中央の「３２」、「６４」と「１２８」との中央の「９６」、「１２８」と「１９２」との中央の「１６０」、「１９２」と「２５５」との中央の「２２４」（もう一つの中央の整数値の「２２３」でも良い。）と言うように、前に得られている値の中央の整数値を求めて、「１」〜「２５４」までの濃度値の配列「１２８，６４，１９２，３２，９６，１６０，２２４，……」を作成する。この配列は予め計算して配列データとして記憶しておき、本処理時に配列データのみ読み込んで用いても良い。
【００９６】
ステップＳ３１０５においては、この配列の先頭から順次、濃度値ｉに設定し、ステップＳ３１１０の２値化処理を行い、配列の最後までステップＳ３１１０の処理が終了すれば、ステップＳ３１３０にて「ＹＥＳ」と判定されて、閾値の設定処理に移る（Ｓ３１５０，Ｓ３１６０，Ｓ３１７０）。
【００９７】
次にステップＳ３１１０の２値化処理について説明する。本処理は、前記実施の形態１の２値化処理とは、ステップＳ１１６の処理が図１３のごとくの処理に代りのみで、後の処理は図３にて説明した内容と同じである。
すなわち、ステップＳ１１４にて所定領域内でないと判定された場合には、ステップＳ４２００の処理が行われ、２値化済みの濃度値、すなわち既に２値化画素マトリックスが求められている濃度値の内で、最近上方２値化画素マトリックスＨと、最近下方２値化画素マトリックスＬとをワーキングメモリ１４の内容から検索する（Ｓ４２００）。
【００９８】
次に、濃度値ｉの均一濃度画素マトリックスについて、誤差拡散法により２値化し、２値化画素マトリックスＦｉを作成する（Ｓ４２１０）。
ただし、次の条件▲１▼，▲２▼の下に２値化される。
▲１▼最近上方２値化画素マトリックスＨと最近下方２値化画素マトリックスＬとの両者にて共に「１」オンが配置されている画素位置は、必ず「１」オンとする。
【００９９】
▲２▼最近上方２値化画素マトリックスＨと最近下方２値化画素マトリックスＬとの両者にて共に「０」オフが配置されている画素位置は、必ず「０」オフとする。
すなわち、通常、誤差拡散法においては、周辺の画素から分配された誤差と注目画素の濃度とを合計した値を、閾値と比較して、閾値以上であれば「１」オン、閾値未満であれば「０」オフに２値化しているが、前記▲１▼または▲２▼の条件に該当する画素の場合には、その画素については閾値との比較をせずに、前記▲１▼または▲２▼の条件通りに、「１」オンまたは「０」オフに設定する。勿論、この閾値と比較しない設定の結果は２値化誤差に反映され、周辺画素への分配の対象となる。
【０１００】
前記条件▲１▼▲２▼により、図１４（ａ）に示すごとく、各画素位置において、濃度値の小さい方から見ると、一旦、オンとなると、以後の濃度値ｉおいては必ずオンとなる。
前記▲１▼▲２▼の条件を用いない通常の２値化を行うと、図１４（ｂ）に示すごとく、オンとなった後も同じ画素位置でオフが出現する。このため、通常の２値化のみでは各画素位置におけるオンの数が、０〜２５５内の特定の値に偏る傾向が有るが、本実施の形態３では、０〜２５５の範囲の各種の値に分散して好適な分布となる。したがって、このようにして得られるディザマトリックスは、閾値の値も０〜２５５の範囲に適度に分散して分布するものとなる。したがって、特に、図１４（ｂ）に示したごとく、各均一濃度画素マトリックス独立に２値化している場合に得られるディザマトリックスに比較して、解像度の向上や紋様の抑制を一層改良させたディザマトリックスを得ることができる。
【０１０１】
更に、本実施の形態３では、既に２値化処理された濃度値の中央の濃度ｉを順次２値化して行くために、最近上方２値化画素マトリックスと最近下方２値化画素マトリックスとのそれぞれの２値分散の影響を同程度受けることになり、中央の濃度ｉの２値化画素マトリックスの２値の分散は比較的均一な分布となる。このようにして求めた、均一な分布と条件▲１▼▲２▼とを満たす２値化画素マトリックスに基づいてディザマトリックスは作成されるので、解像度の向上や紋様の抑制上の効果が更に増大する。
【０１０２】
本実施の形態３では、予め、順次、濃度値ｉに中央の値が得られるように配列を作成していたが、配列を予め作成する代りにステップＳ３１０５の処理にて、中央の値を計算にて順次求めて、濃度値ｉに設定しても良い。
中央の値を計算にて求める一例を図１５に示す。この処理は、図１２におけるステップＳ３１００〜Ｓ３１３０の代りに実行される。本処理において、最初は８つの変数Ｉ0 〜Ｉ7 を「０」で初期化し（Ｓ５０１０〜Ｓ５０８０）、次にＩ0 〜Ｉ7 までを合計して、濃度値ｉに設定する（Ｓ５０９０）。この濃度値ｉが「０」であれば、Ｉ7 を「１２８」増加させ（Ｓ５１２０）、次にＩ7 ＞１２８ではないので（Ｓ５１３０で「ＮＯ」）、ステップＳ５０９０の合計処理で濃度値ｉに「１２８」が設定され、ｉ＝０でなく（Ｓ５１００で「ＮＯ」）、ｉ＝２５５でもないので（Ｓ５１１０で「ＮＯ」）、ステップＳ３１１０の処理（図１２の処理と同じ）がなされる。
【０１０３】
次にステップＳ５１２０の処理にてＩ7 ＝２５６となり、ステップＳ５１３０にて「ＹＥＳ」と判定されて、Ｉ6 が「６４」増加され（Ｓ５１４０）、Ｉ6 ＝６４となり、Ｉ6 ＞６４ではないので（Ｓ５１５０で「ＮＯ」）、ステップＳ５０８０にてＩ7 ＝０に初期化され、ステップＳ５０９０ではｉ＝６４となる。したがって、ステップＳ５１００，Ｓ５１１０からステップＳ３１１０に移って、濃度値ｉ＝６４の均一濃度画素マトリックスについて２値化処理がなされる。
【０１０４】
次に、ステップＳ５１２０にてＩ7 ＝１２８となり、ステップＳ５１３０にて「ＮＯ」と判定されて、ステップＳ５０９０では、Ｉ0 〜Ｉ5 ＝０、Ｉ6 ＝６４、Ｉ7 ＝１２８であることから、ｉ＝１９２となり、ステップＳ５１００，Ｓ５１１０の後、濃度値ｉ＝１９２の均一濃度画素マトリックスについて２値化がなされる。
【０１０５】
次にＩ0 〜Ｉ5 ＝０、Ｉ6 ＝６４、Ｉ7 ＝１２８であることから、ステップＳ５１２０からステップＳ５１６０まで進み、ステップＳ５１６０でＩ5 ＝３２となる。Ｉ5 ＞３２ではないので（Ｓ５１７０で「ＮＯ」）、ステップＳ５０７０，Ｓ５０８０が処理されて、Ｉ0 〜Ｉ4 ＝０、Ｉ5 ＝３２、Ｉ6 ＝０、Ｉ7 ＝０となる。したがって、ステップＳ５０９０ではｉ＝３２となり、ステップＳ３１１０では濃度値ｉ＝３２の均一濃度画素マトリックスが２値化される。
【０１０６】
次にＩ7 ＝０となったので、ステップＳ５１２０，Ｓ５１３０を経て、ステップＳ５０９０に戻る。このとき、Ｉ0 〜Ｉ4 ＝０、Ｉ5 ＝３２、Ｉ6 ＝０、Ｉ7 ＝１２８となる。したがって、ステップＳ５０９０ではｉ＝１６０となり、ステップＳ３１１０では濃度値ｉ＝１６０の均一濃度画素マトリックスが２値化される。
【０１０７】
次に、ステップＳ５１２０〜Ｓ５１５０，Ｓ５０８０が処理されて、Ｉ0 〜Ｉ4 ＝０、Ｉ5 ＝３２、Ｉ6 ＝６４、Ｉ7 ＝０となる。したがって、ステップＳ５０９０ではｉ＝９６となり、ステップＳ３１１０では濃度値ｉ＝９６の均一濃度画素マトリックスが２値化される。
【０１０８】
以後、Ｉ0 〜Ｉ7 の値に応じて、ステップＳ５０２０〜Ｓ５０８０，Ｓ５１２０〜Ｓ５２６０のいずれかの処理により、順次、濃度値ｉに、２２４、１６，１４４，８０，…，１２７が設定されて、濃度値ｉの均一濃度画素マトリックスの２値化と保存が行われ、最後に、Ｉ0 ＝１、Ｉ1 ＝２、Ｉ2 ＝４、Ｉ3 ＝８、Ｉ4 ＝１６、Ｉ5 ＝３２、Ｉ6 ＝６４、Ｉ7 ＝１２８となる。このとき、ステップＳ５０９０にて濃度値ｉ＝２５５が設定されるので、ステップＳ５１１０にて「ＹＥＳ」と判定されて、ステップＳ３１５０の処理に移る。このようにして、計算にても中央の値をたどって、均一濃度画素マトリックスの２値化を行うことができる。
【０１０９】
尚、本実施の形態２では、中央の整数値をたどって２値化処理したが、完全に中央の整数値でなくても、１：２の位置、１：３の位置、２：１の位置、３：１の位置といったように、２値化済みの２つの濃度値の中間の値では有るが、小側あるいは大側に偏った整数値をたどって２値化処理しても良い。
【０１１０】
［その他］
前記実施の形態１〜３において、前記カウント値もそのまま集積結果マトリックスＭ１として用いるのではなく、所定の係数をかけても良い。また、濃度値ｉの均一濃度画素マトリックスの２値化値Ｉixy に対して次式のごとく濃度値ｉ毎に異なる係数ｋ(i) を設定して、全ての均一濃度画素マトリックスの合計値Ｃxyを求めることにより、集積結果マトリックスＭ１を作成しても良い。
【０１１１】
【数１５】

【０１１２】
尚、ｘ，ｙは均一濃度画素マトリックスあるいは集積結果マトリックスＭ１上の画素位置を表す。
均一濃度画素マトリックスの２値化は、オン（「１」）またはオフ（「０」）への２値化であったが、いかなる値の２値化でも良く、いずれか一方の値の個数をカウントして、集積結果マトリックスＭ１を形成し、この集積結果マトリックスＭ１をディザマトリックスに変換すれば良い。また、上述したような個数のカウントではなく、２値化値の合計であっても良い。
【０１１３】
前記実施の形態１〜３において、２値化値を保存する特定の領域Ａは、全ての均一濃度画素マトリックスにおいて同じ位置であったが、均一濃度画素マトリックスに応じて異なっていても良い。これは、均一濃度画素マトリックスを構成する濃度値ｉに応じて、先頭画素ラインＬ０部分で生じている誤差拡散の歪みの後方への影響の程度が異なるためである。
【０１１４】
また、このように濃度値ｉに応じて、先頭画素ラインＬ０部分で生じている誤差拡散の歪みが後方へ影響する程度が異なるため、全ての均一濃度画素マトリックスの２値化において誤差拡散の歪みの影響を無くすため、均一濃度画素マトリックスの大きさを濃度値ｉの値に応じて調整しても良い。
【０１１５】
前記実施の形態１〜３において、所定領域（前方縁部領域ＢＦ）の主走査方向の長さは３画素分であるが、これは係数マトリックスの列数に応じた長さにすることが好ましい。すなわち誤差拡散の影響がある長さに設定することが好ましい。
【０１１６】
前記実施の形態１における２値化は、２値化時に周辺画素から２値化誤差の分配を受けるタイプの誤差拡散法、いわゆる平均誤差最小法であったが、実施の形態２のごとく、注目画素を２値化した場合に２値化の誤差を未だ２値化していない周辺の画素の濃度に分配する方法による誤差拡散法であっても良い。
【０１１７】
また逆に、実施の形態２に平均誤差最小法を採用しても良い。この場合には、係数マトリックスβ，β1 ，β2 ，β3 ，β4 ，β5 については、次の内容とする。
【０１１８】
【数１６】

【０１１９】
【数１７】

【０１２０】
【数１８】

【０１２１】
【数１９】

【０１２２】
【数２０】

【０１２３】
【数２１】

【０１２４】
前記実施の形態３における２値化は、実施の形態１のごとく、２値化時に周辺画素から２値化誤差の分配を受けるタイプの誤差拡散法、いわゆる平均誤差最小法でも良く、実施の形態２のごとく、注目画素を２値化した場合に２値化の誤差を未だ２値化していない周辺の画素の濃度に分配する方法による誤差拡散法であっても良い。
【図面の簡単な説明】
【図１】 実施の形態１のディザマトリックス作成装置の主要ブロック図である。
【図２】 実施の形態１のディザマトリックス作成処理のフローチャートである。
【図３】 実施の形態１の２値化処理のフローチャートである。
【図４】 実施の形態１の均一濃度値画像データを示す説明図である。
【図５】 実施の形態１の均一濃度値の画像と特定領域との関係を示す説明図である。
【図６】 特定領域およびその周辺を拡大して示す説明図である。
【図７】 実施の形態１の均一濃度値画像データを２値化した状態を示す説明図である。
【図８】 実施の形態１の集積結果マトリックスの説明図である。
【図９】 実施の形態２の２値化処理のフローチャートである。
【図１０】 実施の形態２の逆方向２値化処理のフローチャートである。
【図１１】 特定領域およびその周辺を拡大して示す説明図である。
【図１２】 実施の形態３のディザマトリックス作成処理のフローチャートである。
【図１３】 実施の形態３の２値化処理の一部を示すフローチャートである。
【図１４】 実施の形態３の各画素位置のオン・オフ設定状態の説明図である。
【図１５】 実施の形態３の均一濃度画素マトリックスの処理順序を決定する方法の他の例を示すフローチャートである。
【図１６】 従来例を示す説明図である。
【符号の説明】
２…ディザマトリックス作成装置 １０…入力部
１１…マイクロコンピュータ部 １２…ＣＰＵ
１３…プログラムメモリ １４…ワーキングメモリ
１６…ディザマトリックス格納メモリ
１７…出力イメージメモリ １８…システムバス
１９…出力部[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a method of creating a dither matrix for converting a multi-value image into a binary image.
[0002]
[Prior art]
The data constituting the image is usually expressed in shades of about 8 bits. However, in a device that outputs this image data as an image that can be seen by a person, for example, an output device such as a printer, a dye such as ink is used as dots on recording paper because of its simple mechanism and ease of control. So-called binary recording, in which only the attachment is performed, is common.
[0003]
When using such a binary recording output device to record an image expressed in 8-bit grayscale, whether or not to apply ink by simply determining the value of each pixel based on a threshold value. Will be determined. However, by simply binarizing with the threshold value in this way, the rich shade state possessed by the 8-bit shade gradation (256 gradations for one color and 256 gradations for each of the three primary colors, approximately 16.7 million colors) can be obtained. It is impossible to express. Therefore, a method of expressing a shaded state in a pseudo manner with such a binary recording output device has been considered. In other words, this is the concept of area gradation in which shading is expressed by the number of dots arranged in a predetermined area.
[0004]
Various proposals have been made as methods for realizing this concept. For example, an error diffusion method, a random number generation type dither method, a systematic dither method, or the like.
[0005]
[Problems to be solved by the invention]
However, these methods have the following problems.
That is, in the error diffusion method, the binarization process is slow due to error diffusion or error distribution calculation, and a unique pattern is likely to occur in the binarized image. The random number generation type dither method is fast because it uses a dither matrix, but because it uses random numbers, there is a problem that the image quality is very noisy, that is, the image quality is such that a lot of noise is filled with the image. It was.
[0006]
As another method using the dither matrix, there is a spiral type belonging to the systematic dither method. However, this spiral type easily collects dots, and the dense one is recognized as a large dot, so the resolution is easily deteriorated. Among the systematic dither methods, there is a Bayer method, but this Bayer method has good resolution because dots are dispersed, but a unique pattern is generated, and the gradation of dark parts is also poor. Met.
[0007]
When processing an image wider than the size of the dither matrix, as shown in FIG. 16, the partial portions of the image are sequentially binarized for each region DM having the size of the dither matrix. In some cases, color unevenness or light / darkness unevenness occurs in the boundary portions BD1 and BD2 of this processing, and a pseudo boundary line such as a white line appears.
[0008]
The present invention provides a dither matrix that does not become noisy even when the dither method is used, the resolution does not deteriorate, the generation of a pattern is suppressed, and a pseudo boundary line does not occur between regions processed by the dither matrix. It is intended to create.
[0009]
[Means for Solving the Problems and Effects of the Invention]
In the dither matrix creation method of the present invention, each of a plurality of density values discretely set in a predetermined range set within a density range that can be represented by a pixel is converted into all pixels in each of a plurality of uniform density pixel matrices. For each uniform density pixel matrix, the density value is binarized to either on / off by the error diffusion method, and the density value of a specific region in a part of the binarized uniform density pixel matrix Based on this, the threshold value of each pixel in the dither matrix is set.
[0010]
When the image is binarized by the dither matrix having the threshold value set in this way, a binarized image is generated which is not noisy to human eyes, does not deteriorate in resolution, and suppresses generation of a pattern. Can do.
The binary sequence produced by the error diffusion method looks irregular at first glance, but is not at all random. That is, there is a property that it is difficult to form a pattern, but it is difficult to form a noisy image. As described above, the dither matrix creation method of the present invention is created by incorporating the characteristics of the binary sequence in the error diffusion method into the dither matrix by using the binary sequence generated by the error diffusion method. .
[0011]
Therefore, in the dither matrix that enables the binarization calculation processing at high speed, an appropriate threshold value distribution that is irregular but not random is formed by the error diffusion method as described above. It is considered that a binarized image that is not noisy to the eyes, does not deteriorate in resolution, and further suppresses the generation of patterns can be generated.
[0012]
Further, in the present invention, the uniform density pixel matrix is a matrix larger than a required dither matrix, and after binarization by the error diffusion method, a threshold value of each pixel in the dither matrix is determined based on the binarization result. Is set. For this reason, error diffusion or error distribution distortion in the initial stage of error diffusion processing hardly affects the dither matrix, and even if the image is binarized using the completed dither matrix, the occurrence of dot bias and patterns are suppressed. can do. The threshold value of the dither matrix based on the binarization result is obtained by, for example, setting the density value of a specific area in a part of the binarized uniform density pixel matrix to pixels in the same position for all the uniform density pixel matrix. It is also possible to perform integration based on the integration result for each pixel at the same position.
[0013]
Furthermore, in the present invention, the boundary is defined with respect to an edge region in the specific region that is in contact with a boundary between the specific region and an external region other than the specific region, or an edge region of the external region that is in contact with the boundary. Boundary processing that maintains the continuity of the binarization processing that is sandwiched is performed.
[0014]
The reason why unevenness in color and lightness and darkness occurs at the boundary between regions processed by the dither matrix is that there is no continuity between the edges at the left and right edges and the top and bottom edges of the dither matrix. In particular, when the dither matrix is created by the binarization processing by the error diffusion method as described above, the discontinuity of the error diffusion processing between edges in the main scanning direction of the binarization processing by the error diffusion method is color. It greatly affects unevenness of light and darkness. If this is maintained by boundary processing, color unevenness and light / darkness unevenness at the boundary of the region processed by the dither matrix can be suppressed, and generation of a pseudo boundary line can be suppressed.
[0015]
As the boundary processing, in particular, for the rear edge region in the specific region or the front edge region of the external region that is in contact with the boundary that the processing in the main scanning direction of binarization crosses from the specific region to the external region It is preferable to be performed.
As the boundary processing, for example, in the front edge region of the external region, the processing in the main scanning direction of binarization is performed in the front edge region in the specific region in contact with the boundary crossing from the external region to the specific region. A process of copying the immediately preceding binarization result can be used.
[0016]
In this way, by copying the binarization result of the front edge region in the specific region to the front edge region of the external region, the binarization result of the front edge region in the specific region results in an error. It is reflected in the rear edge region in the specific region by error distribution processing in the diffusion method. Therefore, the continuity of the binarization process between the front edge area and the rear edge area in the specific area is maintained, and even when the image is binarized by repeatedly using the formed dither matrix, a pseudo boundary is generated. The generation of lines can be prevented.
[0017]
In this case, the binarization processing further reverses the main scanning direction for a region composed of the rear edge region in the specific region and the front edge region of the external region that are continuous across the boundary. Is preferred. As a result, the binarization result of the front edge area in the specific area is more easily reflected in the rear edge area in the specific area. Therefore, the continuity of the binarization process between the rear edge region in the specific region and the front edge region of the outer region is further strongly maintained, and the generation of the pseudo boundary line is further suppressed.
[0018]
Note that if the specific region does not include the integration result of the first pixel line portion where the error diffusion method has been started, error diffusion or error distribution distortion at the initial stage of error diffusion hardly affects the dither matrix. Thus, even if the image is binarized using the completed dither matrix, it is possible to further suppress the deviation of dots and the generation of patterns. Further, in this case, in order to eliminate the influence of the distortion as much as possible, the specific region is a line farthest from the head pixel line portion, and the integration of the final pixel line portion where the error diffusion method processing is completed is completed. It is preferable to include the results.
[0019]
Further, when binarization processing of a uniform density pixel matrix having a certain density value is performed, each pixel is binarized according to the binarization state of the uniform density pixel matrix that has already been binarized. It is also good. This “binarization processing performed according to the binarization state of the already binarized uniform density pixel matrix” means, for example, binarization processing within the already binarized uniform density pixel matrix There is a method of performing binarization processing according to the binarization state of the uniform density pixel matrix having the density value closest to the uniform density pixel matrix to be attempted. In other words, the binarization process by the error diffusion method of the uniform density pixel matrix is not executed completely independently, but the binarization state in the uniform density pixel matrix of other density values already made is reflected.
[0020]
As a result of this reflection, the integrated state at each pixel position is not biased to a specific value and is dispersed into various values as compared with a case where each pixel pixel matrix is binarized independently. Therefore, in the dither matrix finally obtained based on such an accumulation state, the threshold values are distributed in a moderately distributed manner within a predetermined range.
[0021]
More specifically, in the error diffusion method performed in the uniform density pixel matrix in which the density value i is the density value of all pixels, the density value that is less than the density value i and closest to the density value i is already binarized. If the pixel position that is the same as the pixel position that is turned on in the recently obtained lower binarized pixel matrix is always turned on and binarized by error diffusion processing, the above-described effect can be obtained. In this case, first, the density value of each pixel is turned on / off by the error diffusion method for a uniform density pixel matrix in which the minimum value min of a plurality of discretely set density values is the density value of all pixels. It is also possible to binarize to any of the above, and then binarize the other density as the density value i.
[0022]
Further, in the error diffusion method performed in the uniform density pixel matrix in which the density value i is the density value of all pixels, the density value exceeding the density value i and closest to the density value i is already binarized. In addition, if the pixel position that is the same as the pixel position that is turned off in the upper binarized pixel matrix is always turned off and binarized by error diffusion processing, the above-described effect can be obtained. In this case, first, for a uniform density pixel matrix in which the maximum value max among a plurality of discrete density values is set as the density value of all pixels, the density value of each pixel is turned on / off by the error diffusion method. It is also possible to binarize to any one of the above, and then binarize the other density values as the density value i.
[0023]
Further, in the error diffusion method performed in the uniform density pixel matrix in which the density value i is the density value of all pixels, the density value that is less than the density value i and closest to the density value i has already been binarized. Pixel positions that are also on in the most recently binarized pixel matrix that has already been binarized for the density value that is on and in the lower binarized pixel matrix that exceeds the density value i and is closest to the density value i The same pixel position is always turned on, and the error diffusion processing is performed with the pixel position same as the pixel position that is off in the latest lower binarized pixel matrix and off in the latest upper binarized pixel matrix being off. By binarizing, the above-described effects can be obtained. In this case, first, with respect to two uniform density pixel matrices having the minimum value min and the maximum value max of a plurality of discretely set density values as the density values of all pixels, the error diffusion method is used for each pixel. It is preferable to binarize the density value to either on / off.
[0024]
Further, the density value i is made to correspond to the central integer value of the density values of the nearest lower binarized pixel matrix and the nearest upper binarized pixel matrix, or to almost the middle integer value. Is preferable in terms of improving the resolution and suppressing the pattern of the pseudo halftone image obtained by processing with the obtained dither matrix.
[0025]
The predetermined range may be equal to a density range that can be expressed in a pixel. In this case, examples of the predetermined range include a range of 0 to 255. This is particularly preferable when the data constituting the image to be binarized is represented by 8-bit grayscale.
[0026]
In order to further improve the above-described effect, a plurality of density values discretely set in the predetermined range are set to max-min + 1 set by changing one by one from the minimum value min to the maximum value max. It is preferably a value.
The integration result for each pixel at the same position is, for example, the number of one of the binarized density values at the pixel at the same position, and the threshold value of each pixel of the dither matrix according to this number Can be set. For example, assuming that binary values are represented by “1” (on) and “0” (off), “1” (on) existing at the same position in a plurality of binarized uniform density pixel matrices. The count value may be set as the threshold value at the same position of the dither matrix as it is or by applying a coefficient. Further, the threshold value may be set based on the table from the count value. “0” (off) may be counted instead of “1” (on).
[0027]
In addition to the above-described method, the threshold value is set to 2 when a uniform density pixel matrix having a minimum value min to a uniform density pixel matrix having a maximum value max among a plurality of discretely set density values is viewed. For a pixel position that first appears as “1” (on) in the digitized uniform density pixel matrix, a threshold value is set according to the density value of the uniform density pixel matrix that first becomes “1” (on). It may be set. For example, the pixel positions may be sequentially checked from the smaller density value, and the density value of the uniform density pixel matrix in which “1” (ON) first appears may be set as a threshold value. Alternatively, a threshold value may be set by multiplying the density value by a predetermined coefficient.
[0028]
In addition, the threshold value is set as a binary uniform density when viewed from a uniform density pixel matrix having a maximum value max to a uniform density pixel matrix having a minimum value min among a plurality of density values set discretely. The threshold value may be set according to the density value of the uniform density pixel matrix that first becomes “0” (off) for the pixel position that appears first as “0” (off) in the pixel matrix. good. For example, the pixel positions may be sequentially checked from the larger density value, and the density value of the uniform density pixel matrix in which “0” (off) first appears may be set as a threshold value as it is. Alternatively, a threshold value may be set by multiplying the density value by a predetermined coefficient.
[0029]
DETAILED DESCRIPTION OF THE INVENTION
[Embodiment 1]
FIG. 1 is a main block diagram of the dither matrix creation device 2.
The microcomputer unit 11 constituting the main body of the dither matrix creating apparatus 2 includes a CPU 12, a program memory 13 made of ROM, a working memory 14 made of RAM, a dither matrix storage memory 16 made of RAM, and an output image memory 17 made of RAM. ing. An input unit 10 and an output unit 19 are connected to the microcomputer unit 11 via a system bus 18.
[0030]
The input unit 10 includes an interface for inputting from a keyboard or an external storage device, and inputs data and instructions necessary for creating a dither matrix. In addition, the input unit 10 inputs image data having a gradation in which each pixel is represented by 8 bits in order to confirm the effect of the created dither matrix.
[0031]
The CPU 12 executes creation of a dither matrix as described later, and the program memory 13 stores programs for performing various controls performed by the CPU 12. The working memory 14 temporarily stores data necessary for the CPU 12 to execute a program stored in the program memory 13.
[0032]
The dither matrix storage memory 16 stores dither matrix data formed by the processing of the CPU 12.
The output image memory 17 uses the dither matrix stored in the dither matrix storage memory 16 by the CPU 12 to binarize image data having a gradation represented by 8 bits. Store image data.
[0033]
The output unit 19 is a device that prints the binarized image data stored in the output image memory 17 with or without dots by an electrophotographic method.
Next, a dither matrix creation process executed by the dither matrix creation device 2 will be described. A flowchart of the dither matrix creation process is shown in FIG.
[0034]
First, “0” is set to the density value i (S100). Next, the uniform density pixel matrix in which all pixels have density value i = 0 is binarized (S110). As shown in FIG. 4, 256 uniform density pixel matrices D0 to D255 of i = 0 to 255 may be used by reading the uniform density pixel matrix into the working memory 14.
[0035]
Since all the pixels have the same density value i in each uniform density pixel matrix, in addition to arranging 256 types of uniform density pixel matrices in this way, all density values are used when the density value of each pixel is used for calculation in the error diffusion method. Since it may be handled as the value i, only 256 values from i = 0 to 255 may be stored. Further, a lower limit value “0”, an upper limit value “255” of the density value i, and a step value Sp (processing for determining the value of the density value i discretely present between the lower limit value and the upper limit value are processed. Only a value indicating the interval between the density values i (here, Sp = 1) may be used.
[0036]
The details of the binarization process in step S110 are shown in the flowchart of FIG.
First, the density value (equal to the density value i) of the pixel of interest in the uniform density pixel matrix having the density value i = 0 is read (S112). The pixel position of the first target pixel is the origin position (0, 0). Thereafter, every time the processing returns to step S112, the position of the pixel of interest is designated with the x direction as the main scanning direction and the y direction as the sub scanning direction, and the binarization processing proceeds.
[0037]
Next, it is determined whether or not the pixel position is within a predetermined area (S114). In the first embodiment, the predetermined area is an area that exists immediately after the main scanning direction of the specific area A set in a part of each uniform density pixel matrix D0 to D255 shown in FIG. As shown in the enlarged view of FIG. 6, assuming that the area of the uniform density pixel matrix D0 to D255 other than the specific area A is the external area B, the binarized main scanning direction in the boundary between the specific area A and the external area B Means the front edge region BF of the outer region B that is in contact with the boundary CR crossing from the specific region A to the outer region B.
[0038]
If the target pixel is not within the predetermined region (front edge region BF) (“NO” in S114), then the target pixel is binarized by a normal error diffusion method and the binarization error is Calculated.
That is, first, as shown in the following equation 1, the density value i is corrected by the binarization error sum E of the peripheral pixels, and the corrected density value I is calculated.
[0039]
[Expression 1]

[0040]
The binarization error sum E is calculated based on the coefficient matrix α and the binarization error e of the surrounding pixels as shown in the following equation 2.
[0041]
[Expression 2]

[0042]
In the first embodiment, the following matrix is used as the coefficient matrix α. αab represents the coefficient at the position (a, b) of the coefficient matrix α. Eab represents the binarization error of the peripheral pixel at the position corresponding to the coefficient at the position (a, b) of the coefficient matrix α.
[0043]
[Equation 3]

[0044]
Here, * represents the position of the target pixel.
The corrected density value I obtained in this way is compared with a threshold value t (for example, t = 128 in the first embodiment), and if I> t, the target pixel is binarized to on (“1”). If I ≦ t, the target pixel is binarized to off (“0”).
[0045]
Next, when the target pixel is binarized to ON, the binarization error e of the target pixel is obtained by the calculation of the following equation.
[0046]
[Expression 4]

[0047]
Further, when the pixel of interest is binarized off, I is set to the binarization error e of the pixel of interest as in the following equation.
[0048]
[Equation 5]

[0049]
In this way, the binarization process (S116) by the normal error diffusion method is terminated, and then it is determined whether or not the current target pixel is included in the specific area A (S118). As shown in FIG. 5, the specific area A is an area where binarized value counting processing is performed in order to form the integration result matrix M1. In the first embodiment, all the uniform density pixel matrices D0 to D0 are processed. It exists in the same position in D255.
[0050]
The specific area A does not include the first pixel line L0 portion in the error diffusion process, but includes the final pixel line Lx portion. This is because in the first pixel line L0 portion, there is no previous line, so in the binarization error diffusion, the binarization error distribution is distorted compared to the back line, and the dot concentration and This is because a specific pattern is likely to occur, and the influence of the distortion of the binarized error diffusion decreases as the distance from the head pixel line L0 increases backward, and the influence is the least in the final pixel line Lx portion.
[0051]
If it is determined that the current pixel of interest is included in the specific area A (“YES” in S118), the binarized value of the pixel of interest is stored in the working memory 14 (S120). If it is determined that the current pixel of interest is not included in the specific area A (“NO” in S118), the binarized value of the pixel of interest is not stored.
[0052]
Next, it is determined whether or not there is an unprocessed pixel in the density value i (S122). If there is an unprocessed pixel (“YES” in S122), the process returns to step S112 again, and the unprocessed pixel is selected as the target pixel. The above processing is performed.
If it is determined that the target pixel is within the predetermined area (front edge area BF) (“YES” in S114), the following binarization process is performed (S124).
[0053]
That is, for the current pixel of interest, the calculations of the above formulas 1, 2, 3, and 4 are performed, but the comparison between the corrected density value I for binarization and the threshold value t is not performed, and m pixels in the main scanning direction. The binarization result for the pixel at the returned position is directly adopted as the binarization result for the current pixel of interest. If this binarization result is on, the operation of Expression 3 is performed, and if it is off, the operation of Expression 4 is performed. That is, the error diffusion method according to the binarization result of m pixels before is executed without performing comparison with the threshold value t.
[0054]
The value m is the length of the specific area A in the main scanning direction. Therefore, for the pixels in the front edge region BF, the binarization in the front edge region AF that is in contact with the boundary CF where the binarization process in the main scanning direction crosses from the external region B to the specific region A within the specific region A is performed. The conversion result is adopted as it is. That is, in FIG. 6, the front edge region BF including the three pixel columns Q1, Q2, and Q3 that are in contact with the outside of the right boundary CR of the specific region A has three pixels that are in contact with the inside of the left boundary CF of the specific region A. The binarized state is exactly the same as that of the front edge region AF composed of the columns P1, P2, and P3. For example, for the binarization of the pixel b1, the binarization result of the pixel a1 is used as it is.
[0055]
In this way, each pixel is binarized, and if binarization is completed for all the pixels of the uniform density pixel matrix of density value i (“NO” in S122), as shown in FIG. It is determined whether i = 255 (S130). Since the density value i is initially 0 (“NO” in S130), the density value i is then incremented (S140). Therefore, the above-described binarization process is executed on the uniform density pixel matrix in which all the pixels have the density value i = 1 (S110).
[0056]
Thereafter, the process of sequentially incrementing the density value i (S140) and binarizing the uniform density pixel matrix of the corresponding density value i as described above (S110) is repeated.
When the binarization processing (S110) of the uniform density pixel matrix having the density value i = 255 is completed, 256 binarized pixel matrices having density values i = 0 to 255 are stored in the working memory 14 as shown in FIG. F0 to F255 are formed.
[0057]
Next, since the density value i = 255 (“YES” in S130), for all the binarized pixel matrices, the number of pixels binarized on among the pixels at the same position is counted, and this count value An integration result matrix M1 is formed with S as an element (S150).
[0058]
That is, as shown in FIG. 7, assuming that the upper left corner is the origin (0, 0), the horizontal direction is the x axis, and the vertical direction is the y axis, first, (0, 0) of all the binarized pixel matrices F0 to F255. The pixels that are turned on are counted. The count result is stored in the same position (0, 0) of the integration result matrix M1 prepared in the working memory 14 as shown in FIG. This counting process is performed for each pixel position, and all the integration result matrix M1 is filled. This count process is binarized to “1” when each pixel is on, and to “0” when it is off, so the numerical values of the pixels at the same position may be summed up.
[0059]
Next, threshold values for the dither matrix are set in order from the pixel position having a low count value in the integration result matrix M1 (S160), and the matrix composed of the threshold values is stored in the dither matrix storage memory 16 as a dither matrix (S170). .
[0060]
As a method for setting the threshold value, the count value itself may be set as the threshold value. In this case, since the count process in step S150 corresponds to the threshold setting process for the dither matrix, no particular process is performed in step S160. Then, the integration result matrix M1 itself is stored as a dither matrix in the dither matrix storage memory 16 (S170).
[0061]
In step S160, instead of simply using the count values themselves of the integration result matrix M1 as threshold values, the count values of the integration result matrix M1 are set as threshold values according to a table of count values and threshold values prepared in advance. It may be converted.
In step S160, the threshold values may be set to “0, 1, 2,...” From the lowest count value. 254, 253, ... "may be set. In this case, if there are two or more of the same count value, they may all be given the same threshold value, or the same count value may be ordered and the threshold value set in the same way as different count values. You may do it.
[0062]
Using the dither matrix formed in this way, a halftone color image is binarized and recorded by a color printer, so that it is not noisy to human eyes and resolution is not deteriorated. Thus, it was possible to generate a pseudo-halftone binarized image in which the generation of patterns was suppressed. Of course, this method is a dither method, and many calculations are not performed unlike the error diffusion method, so that it can be processed quickly.
[0063]
Furthermore, in the first embodiment, the binarized pixel matrix F0 to F255 of the specific area A that does not include the first pixel line L0 and includes the final pixel line Lx is obtained from the uniform density pixel matrix, and this binarized pixel. An integration result matrix M1 smaller than the uniform density pixel matrix is obtained from the matrices F0 to F255, and a dither matrix is formed based on the integration result matrix M1. For this reason, it is difficult to be affected by the distortion of error distribution occurring in the first pixel line L0, and a pseudo-halftone binary image in which the generation of patterns is further suppressed can be generated.
[0064]
Further, as described above, the binarization result of the front edge area AF of the specific area A is directly adopted for the pixels in the front edge area BF. Therefore, in the error distribution, a binarization error is distributed from the peripheral pixel at the position corresponding to the shape of the coefficient matrix α to the target pixel, but the rear edge region AR of the specific region A that is in contact with the left side of the boundary CR. However, the binarization error distribution from the pixels in the front edge region BF is directly or indirectly affected. Since the front edge area BF is the same as the binarization result of the front edge area AF of the specific area A, the front edge area AF of the specific area A and the rear edge area AR of the specific area A The continuity of the process is maintained. Therefore, even if the halftone image is binarized by using the dither matrix created by the following process, it is possible to suppress color unevenness and light / dark unevenness between the areas processed by the dither matrix, and the pseudo boundary Can be suppressed.
[0065]
The determination of unprocessed pixels in step S122 may be determination of whether or not there are unprocessed pixels in the specific area A.
[Embodiment 2]
The second embodiment is different from the first embodiment in that the binarization process in step S110 is as shown in FIGS. 9 and 10, and the error diffusion method performed in steps S1160, S2040, and S2050. The difference is that, when the pixel of interest is binarized, the binarization error is distributed to the density of surrounding pixels that have not yet been binarized.
[0066]
In FIG. 9, steps S1120, S1180, S1200, and S1220 are the same as steps S112, S118, S120, and S122 shown in FIG. 3 of the first embodiment. In the following, a description will be given focusing on differences from the first embodiment.
[0067]
First, after step S1120, it is determined whether or not it is the reverse rotation region RV (S1122). If it is the reverse rotation region RV (“YES” in S1122), reverse binarization processing (S2000) is performed.
As shown in FIG. 11, the reverse region RV includes the pixel row Pm-3 immediately before the rear edge region AR of the specific region A and the entire rear edge region AR of the specific region A (pixel rows Pm-2, Pm- 1, Pm) and the entire front edge region BF of the outer region B (pixel columns Q1, Q2, Q3).
[0068]
Therefore, while the binarization process performed in the main scanning direction does not reach the pixel column Pm-3 (“NO” in S1122), the processes of steps S1160 to S1220 are executed, and the target pixel is detected by a normal error diffusion method. While binarization is performed, a binarization error is calculated and distributed to surrounding pixels that have not yet been binarized.
[0069]
That is, as shown in the following expression 5, the density value i is corrected by the binarization error sum E distributed from the already binarized peripheral pixels to the target pixel, and the corrected density value I is calculated.
[0070]
[Formula 6]

[0071]
The corrected density value I obtained in this way is compared with a threshold value t (for example, t = 128 in the second embodiment), and if I> t, the target pixel is binarized to ON (“1”). If I ≦ t, the target pixel is binarized to off (“0”).
Next, when the pixel of interest is binarized to ON, the binarization error e of the pixel of interest is obtained by the calculation of the following equation (6).
[0072]
[Expression 7]

[0073]
Further, when the pixel of interest is binarized off, I is set to the binarization error e of the pixel of interest as shown in the following equation (7).
[0074]
[Equation 8]

[0075]
Next, the binarization error e is distributed to the unbinarized peripheral pixels based on the coefficient matrix β shown below.
[0076]
[Equation 9]

[0077]
Here, * represents the position of the target pixel, and the numerical value represents that the binarization error e obtained by multiplying the numerical value by the coefficient is distributed to the target pixel at the position corresponding to the numerical value. The processing of the next steps S1180 to S1220 is the same as steps S118 to S122 of the first embodiment.
[0078]
When the binarization processing performed in the main scanning direction reaches the pixel column Pm-3, that is, when the reverse rotation region RV is entered (“YES” in S1122), the processing in step S2000 is executed.
As shown in the flowchart of FIG. 10, first, a binarization target pixel is set (S2010). For this setting, a pixel row consisting of “..., A, b, c, d, e,..., R, S, T, U, V, W, X, Y, Z,. To explain, when the pixel S is set as the target pixel in the loop returning from step S1220 to step S1120, a process of setting the pixel Y as the actual pixel to be binarized is performed separately from the setting. . Next, when the pixel T is set as the target pixel, the pixel X is set as the binarization target pixel, and when the pixel U is set as the target pixel, the pixel W is set as the binarization target pixel. When the pixel V is set as the target pixel, the pixel V is set as the binarization processing target pixel, and when the pixel W is set as the target pixel, the pixel U is set as the binarization processing target pixel. When the pixel X is set, the pixel T is set as the binarization processing target pixel, and when the pixel Y is set as the target pixel, the pixel S is set as the binarization processing target pixel. That is, in the reverse rotation region RV, the binarization process proceeds in the opposite direction to the other regions.
[0079]
Next, the density value of the binarization target pixel and the binarization error sum E are newly read (S2015). Then, for the binarization process by the error diffusion method, a coefficient matrix corresponding to the binarization process target pixel position is designated (S2020). The reason why the coefficient matrix β used outside the reverse region RV is not used is that the arrangement of the non-binarized peripheral pixels with respect to the binarization target pixel is different because the main scanning direction is reversed.
[0080]
For the seven pixel columns Pm-3, Pm-2, Pm-1, Pm, Q1, Q2, and Q3 in the reverse region RV, the coefficient matrix is adapted to the arrangement of the unbinarized peripheral pixels, respectively. A coefficient matrix β1, β2, β3, β4, β5 is selected and applied. An example of each coefficient matrix β1, β2, β3, β4, β5 is shown below.
[0081]
[Expression 10]

[0082]
[Expression 11]

[0083]
[Expression 12]

[0084]
[Formula 13]

[0085]
[Expression 14]

[0086]
A coefficient matrix β5 is applied to the pixel column Pm-3 in the reverse region RV, a coefficient matrix β4 is applied to the pixel column Pm-2, and a coefficient matrix β3 is applied to the pixel columns Pm-1, Pm, and Q1. The coefficient matrix β2 is applied to the pixel column Q2, and the coefficient matrix β1 is applied to the pixel column Q3.
[0087]
Next, step S2040 or step S2050 is executed in the determination of step S2030. The processing in step S2040 is the same as the processing in step S1160 described above except that the coefficient matrix specified in step S2020 is used as the coefficient matrix.
[0088]
That is, for the current pixel of interest, the calculation of the above-mentioned formulas 5, 6, and 7 and the distribution of the binarization error e are performed, but the comparison between the correction density value I for binarization and the threshold value t is not performed. The binarization result for the pixel at the position returned by m pixels in the main scanning direction is directly adopted as the binarization result of the current pixel of interest. If this binarization result is on, the calculation of Expression 6 is performed, and if it is off, the calculation of Expression 7 is performed. That is, without making a comparison with the threshold value t, a binarization error is calculated according to the binarization result of m pixels before, and distributed to surrounding pixels based on the coefficient matrix β.
[0089]
The value m is the length of the specific area A in the main scanning direction. Therefore, for the pixels in the front edge region BF, the binarization in the front edge region AF that is in contact with the boundary CF where the binarization process in the main scanning direction crosses from the external region B to the specific region A within the specific region A is performed. The conversion result is adopted as it is. That is, in FIG. 6, the front edge region BF including the three pixel columns Q1, Q2, and Q3 that are in contact with the outside of the right boundary CR of the specific region A has three pixels that are in contact with the inside of the left boundary CF of the specific region A. The binarized state is exactly the same as that of the front edge region AF composed of the columns P1, P2, and P3. For example, the binarization of the pixel W uses the binarization result of the pixel a as it is.
[0090]
That is, after performing the binarization processing of the pixel R, the pixel Y is binarized next, but the pixel Y is in a predetermined region (front edge region BF) (“YES” in S2030). The binarization itself is set to the same value as the pixel c for which the binarization result has already been obtained, and the binarization error is distributed by the coefficient matrix β1. Next, the pixel X is binarized. Here, the binarization itself is set to the same value as the pixel b for which the binarization result has already been obtained, and the binarization error is distributed by the coefficient matrix β2. . Next, the pixel W is binarized. Here, the binarization itself is set to the same value as the pixel a for which the binarization result has already been obtained, and the binarization error is distributed by the coefficient matrix β3. .
Next, the pixel V is binarized. However, since the pixel V is not in the predetermined region (front edge region BF) (“NO” in S2030), the pixel V is binarized by comparison with the threshold value t. Is distributed by the coefficient matrix β3. The processing for the next pixel U is the same as that for the pixel V. Next, the pixel T is binarized. Again, the pixel T is binarized by comparison with the threshold value t, and the binarization error is distributed by the coefficient matrix β4. Next, the pixel S is binarized. Here, too, the pixel S is binarized by comparison with the threshold value t, and the binarization error is distributed by the coefficient matrix β5.
[0091]
Then, since the target pixel goes out of the reverse rotation region RV and becomes the pixel Z (“NO” in S1122), the pixel Z itself is binarized, and is left to the process of step S1160.
Since the second embodiment is configured as described above, the same effect as that of the first embodiment is produced, and the binarization main scanning direction is reversed in the reverse rotation region RV across the boundary CR. Therefore, as compared with the first embodiment, the binarization is performed on the rear edge region AR of the specific region A in contact with the left side of the boundary CR from the pixel in the front edge region BF in contact with the right side of the boundary CR. The influence of error distribution can be made sufficiently large. That is, the binarization error distribution from the front edge area AF to the rear edge area AR is sufficiently effective. For this reason, in the binarization processing by the error diffusion method, the continuity of processing between the front edge area AF of the specific area A and the rear edge area AR of the specific area A is sufficiently maintained. Even if the halftone image is binarized by using the dither matrix created by the processing of No. 2, it is possible to sufficiently suppress the generation of a pseudo boundary between processing regions by the dither matrix.
[0092]
[Embodiment 3]
The third embodiment is different from the first and second embodiments in the dither matrix creation processing, and is that the binarization processing of each pixel of the uniform density pixel matrix is performed as follows. That is, ON and density in the most recent lower binarized pixel matrix L obtained by binarizing a uniform density pixel matrix having a density value less than the density value i of the uniform density pixel matrix to be binarized and closest to the density value i. In the most recently binarized pixel matrix H obtained by binarizing the uniform density pixel matrix having a density value exceeding the value i and closest to the density value i, the same pixel position as the on pixel position is always turned on. Then, error diffusion processing is performed under the condition that the pixel position that is off in the latest lower binarized pixel matrix and the same pixel position that is off in the latest upper binarized pixel matrix is always turned off. Furthermore, the binarization processing order of each uniform density pixel matrix is such that the density value i of the uniform density pixel matrix to be processed is the center of the density values of the most recently lower binarization pixel matrix and the most recent binarization pixel matrix. Another difference is that processing is sequentially performed so as to correspond to an integer value.
[0093]
The dither matrix creation process of the third embodiment is shown in FIGS.
First, a uniform density pixel matrix in which all pixels have a density value i = 0 and a uniform density pixel matrix in which all pixels have a density value i = 255 are binarized by an error diffusion method (S3080). As shown in FIG. 4, this uniform density pixel matrix may be used by reading two types of uniform density pixel matrices D0 and D255 of i = 0 and 255 into the working memory 14. Then, the binarized pixel matrices F0 and F255 (FIG. 7) are stored in the working memory 14 (S3090). However, the binarization result by the error diffusion method of the uniform density pixel matrix having the density values i = 0 and 255 performed in step S3080 indicates that all pixels are “0” (off) when i = 0, and i = 255. In this case, since all the pixels are “1” (on), a binary pixel matrix in which all the pixels are off and a binary pixel matrix in which all the pixels are on are provided in advance without performing calculation by the error diffusion method. In addition, these matrices may be used.
[0094]
Next, a density value array is created for the density values 1 to 254 (S3100), and then the density values i are read one by one from the array (S3105), and the process of step S3110 is performed to obtain all the density values in the array. If the process is completed for the value, it is determined in step S3130 that the array is completed, and the processes of steps S3150, S3160, and S3170 are performed. The processes in steps S3150, S3160, and S3170 are the same as steps S150, S160, and S170 in the first embodiment.
[0095]
The array of density values in step S3100 is created so that the density value at the center of the binarized density values is set to i. That is, the beginning of the array is “128” which is the central integer value of “0” and “255” (may be another central integer value “127”), and the next is “0” and “128”. "64" in the middle of "128", "192" in the middle of "128" and "255" (another integer value of "191" may be acceptable), and the next is "0" and "64" Central “32”, “64” and “128” in the middle “96”, “128” and “192” in the middle “160”, “192” and “255” in the middle “224” (Another central integer value “223” may be used.) As described above, the central integer value of the previously obtained values is obtained, and an array of density values from “1” to “254” is obtained. “128, 64, 192, 32, 96, 160, 224,...” Are created. This array may be calculated in advance and stored as array data, and only the array data may be read and used during this processing.
[0096]
In step S3105, the density value i is sequentially set from the beginning of this array, the binarization process of step S3110 is performed, and if the process of step S3110 is completed until the end of the array, “YES” is determined in step S3130. After the determination, the process proceeds to threshold value setting processing (S3150, S3160, S3170).
[0097]
Next, the binarization process in step S3110 will be described. In this process, the binarization process of the first embodiment is the same as the process described in FIG. 3 except that the process in step S116 is replaced with the process shown in FIG.
That is, if it is determined in step S114 that it is not within the predetermined area, the process of step S4200 is performed, and binarized density values, that is, among the density values for which a binarized pixel matrix has already been obtained. Thus, the latest upper binarized pixel matrix H and the latest lower binarized pixel matrix L are searched from the contents of the working memory 14 (S4200).
[0098]
Next, the uniform density pixel matrix having the density value i is binarized by the error diffusion method to create a binarized pixel matrix Fi (S4210).
However, it is binarized under the following conditions (1) and (2).
{Circle around (1)} A pixel position where “1” ON is arranged in both the most recently binarized pixel matrix H and the most recently binarized pixel matrix L is always “1” ON.
[0099]
{Circle around (2)} Pixel positions where “0” off is arranged in both the most recently binarized pixel matrix H and the most recently binarized pixel matrix L are always “0” off.
That is, normally, in the error diffusion method, the sum of the error distributed from the surrounding pixels and the density of the pixel of interest is compared with a threshold value. For example, in the case of a pixel corresponding to the above condition (1) or (2), the pixel is not compared with the threshold value without being compared with the above (1) or (2). Set “1” on or “0” off according to condition (2). Of course, the result of setting that is not compared with this threshold value is reflected in the binarization error, and is subject to distribution to surrounding pixels.
[0100]
According to the conditions (1) and (2), as shown in FIG. 14 (a), when viewed from the smaller density value at each pixel position, once turned on, the subsequent density value i is always turned on. Become.
When normal binarization is performed without using the above conditions (1) and (2), as shown in FIG. 14B, after turning on, off appears at the same pixel position. For this reason, the number of ONs at each pixel position tends to be biased to a specific value in the range of 0 to 255 with only normal binarization, but in the third embodiment, various values in the range of 0 to 255 are included. To a suitable distribution. Therefore, the dither matrix obtained in this way is distributed in a moderately distributed manner in the range of the threshold value from 0 to 255. Therefore, in particular, as shown in FIG. 14B, the dither with improved resolution and suppression of the pattern is further improved as compared with the dither matrix obtained when each uniform density pixel matrix is binarized independently. A matrix can be obtained.
[0101]
Further, in the third embodiment, in order to sequentially binarize the density i at the center of the already binarized density values, the latest upper binarized pixel matrix and the latest lower binarized pixel matrix The binary dispersion of the binarized pixel matrix having the central density i has a relatively uniform distribution. Since the dither matrix is created based on the binarized pixel matrix satisfying the uniform distribution and the conditions (1) and (2) obtained in this way, the effect of improving the resolution and suppressing the pattern further increases. To do.
[0102]
In the third embodiment, the array is created in advance so that the center value can be obtained sequentially for the density value i. However, instead of creating the array in advance, the center value is calculated in the process of step S3105. May be sequentially obtained and set to the density value i.
An example of obtaining the central value by calculation is shown in FIG. This process is executed instead of steps S3100 to S3130 in FIG. In this process, first, eight variables I0 to I7 are initialized with "0" (S5010 to S5080), then I0 to I7 are summed and set to a density value i (S5090). If the density value i is “0”, I7 is increased by “128” (S5120), and then I7> 128 is not satisfied (“NO” in S5130), so the density value i is set to “ 128 ”is set, i is not 0 (“ NO ”in S5100), and i is not 255 (“ NO ”in S5110), so the processing in step S3110 (same as the processing in FIG. 12) is performed.
[0103]
Next, in step S5120, I7 = 256, it is determined "YES" in step S5130, I6 is increased by "64" (S5140), I6 = 64, and I6> 64 is not satisfied (in S5150). “NO”), it is initialized to I7 = 0 in step S5080, and i = 64 in step S5090. Accordingly, the process proceeds from steps S5100 and S5110 to step S3110, and binarization processing is performed on the uniform density pixel matrix having density value i = 64.
[0104]
Next, in step S5120, I7 = 128, and "NO" is determined in step S5130. In step S5090, I0 to I5 = 0, I6 = 64, and I7 = 128, i = 192. After steps S5100 and S5110, binarization is performed on the uniform density pixel matrix having density value i = 192.
[0105]
Next, since I0 to I5 = 0, I6 = 64, and I7 = 128, the process proceeds from step S5120 to step S5160, and in step S5160, I5 = 32. Since I5> 32 is not satisfied (“NO” in S5170), steps S5070 and S5080 are processed, and I0 to I4 = 0, I5 = 32, I6 = 0, and I7 = 0. Accordingly, i = 32 in step S5090, and the uniform density pixel matrix having the density value i = 32 is binarized in step S3110.
[0106]
Next, since I7 = 0, the process returns to step S5090 via steps S5120 and S5130. At this time, I0 to I4 = 0, I5 = 32, I6 = 0, and I7 = 128. Accordingly, i = 160 in step S5090, and the uniform density pixel matrix having the density value i = 160 is binarized in step S3110.
[0107]
Next, steps S5120 to S5150 and S5080 are processed so that I0 to I4 = 0, I5 = 32, I6 = 64, and I7 = 0. Therefore, i = 96 is obtained in step S5090, and the uniform density pixel matrix having the density value i = 96 is binarized in step S3110.
[0108]
Thereafter, 224, 16, 144, 80,..., 127 are sequentially set to the density value i by any one of steps S5020 to S5080 and S5120 to S5260 according to the values of I0 to I7. The binarization and storage of the uniform density pixel matrix of value i is performed, and finally I0 = 1, I1 = 2, I2 = 4, I3 = 8, I4 = 16, I5 = 32, I6 = 64, I7 = 128. At this time, since the density value i = 255 is set in step S5090, “YES” is determined in step S5110, and the process proceeds to step S3150. In this manner, the uniform density pixel matrix can be binarized by following the central value in the calculation.
[0109]
In the second embodiment, binarization processing is performed by following the central integer value. However, even if it is not completely the central integer value, the 1: 2 position, 1: 3 position, 2: 1 Although it is an intermediate value between two binarized density values, such as a position and a 3: 1 position, binarization processing may be performed by following an integer value biased to the small side or the large side.
[0110]
[Others]
In the first to third embodiments, the count value is not directly used as the integration result matrix M1, but may be multiplied by a predetermined coefficient. Also, a different coefficient k (i) is set for each density value i for the binarized value Iixy of the uniform density pixel matrix of density value i as shown in the following formula, and the total value Cxy of all uniform density pixel matrices is set. The accumulation result matrix M1 may be created by obtaining the value.
[0111]
[Expression 15]

[0112]
Note that x and y represent pixel positions on the uniform density pixel matrix or the integration result matrix M1.
The binarization of the uniform density pixel matrix was binarization to ON (“1”) or OFF (“0”), but any value may be binarized. Counting may be performed to form an accumulation result matrix M1, and the accumulation result matrix M1 may be converted into a dither matrix. Further, the total of binarized values may be used instead of the number of counts as described above.
[0113]
In the first to third embodiments, the specific area A for storing the binarized values is the same position in all the uniform density pixel matrices, but may be different depending on the uniform density pixel matrix. This is because the degree of the influence of the error diffusion distortion occurring in the first pixel line L0 portion on the rear side differs depending on the density value i constituting the uniform density pixel matrix.
[0114]
Further, since the degree of influence of error diffusion distortion occurring in the first pixel line L0 portion in the backward direction varies depending on the density value i in this way, error diffusion distortion occurs in binarization of all uniform density pixel matrices. In order to eliminate the influence of the above, the size of the uniform density pixel matrix may be adjusted according to the density value i.
[0115]
In the first to third embodiments, the length of the predetermined region (front edge region BF) in the main scanning direction is three pixels, but this is preferably a length corresponding to the number of columns of the coefficient matrix. . That is, it is preferable to set the length to have an effect of error diffusion.
[0116]
The binarization in the first embodiment is an error diffusion method in which binarization error is distributed from surrounding pixels at the time of binarization, a so-called average error minimum method. When the pixels are binarized, an error diffusion method may be used in which binarization errors are distributed to the density of surrounding pixels that have not yet been binarized.
[0117]
Conversely, the average error minimum method may be employed in the second embodiment. In this case, the coefficient matrices β, β1, β2, β3, β4, β5 have the following contents.
[0118]
[Expression 16]

[0119]
[Expression 17]

[0120]
[Expression 18]

[0121]
[Equation 19]

[0122]
[Expression 20]

[0123]
[Expression 21]

[0124]
The binarization in the third embodiment may be an error diffusion method of receiving a binarization error distribution from surrounding pixels at the time of binarization, that is, a so-called average error minimum method, as in the first embodiment. As in 2, when the pixel of interest is binarized, an error diffusion method based on a method of distributing binarization errors to the density of surrounding pixels that have not yet been binarized may be used.
[Brief description of the drawings]
FIG. 1 is a main block diagram of a dither matrix creating apparatus according to a first embodiment.
FIG. 2 is a flowchart of a dither matrix creation process according to the first embodiment.
FIG. 3 is a flowchart of binarization processing according to the first embodiment.
4 is an explanatory diagram illustrating uniform density value image data according to Embodiment 1. FIG.
FIG. 5 is an explanatory diagram illustrating a relationship between a uniform density value image and a specific region according to the first embodiment;
FIG. 6 is an explanatory diagram showing an enlargement of a specific region and its periphery.
7 is an explanatory diagram showing a state in which uniform density value image data according to the first embodiment is binarized. FIG.
FIG. 8 is an explanatory diagram of an integration result matrix according to the first embodiment.
FIG. 9 is a flowchart of binarization processing according to the second embodiment.
10 is a flowchart of backward binarization processing according to Embodiment 2. FIG.
FIG. 11 is an explanatory diagram showing an enlargement of a specific area and its periphery.
FIG. 12 is a flowchart of a dither matrix creation process according to the third embodiment.
13 is a flowchart showing a part of binarization processing according to Embodiment 3. FIG.
FIG. 14 is an explanatory diagram of an on / off setting state of each pixel position according to the third embodiment.
FIG. 15 is a flowchart illustrating another example of the method for determining the processing order of the uniform density pixel matrix according to the third embodiment.
FIG. 16 is an explanatory diagram showing a conventional example.
[Explanation of symbols]
2 ... Dither matrix creation device 10 ... Input section
11 ... Microcomputer part 12 ... CPU
13 ... Program memory 14 ... Working memory
16 ... Dither matrix storage memory
17 ... Output image memory 18 ... System bus
19 ... Output section

Claims (8)

The uniform density pixel is obtained by setting each of a plurality of density values discretely set in a predetermined range set in a density range that can be expressed in the pixel as a density value of all pixels in each of the plurality of uniform density pixel matrices. For each matrix, the density value is binarized to one of ON / OFF by an error diffusion method, and each dither matrix in the dither matrix is based on the density value of a specific area of a part of the binarized uniform density pixel matrix. A dither matrix creation method for setting pixel thresholds,
The binarization that sandwiches the boundary with respect to an edge region in the specific region that touches a boundary between the specific region and an external region other than the specific region, or an edge region of the external region that touches the boundary A dither matrix creation method characterized by performing boundary processing to maintain continuity of processing. The density values of a specific region of a part of the binarized uniform density pixel matrix are accumulated for each pixel at the same position for all of the uniform density pixel matrix, and based on the accumulation result for each pixel at the same position. 2. The method of creating a dither matrix according to claim 1, wherein a threshold value of each pixel in the dither matrix is set. 3. The dither matrix generating method according to claim 1, wherein the specific region does not include a result of integration of a head pixel line portion where the error diffusion method has been started. The dither matrix creating method according to claim 1, wherein the specific region includes a result of integration of a final pixel line portion where the error diffusion processing is completed. The boundary processing is performed on the rear edge region in the specific region or the front edge region of the external region where the processing in the main scanning direction of binarization touches the boundary crossing from the specific region to the external region. The dither matrix creating method according to claim 1, wherein the dither matrix is created. The boundary processing is performed immediately before the front edge region in the specific region, in the front edge region of the external region, in the front edge region in the specific region where the processing in the main scanning direction of binarization touches the boundary crossing from the external region to the specific region. 6. The dither matrix creating method according to claim 5, wherein the dither matrix creating process is a process of copying the value result. The binarization process reverses the main scanning direction for a region composed of a rear edge region in the specific region and a front edge region of the external region that are continuous across the boundary. Item 7. The method of creating a dither matrix according to item 6. The dither matrix creating method according to claim 1, wherein the predetermined range is equal to a density range that can be expressed in the pixel.

Images