JP4141824B2 - Dot dispersion type mask, image display device, image processing device and method, image processing program, dot dispersion type mask creation method, dot dispersion type mask creation program, and computer-readable recording medium - Google Patents

Description

【００７０】
しかし、入力側から出力側、すなわち一次マトリクス側から出力マトリクス側へマッピングを行うと、回転がすべて終了したときに出力マトリクスの中には閾値が入っていない部分が現れてしまう。このため、出力側を基準として、逆変換によって座標を求める手法を採る。この逆変換の変換式は、次の式（２）のようになる。また、逆変換の様子は、図１１に示すようになる。
【００７１】
【数２】

【００７２】
上記のようにして変換先の座標を求めた際、出力マトリクス側から一次マトリクス側への変換後の点が同じ位置にマッピングされることが生じ得る。そこで、本実施の形態では、一次マトリクスの閾値は、一度しか選択できないこととし、既に選択されていた場合は、その位置の近傍からまだ選択されていない閾値を探索する。
【００７３】
（３）変換後の点が同じ位置にマッピングされた場合は近傍の閾値を探索する処理は、次のように行われる。すなわち、変換後の点が同じ位置にマッピングされた際に、その位置の近傍からまだ選択されていない閾値は、図１２に示すように、最初に８近傍の中からまだ選択されていない閾値を探索する。次に、該当したものの中からユークリッド距離が最小となる閾値を選択する（ただし、ここでは式（２）で求めた際の浮動小数点型の座標との距離をいう）。仮に、８近傍で見つからなかった場合には、１６近傍、２４近傍と範囲を広げて同様に探索を行う。
【００７４】
上記方法によりマトリクスを作成すれば、基礎となるディザマトリクスの規則的成分は、回転した形で規則性を保ったまま残すことができる。また、回転させた際に閾値の抜け、および重複が生じないように、探索して閾値を敷き詰めることにより、不規則的な成分を付加することができる。
【００７５】
これらの規則的成分と不規則的成分の強度を比較すれば、基礎となるディザマトリクスが規則的であって、その成分の多くが回転した形で保持されるため、規則的成分の強度が不規則的成分の強度よりも強くなる。これら成分の強度の大小は、先に図３〜図６を用いて説明したように、周波数領域での振幅もしくは振幅の２乗のエネルギーの大小として表現することができる。
【００７６】
次に、回転変換における回転の角度毎に上記の探索回数を調査した結果について説明する。図１３に示すように、０度と９０度付近で探索回数が少なく、同じ位置にマッピングされることが少ないことが確認できる。
【００７７】
また、上記のように回転行列を用いた方法によって、ドットパターンが分散したディザマトリクス（ドット分散型マスク）が得られる。ここでは、上記のように、入力ディザマトリクスにＢａｙｅｒ型（４×４）を使用し、４×４と１３２×１６２（２インチのＬＣＤパネルの代表的なサイズ）それぞれのサイズでディザマトリクスを作成した。閾値が０のドットだけをＯＮにしたドットパターンを図１４に示す。図１４（ｂ）、（ｃ）は、それぞれ４×４のサイズで回転角が１０度、３０度のドットパターンを示す。同図（ｄ）、（ｅ）は１３２×１６２のサイズで回転角が１０度、３０度のドットパターンを示す。同図（ａ）は回転角が０°すなわちＢａｙｅｒ型（４×４）をそのまま使用して得られるドットパターンを示す。ただし、いずれのパターンについても閾値が０のドットをタイル状に敷き詰めて１３２×１６２にしたものを示す。
【００７８】
図１４（ｂ）、（ｃ）からわかるように、４×４のサイズで回転した場合は、所定の回転角を指定して回転操作を行っても、ドットパターンはほとんど回転しない。これは、入力ディザマトリクスにＢａｙｅｒ型の４×４を使用したときには、回転操作を有効にするために、ある程度の広さの敷き詰め範囲が必要であることを表す。ここで敷き詰め範囲とは、図１０のフローチャートの説明における出力マトリクスのサイズのことを指す。本発明者は、基礎となる入力ディザマトリクスのサイズとして４×４のサイズを用いる場合には、出力マトリクスのサイズとしておよそ１６×１６以上を用いれば、回転操作が有効にはたらくことを経験的に確認している。また、１６×１６以上であればそれ以上のどのサイズに関しても、図１４（ｄ）、（ｅ）に示した１３２×１６２のパターンとほぼ同様なパターンを発生させることが可能であることも確認している。
【００７９】
また、回転角の違いによるドットパターンの特徴をみると、１３２×１６２の結果から、回転角が３０度の方が１０度のときよりも乱れたドットパターンを示した。このことは、３０度の方が、点が同じ位置にマッピングされることが多いからであり、図１３において、３０度において探索回数が多いことから確認することができる。
【００８０】
次に、最良な回転角を決定するために、官能評価実験を行った。これは、ドットパターンを観察することにより行う実験である。具体的方法は、０度から４５度の範囲で回転角の異なるディザマトリクスを作成し、得られたディザマトリクスによる画質を、一対比較法により評価を行い、最良な回転角を決定した。評価結果は、図１５に示すように、回転角が１０度のとき最も心理評価値が高くなった。図１５における縦軸の心理評価値は、値が大きいほど他に比べて好ましい画質であることを表している。すなわち、回転範囲を０度から９０度として考えると、１０度または８０度が望ましい角度である。このため、本実施の形態では、回転角を５度から２０度または７０度から８５度の範囲内で回転することとしている。さらに好ましくは、１０度または８０度で回転する。上記のように、本発明者は、５度から２０度または７０度から８５度の範囲内で回転することによって、画質に対する心理表価値が最も高くなることを見出している。これにより、画質の向上を図ることができるマスクの作成が可能となる。
【００８１】
（第２の作成方法）
次に、第１の実施形態に係るディザ閾値マスクの第２の作成方法について説明する。上記の第１の作成方法では、入力マトリクスとしてＢａｙｅｒ型のディザマトリクスを使用し、これを回転処理することにより規則性と不規則性を併せ持つディザ閾値マトリクスを作成した。回転処理は回転行列を用いて入力マトリクスの閾値を変換することにより行った。
【００８２】
これに対し、第２の作成方法は、入力マトリクスとしてＢａｙｅｒ型のディザマトリクスを用いること、及び、回転処理により規則性と不規則性を併せ持つディザ閾値マトリクスを作成することは第１の作成方法と同様であるが、回転処理として、回転行列を用いるのではなく、ディジタル線分及び境界追跡のアルゴリズムを使用する点が第１の作成方法とは異なる。
【００８３】
図１９に、ディザ閾値マスクの第２の作成方法のフローチャートを示す。ここでも、ディザ閾値マスクをマトリクスと呼ぶ。まず、入力マトリクスとしてＢａｙｅｒ型のディザマトリクスを読み込み（ステップＳ１１）、所定サイズの一次マトリクス及び出力マトリクスの領域を確保する（ステップＳ１２）。一次マトリクスは、第１の作成方法と同様に、マスク作成作業のために使用される作業領域と考えることができる。図２０を参照すると、図２０（ａ）に模式的に示すＢａｙｅｒ型などのディザマトリクスが用意される。そして、図２０（ｂ）に示すように、ある座標系（ここではＸＹ座標系）上に所定サイズの一次マトリクスの領域を確保し、そこに複数個のディザマトリクスを敷き詰めることにより所定サイズの一次マトリクスの領域を形成する。また、出力マトリクスのサイズに相当する記憶領域をメモリ上などに確保する。
【００８４】
次に、回転処理の回転角をθとすると、図２０（ｃ）に示すように、一次マトリクスの座標系内に設定された原点Ｏを通り、当該座標系上の１つの軸（例えばＸ軸とする）に対して角度θを有する線分（回転後基準線分７２）を引き、一次マトリクス内でその線分上に存在する各点の閾値を出力マトリクス内の閾値として保存する（ステップＳ１３）。これにより、角度θの回転処理により得られる出力マトリクス内の回転後基準線分７２上の閾値が得られたことになる。
【００８５】
次に、こうして得られた出力マトリクス内の回転後基準線分７２を基準として、境界追跡処理により出力マトリクス内の他の閾値を求めていく。具体的には、まず、境界追跡のスタート位置を決定してその閾値を求め（ステップＳ１４）、そのスタート位置を開始点として、ステップＳ１３で求めた回転後基準線分に沿って閾値を入力マトリクスから取得し、出力マトリクスの閾値として保存していく（ステップＳ１５）。
【００８６】
こうして、最初に得られた回転後基準線分を基準として、その境界に沿って順に閾値を決定していくことにより、所定サイズの出力マトリクスの閾値が順に決定されていく。そして、出力マトリクスの閾値が全て決定され、保存された時点で（ステップＳ１６；Yes）、処理は終了する。
【００８７】
図２１に、ステップＳ１４で行われる境界追跡処理の例を示す。境界追跡処理とは、既に得られた線分に沿って、当該線分に隣接する境界点を決定し、その境界点に閾値を決定してゆく処理である。図２１（ａ）には傾きが異なる２つの線分（線分９０と線分９１）が示されている。いま、ステップＳ１３において回転後基準線分７２として線分９０が得られたとする。図２１（ａ）においては、線分に隣接して既に決定された境界点と、それら境界点に基づいて決定された境界候補点とが示されている。右上がりの線分の場合、図２１（ｂ）に示す境界候補マトリクスを用い、これを１つの境界点に当てはめることにより、その回りに境界候補点が得られる。図２１（ｂ）に例示する境界候補マトリクスは、左下に１つの境界点を有し、その他の３つの隣接点を境界候補点とするマトリクスである。
【００８８】
境界候補マトリクスを用いて得られた境界候補点のうち、線分自体を構成する点ではなく、かつ、線分に隣接する点を境界点と決定し、一次マトリクス上の当該点に対応する閾値を出力マトリクス内の回転後基準線分７２に隣接する閾値として保存する。回転後基準線分７２に沿って境界点が決定され、その閾値を保存することにより、回転後基準線分７２に沿って１行の閾値が保存されることになる。
【００８９】
次に、そうして決定された境界点が構成する線分に対してさらに同様の境界追跡処理を行うことにより、回転後基準線分７２から順に１行ずつ出力マトリクス内の閾値が決定されていき、やがて出力マトリクス内の全ての閾値が保存されて出力マトリクスが完成する。図２１（ｃ）は、回転後基準線分７２を基準に、座標系のＹ方向へ２行分の境界追跡処理を実行し、境界点を決定した例を示す。
【００９０】
次に、上述の境界追跡処理において、境界点に決定された画素の閾値を取得する方法について図２２を参照して説明する。いま、図２２に示すように回転角θで回転後基準線分７２が得られ、その回転後基準線分７２を基準にして境界追跡処理により出力マトリクスの閾値を決定する場合を想定する。境界追跡処理では、回転後基準線分７２に平行にＹｒ軸（Ｙ軸を回転角θだけ回転させて得られる軸）方向に１行ずつ閾値を決定していくことになり、いま回転後基準線分７２に平行な平行直線７３に沿った点７８の出力マトリクス内における閾値を決定するものと仮定する。
【００９１】
この場合の処理フローチャートを図２３に示す。まず、回転後基準線分７２に平行な平行直線７３を求め（ステップＳ２１）、次に、平行直線７３に垂直な直線（Ｙｒ軸）を求める（ステップＳ２２）。そして、それらの交点７４を求め（ステップＳ２３）、その交点からの閾値決定対象である点７８までの距離を求め（ステップＳ２４）、その距離を画素の個数に換算する（ステップＳ２５）。図２２の例では、交点７４から点７８までの個数はｎ個である。そして、一次マトリクス上で交点７４からＸ軸方向にｎ個の距離に存在する点７７の閾値を、点７８の閾値として保存する（ステップＳ２６）。こうして、回転後の境界追跡処理において、境界点の閾値を決定していく。このように、交点７４からの距離を基準として閾値を決定するのは、回転角θに正しく対応する閾値を取得するためである。例えば、上記の例において、出力マトリクス内の点７８に対応する位置に、一次マトリクス内の点７８に対応する閾値を保存してしまうと、回転角θの回転後の閾値を保存することができない。上記のように回転角θだけ回転したＹｒ軸上の点７４からの距離（画素の個数）が等しくなるように一次マトリクスから閾値を取得すれば、正しくθだけ回転した後の閾値を取得することができる。
【００９２】
なお、上記の例では、基準線分を一次マトリクスの座標系のＸ軸方向（即ち、一次マトリクスの列方向）にとり、回転後基準線分７２を基準として行方向に順に出力マトリクスの閾値を求めている。その代わりに、基準線分を一次マトリクスの座標系のＹ軸方向（即ち、一次マトリクスの行方向）にとり、回転後基準線分を基準として列方向に順に出力マトリクスの閾値を求めてもよい。
【００９３】
（第３の作成方法）
次に、本発明のディザ閾値マスクの第３の作成方法について説明する。上記の第１及び第２の作成方法では、Ｂａｙｅｒ型などの入力ディザマトリクスに対して回転処理を行うことによりディザ閾値マスクを作成したが、本実施形態では、入力ディザマトリクスに対してシフト処理を施すことによりディザ閾値マスクを作成する。このシフト処理によっても、本発明の特徴である、規則性と不規則性を併せ持つディザ閾値マスクを作成することができる。
【００９４】
シフト処理によるディザ閾値マスクの作成方法を、図２４及び図２５を参照して説明する。なお、この方法においても、入力マトリクスとしては、図２５（ａ）に示すＢａｙｅｒ型のディザマトリクスを使用するものとする。また、入力マトリクスを展開してシフト処理を行うために、出力マトリクスよりもサイズの大きな作業領域が使用される。
【００９５】
まず、入力マトリクスを作業領域に展開する（ステップＳ３１）。この例では、図２５（ｂ）に示すように、出力マトリクスを８×８ドットのサイズと仮定しており、図２５（ａ）に示すＢａｙｅｒ型の入力ディザマトリクスを４個組み合わせて、８×８ドットのマトリクスを作業領域に展開している。なお、出力マトリクスがこれより大きい場合は、より多数の入力ディザマトリクスを展開してシフト処理を行うことになる。
【００９６】
次に、シフト処理におけるシフト量を設定する（ステップＳ３２）。シフト処理においては、ステップＳ３１で作業領域上に展開したマトリクスを列（Ｘ軸）方向及び行（Ｙ軸）方向に所定量ずつシフトしてマトリクスの変形を行う。よって、シフト量は、展開されたマトリクスの行毎にＸ方向シフト量が設定され、展開されたマトリクスの列毎にＹ方向シフト量が設定される。図２５（ｃ）は第０行から第３行まではＸ方向シフト量０、第４行から第７行まではＸ方向シフト量１でシフト処理を行った結果を模式的に示す。また、図２５（ｄ）は、第０列から第３列まではＹ方向シフト量０、第４列から第７列まではＹ方向シフト量１でシフト処理を行った結果を模式的に示す。なお、Ｘ方向シフト量、Ｙ方向シフト量ともに任意に設定することができ、Ｘ方向シフト量とＹ方向シフト量が異なっていても構わない。
【００９７】
こうして、シフト量が設定されると、次にそのシフト量に従ってシフト操作が実行される（ステップＳ３３）。具体的には、図２５（ｅ）に示すように、まずＸ方向シフト操作が実行され、第４行から第７行の各ドットがＸ方向に１ずつシフトされる。なお、シフトによってＸ方向側にはみ出たドット列は図２５（ｅ）中の矢印に示すように、Ｘ軸の負方向に移動されて、最終的には８×８ドットの正方形のマトリクスとされる。
【００９８】
次に、Ｘ方向のシフト操作が完了した状態のマトリクスに対して、Ｙ方向シフト操作が実行される。即ち、図２５（ｆ）に示すように、第４列から第７列の各ドットがＹ方向下方に１ずつシフトされ、同様にはみ出た部分が矢印に示すようにＹ軸の上方向に移動されて正方形のマトリクスが作成される。こうして、図２５（ｇ）に示すようにディザ閾値マトリクスが作業領域上で得られ、各閾値を保存して（ステップＳ３４）、ディザ閾値マトリクスの作成が完了する。
【００９９】
なお、上記の例では、Ｘ方向シフト操作の先に行い、次にＹ方向シフト操作を行っているが、逆にＹ方向操作を先に行い、その後でＸ方向シフト操作を行うこととしてもよい。この順序を入れ替えることにより、異なるディザ閾値マトリクスが得られる。図４０から図４３にシフト処理により作成されたディザ閾値マスクの周波数特性を示す。
【０１００】
次に、好適なシフト量について説明する。本発明のディザ閾値マトリクスでは、入力マトリクスとして規則性を有するＢａｙｅｒ型を使用し、これに対して回転処理又はシフト処理を施して不規則性を導入している。本実施形態では、シフト処理により不規則性を導入するので、シフト量の設定が、作成されるディザ閾値マスクの性能に大きな影響を与える。
【０１０１】
図２７（ａ）に、入力マトリクスに対してシフト処理により不規則性を導入した一例を示す。この例では、Ｂａｙｅｒ型ディザマトリクスを横方向に連結し、第０行のドットに対して、３つおきに２つのドットをＹ方向に「−１」シフトした例である。このようにシフト処理により、入力マトリクスが有する規則的な閾値配列パターンがくずれ、不規則な閾値配列パターンが生まれている。
【０１０２】
図２７（ｂ）乃至（ｅ）は、シフト量を変えた場合のドットのシフト例を示す。図２７（ｂ）は１ドット単位でＹ方向に「−１」だけシフトした例であり、図２７（ｃ）は２ドット単位でＹ方向に「−１」だけシフトした例であり、図２７（ｄ）は３ドット単位でＹ方向に「−１」だけシフトした例であり、図２７（ｅ）は４ドット単位でＹ方向に「−１」だけシフトした例である。このように、シフト量を調整すると、そのシフト操作により閾値列が移動する角度αが決まる。図２７（ｂ）の例では角度αは４５度であり、図２７（ｅ）の例では角度αは約１４度である。図２７（ｂ）〜（ｅ）における幅を変化させた場合の角度αの変化を図２６に示す。図２６からわかるように、幅の値を大きくするほど、閾値が移動する角度αの値は小さくなる。
【０１０３】
ここで、先に述べた回転処理によるディザ閾値マトリクスの作成においては、官能評価などの結果、所望の不規則性を有するディザ閾値マトリクスを作成するための好ましい回転角は５〜２０度又は７０〜８５度であり、特に１０度又は８０度付近が好適であることがわかっている。この観点から、シフト処理によりディザ閾値マトリクスを作成する際にも、１０度又は８０度付近の回転角αをとる幅の値が良好であることがわかる。具体的は、図２６より、幅が３〜５の間となるようにシフト量を決定すれば、回転角αが１０度前後になり、好ましいことがわかる。なお、実際には、幅３〜５の間の幅を同じ幅がＸ方向又はＹ方向に連続しないように組み合わせ、ディザ閾値マトリクス全体として約１０度程度の回転角に相当するような配列を作成することがさらに好ましい。
【０１０４】
前述のように、１つのディザ閾値マスク内ではＸ方向及びＹ方向のそれぞれにおいて、ドット位置毎に異なるシフト量を設定することができる。例えば図２５（ｃ）の例では、第０行から第３行までのＸ方向シフト量は０であり、第４行から第７行列までのＸ方向シフト量は１であり、それらは異なっている。但し、そのように異なるシフト量を設定する場合でも、隣接する部分におけるシフト量は±１の範囲内とすることが好ましい。これは、隣接する部分におけるシフト量の絶対値が２以上となると、シフト処理により入力マトリクスが有する規則性が急激に変化するためである。一方、隣接する部分におけるシフト量が±１の範囲内となるようにシフト量を設定すれば、もともと規則性を有する入力マトリクスの所定部分がＸ方向又はＹ方向へ徐々にシフトした状態となり、作成されたディザ閾値マトリクス内における規則性を有する部分と不規則性を有する部分の割合及び位置関係が適切となるからである。
【０１０５】
また、作成されたディザ閾値マトリクスを見た場合、その左端と右端のシフト量は±１以内であることが好ましく、上端と下端のシフト量も±１以内であることが好ましい。そのようにシフト処理におけるＸ方向及びＹ方向のシフト量を設定すれば、作成されたディザ閾値マトリクスを複数個隣接配置して使用する場合に、それらの境界領域において規則性を有する部分と不規則性を有する部分とが適切に配置されることになる。
【０１０６】
［適用例］
本実施の形態のディザ閾値マスクは、表示階調数が少ない（２５６色、４０９６色）ＬＣＤパネルでのディザによる面積階調表現に好適である。すなわちＬＣＤにおいて液晶セルを駆動させるだけで階調表現を行うとすると、駆動による消費電力が大きくなると共に、駆動回路が複雑化する。このため、表示階調数が少ないＬＣＤパネルにおいて、表現可能階調数を補うために本実施形態のディザマスクを用いると、消費電力が低く、また駆動回路が複雑化しないＬＣＤパネルを提供することが可能となる。
【０１０７】
また、本実施形態のディザ閾値マスクは、一個の素子がある程度の画素面積を持つＬＣＤパネルにおいて用いるのが好適である。すなわち、ＬＣＤパネルにおいて画素面積を小さくして解像度を高くすると、画素中における駆動素子および配線の面積の割合が増加し、画素中の表示部の面積の割合が減少するため、ＬＣＤの表示する明るさが暗くなる。このため、解像度を例えば１００〜１５０ｄｐｉ程度としたＬＣＤに対して用いると、表示部の明るさ確保のために有効である。このとき、解像度が低いときには、人間の視覚特性は、ドットパターンにおける規則的なパターンやウォーム模様を認知しやすいが、本実施形態のディザ閾値マスクは、このようなパターンを発生させないため、画質の向上が効果的にはかれる。
【０１０８】
なお、本実施の形態に係るディザ閾値マスクは、ＬＣＤに限らず有機ＥＬディスプレイその他の表示装置、さらにはプリンタの印画部等にも適用可能であることはいうまでもない。
【０１０９】
さらに、本発明の本質は、複数の閾値を持ったディザマスクにおいて、そのディザマスクが持つ特性に特徴があるものである。したがって、すでに提案されているさまざまな画像処理のなかで、処理の一部としてディザ処理を用いるディザ応用処理、あるいはディザマスクを用いて不規則性ノイズ成分を付加する場合など、ディザマスクを使った各種処理に対して、本発明のディザマスクは適用可能である。
【０１１０】
【発明の効果】
以上説明したように、本発明のドット分散型マスクは、連続階調の画像データにおける各画素の階調値を２値または多値に変換するためのドット分散型マスクであって、変換により得られるドットパターンに規則性および不規則性が現れるような閾値の配列を有することを特徴としている。
【０１１１】
このように、ドット分散型を採るので、それぞれのドットが空間的に分離したドットパターンを生じる。このため、ドットの隣接が生じにくい。また、ドットパターンに規則性と共に不規則性を有するので、Ｂａｙｅｒの様な規則的パターンやブルーノイズマスクのような中濃度でのウォーム模様が生じない。従って、好ましい中間調表現が可能となり、画質の向上を図ることができる。
【図面の簡単な説明】
【図１】本実施の形態に係る画像表示装置の概略構成を示す図である。
【図２】ディザ閾値マスクの機能を示す図である。
【図３】本実施の形態に係るディザ閾値マスクの周波数特性を示す図である。
【図４】本実施の形態に係るディザ閾値マスクの周波数特性を示す図である。
【図５】本実施の形態に係るディザ閾値マスクの周波数特性を示す図である。
【図６】本実施の形態に係るディザ閾値マスクの周波数特性を示す図である。
【図７】Ｂａｙｅｒ型マスクによるハーフトーン結果を示す図である。
【図８】ブルーノイズマスクによるハーフトーン結果を示す図である。
【図９】本実施の形態に係るディザ閾値マスクによるハーフトーン結果を示す図である。
【図１０】本実施の形態に係るディザ閾値マトリクスの第１の作成方法のフローチャートである。
【図１１】回転変換の様子を示す図である。
【図１２】閾値の探索順序を示す図である。
【図１３】回転角度と探索回数との関係を示す図である。
【図１４】（ａ） 回転角が０度の場合のディザ閾値マスクによるドットパターンを示す図である。
（ｂ） サイズが４×４で、回転角が１０度の場合のディザ閾値マスクによるドットパターンを示す図である。
（ｃ） サイズが４×４で、回転角が３０度の場合のディザ閾値マスクによるドットパターンを示す図である。
（ｄ） サイズが１３２×１６２で、回転角が１０度の場合のディザ閾値マスクによるドットパターンを示す図である。
（ｅ） サイズが１３２×１６２で、回転角が３０度の場合のディザ閾値マスクによるドットパターンを示す図である。
【図１５】回転角毎のドットパターンに対する官能評価実験の結果を示す図である。
【図１６】２値ディザを説明するためのマトリクス図である。
【図１７】多値ディザを説明するためのマトリクス図である。
【図１８】Ｒ，Ｇ，Ｂ各色に別々のディザマトリクスを対応付けるための装置のブロック図である。
【図１９】ディザ閾値マスクの第２の作成方法のフローチャートである。
【図２０】ディザ閾値マスクの第２の作成方法における回転処理及び境界追跡処理を説明する図である。
【図２１】境界追跡処理における境界点決定方法を模式的に示す図である。
【図２２】境界追跡処理における閾値の決定・保存方法を説明する図である。
【図２３】境界追跡処理における閾値の決定・保存方法のフローチャートである。
【図２４】ディザ閾値マスクの第３の作成方法のフローチャートである。
【図２５】シフト処理の例を説明する図である。
【図２６】シフト処理におけるシフト幅と、対応する回転角との関係を示す図表である。
【図２７】シフト処理におけるシフト例を示す図である。
【図２８】Ｂａｙｅｒ型マスクの周波数特性を示す図である。
【図２９】Ｂａｙｅｒ型マスクの周波数特性を示す図である。
【図３０】Ｂａｙｅｒ型マスクの周波数特性を示す図である。
【図３１】Ｂａｙｅｒ型マスクの周波数特性を示す図である。
【図３２】ブルーノイズマスクの周波数特性を示す図である。
【図３３】ブルーノイズマスクの周波数特性を示す図である。
【図３４】ブルーノイズマスクの周波数特性を示す図である。
【図３５】ブルーノイズマスクの周波数特性を示す図である。
【図３６】特開平６−２２１２３号公報に開示されているマスクの周波数特性を示す図である。
【図３７】特開平６−２２１２３号公報に開示されているマスクの周波数特性を示す図である。
【図３８】特開平６−２２１２３号公報に開示されているマスクの周波数特性を示す図である。
【図３９】特開平６−２２１２３号公報に開示されているマスクの周波数特性を示す図である。
【図４０】シフト処理を利用して作成したディザ閾値マスクの周波数特性を示す図である。
【図４１】シフト処理を利用して作成したディザ閾値マスクの周波数特性を示す図である。
【図４２】シフト処理を利用して作成したディザ閾値マスクの周波数特性を示す図である。
【図４３】シフト処理を利用して作成したディザ閾値マスクの周波数特性を示す図である。
【符号の説明】
１００ 画像処理装置
１０１ 入力手段
１０２ ディザ処理手段
１０３ ＬＣＤパネル[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a dot-dispersed mask, an image display device, a dot-dispersed mask creating method, a dot-dispersed mask for converting the gradation value of each pixel in continuous-tone image data into binary or multi-value, The present invention relates to a creation program and a computer-readable recording medium.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, monochrome or color liquid crystal display (LCD) panels are used in image display units such as cellular phones. In this LCD panel, the transmissivity of the liquid crystal is changed by turning on / off the driving voltage to the liquid crystal cells arranged in a matrix, and a two-gradation or multi-gradation image is displayed. Recently, with the increase in the number of functions such as an Internet function in a mobile phone, it is required to display more information including images on a LCD panel with multiple gradations and high image quality.
[0003]
An LCD panel used in a cellular phone requires display with low power consumption for the purpose of cellular phone. For this reason, when displaying with multiple gradations and high image quality, there are cases where area gradation expression of image data is used together with gradation expression by driving a conventional liquid crystal cell. The area gradation expression of image data is a gradation expression method using human visual characteristics. Since gradation is expressed by data processing, power consumption applied to the liquid crystal cell driving portion can be reduced.
[0004]
In order to express an image of continuous color tone by area gradation expression of image data, the original image needs to be halftoned. When halftoning is performed, a technique using a dither addition mask is known. For example, “An optimal method for two-level rendition of continuous-tone pictures” by Bayer (IEEE International Communication Conference, 1973). (January 11-15), the human visual system takes advantage of the reduced perceptual power for very high frequency signals, and the fixed size is 2m x 2m and 2 (m + 1) It is described that a threshold reference is set using a set of halftone masks (Bayer masks) to capture low spectral components at low frequencies.
[0005]
Y. Yao and K.K. J. et al. “Digital Halftoning a Blue Noise Mask” (Proc. SPIE1452) by Parker describes a “blue noise mask method” that generates a dot pattern that is more inconspicuous in the human visual system than the Bayer mask.
[0006]
Japanese Patent Laid-Open No. 6-22123 discloses an improved technique for a Bayer mask that has a size of, for example, 13 × 9 and spreads a threshold value according to a predetermined rule.
[0007]
[Problems to be solved by the invention]
However, the Bayer mask generates a regular pattern as shown in FIG. This will be described using the frequency characteristics of the Bayer mask shown in FIGS. In each of FIGS. 28 to 31, (a) shows a dot pattern converted into a binary value by a Bayer mask. The dot pattern density has 4 levels of 1/8, 2/8, 3/8, and 4/8, which are shown in FIGS. Each figure (b) shows the amplitude of a two-dimensional FFT obtained by Fourier transforming these dot patterns, and each figure (c) shows a three-dimensional plot of the FFT. In each of the drawings (b) and (c), the fact that the frequency component appears in isolation indicates that the dot pattern contains a lot of this frequency, and the periodicity or regularity of the dot pattern is strong at this frequency. Represents that. As described above, in the Bayer mask, only frequency components exhibiting regularity exist, so that a regular pattern is generated in an image binarized using the Bayer mask.
[0008]
On the other hand, in the blue noise mask, as shown in FIG. 8, a pattern like a snake skin (warm pattern) is generated. This will be described using the frequency characteristics of the blue noise mask shown in FIGS. In each of FIGS. 32 to 35, (a) shows a dot pattern converted into a binary value by a blue noise mask. The density of the dot pattern has four levels of 1/8, 2/8, 3/8, and 4/8, and each is shown in FIGS. Each figure (b) shows the amplitude of a two-dimensional FFT obtained by Fourier transforming these dot patterns, and each figure (c) shows a three-dimensional plot of the FFT. In each of FIGS. (B) and (c), the absence of a conspicuous frequency component other than the component having a frequency of 0 (direct current) indicates that the dot pattern has almost no periodicity or regularity. In this way, the blue noise mask includes only completely irregular components other than the component having a frequency of 0 (direct current). In the region of low density (for example, g = 1/8) or high density (for example, g = 7/8), dot regularity does not appear and a good image can be obtained, but medium density (approximately 1/4 to 3 / In the area 4), the irregular pattern causes dots to appear irregularly and a warm pattern appears.
[0009]
Since these patterns are conspicuous in an image display device with a small number of display gradations, it is desired to improve this point and to improve the quality of images.
[0010]
The mask disclosed in Japanese Patent Laid-Open No. 6-22123 has frequency characteristics as shown in FIGS. In each of FIGS. 36 to 39, (a) shows a dot pattern converted into a binary value by the mask. The density of the dot pattern has four levels of 1/8, 2/8, 3/8, and 4/8, and each is shown in FIGS. Each figure (b) shows the amplitude of a two-dimensional FFT obtained by Fourier transforming these dot patterns, and each figure (c) shows a three-dimensional plot of the FFT. The feature of this mask is that it has a frequency component showing regularity and a frequency component showing irregularity. For example, as shown in the frequency components surrounded by black circles in each figure (c), the peak is large. Since the frequency component appears at a specific position, regularity corresponding to the frequency component as shown in each figure (a) appears. Further, regarding each figure (c), since the irregular component appears in the low frequency region as shown by the white arrow in FIG. 39 (c), there are many low frequency components. There is a strong tendency for the intervals of
[0011]
The present invention has been made in view of such circumstances, and is a dot dispersion type dot dispersion type mask in which a regular pattern or a warm pattern does not appear in the dot pattern and can improve image quality. An object of the present invention is to provide an image display device, a dot dispersion mask creation method, a dot dispersion mask creation program, and a computer-readable recording medium.
[0012]
[Means for Solving the Problems]
In one aspect of the present invention, a dot dispersion type mask is a dot dispersion type mask for converting the gradation value of each pixel in continuous tone image data to binary or multivalue, and is obtained by conversion. It is characterized by having an array of threshold values such that regularity and irregularity appear in the dot pattern.
[0013]
Thus, since the dot dispersion type is adopted, a dot pattern in which each dot is spatially separated is generated. For this reason, dot adjacency is unlikely to occur. Further, since the dot pattern has irregularity as well as regularity, a regular pattern such as Bayer and a warm pattern with a medium density such as a blue noise mask do not occur. Therefore, preferable halftone expression is possible, and the image quality can be improved.
[0014]
One aspect of the above-described dot dispersion mask of the present invention is characterized in that in the frequency region including each frequency component of the dot pattern, the frequency components exhibiting irregularity are scattered in a high frequency region.
[0015]
Since the frequency components exhibiting irregularities are scattered in the high frequency region, there is no pattern regularity due to the presence of specific frequency components, and dot concentration due to the presence of low frequency components. Image quality deterioration can be prevented.
[0016]
In one aspect of the dot dispersion mask of the present invention described above, in the frequency region including each frequency component of the dot pattern, the strength of the frequency component showing regularity is larger than the strength of the frequency component showing irregularity. It is characterized by.
[0017]
As a result, it is possible to eliminate the drawbacks of the blue noise mask composed only of irregular components. That is, since the intensity of the frequency component indicating regularity is greater than the intensity of the frequency component indicating irregularity, it is possible to obtain an image in which the irregular components are scattered while leaving the regularity in the dot pattern. As a result, it is possible to improve the image quality by preventing the occurrence of a warm pattern.
[0018]
One aspect of the above-described dot-dispersed mask of the present invention is characterized in that in the frequency region including each frequency component of the dot pattern, the frequency component showing regularity is locally present in the frequency region.
[0019]
As a result, it is possible to alleviate the disadvantage of a mask having a threshold arrangement such that regularity appears in the dot pattern (for example, a Bayer mask), and because it has a regular component, only an irregular component can be obtained. It is possible to prevent the occurrence of a warm pattern in a blue noise mask including the same. As a result, the image quality can be improved.
[0020]
In another aspect of the present invention, the dot dispersion mask is a dot dispersion mask for converting the gradation value of each pixel in the continuous gradation image data into a binary value or a multi-value, and each of them has a predetermined rule. A regular pattern threshold value group in which a plurality of threshold values are arranged with a certain property and an irregular pattern threshold value group in which a plurality of threshold values are arranged so as not to have the predetermined regularity are combined. And
[0021]
According to this dot dispersion type mask, a plurality of threshold values are arranged with a predetermined regularity to form a regularity pattern, and a plurality of threshold values are arranged so as not to have a regularity and irregularities are formed. A pattern is formed, and both are combined to form a mask. Since the dot pattern has irregularity as well as regularity in this way, a regular pattern such as a Bayer-type dither matrix or a warm pattern with a medium density such as a blue noise mask does not occur. Therefore, preferable halftone expression is possible, and the image quality can be improved.
[0022]
In one aspect of the dot dispersion mask, the dot dispersion mask is created using a basic threshold pattern in which thresholds are arranged according to the predetermined regularity, and the regular pattern threshold group is held by the basic threshold pattern. A threshold value array having the same regularity as a part of the threshold value array. In this way, the regularity of the dot dispersion mask can be maintained by having the same regularity as that of a part of the basic threshold pattern in which thresholds are arranged with a predetermined regularity.
[0023]
In another aspect of the dot dispersion mask, the irregular pattern threshold value group is configured by arranging the threshold values held by the basic threshold pattern so as not to have the predetermined regularity. As a result, it is possible to create a desired dot dispersion mask using only the threshold value of the regular dot pattern prepared in advance.
[0024]
In another aspect of the above-described dot dispersion mask, the regular pattern threshold value group includes a dot that generates a threshold value after rotation and a threshold value before rotation when the basic threshold value pattern is rotated by a predetermined rotation angle. It is characterized in that it is constituted by a threshold value at which dots to be generated coincide. In this case, the dot dispersion mask can be easily created while maintaining regularity by rotating the basic threshold pattern.
[0025]
In another aspect of the dot dispersion mask, the dot dispersion mask is created using a basic threshold pattern in which thresholds are arranged according to the predetermined regularity, and the regular pattern threshold group includes the basic threshold pattern. The threshold value group is constituted by a threshold value that maintains the predetermined regularity when rotated by a predetermined rotation angle, and the irregular pattern threshold value group includes the predetermined rule when the basic threshold pattern is rotated by a predetermined rotation angle. It is characterized by comprising a threshold value that does not maintain the property. Therefore, a part having regularity and a part having irregularity can be created by rotating the basic threshold pattern.
[0026]
In another aspect of the dot dispersion mask, the dot dispersion mask is created using a basic threshold pattern in which thresholds are arranged according to the predetermined regularity, and the regular pattern threshold group includes the basic threshold pattern. The threshold is configured to maintain the predetermined regularity when partially shifted by a predetermined shift amount on the plane of the pattern, and the irregular pattern threshold group includes the basic threshold pattern as the plane of the pattern. It is characterized by comprising a threshold that does not maintain the predetermined regularity when partially shifted by a predetermined shift amount. Therefore, a portion having regularity and a portion having irregularity can be created by the shift process of the basic threshold pattern.
[0027]
According to another aspect of the present invention, an image display device includes a storage unit that stores the dot dispersion type mask described above and a display unit capable of monochrome or color display, and the dot dispersion type mask. An image based on the image data processed by the above is displayed on the display unit.
[0028]
Since the mask of the present image display device adopts a dot dispersion type, a dot pattern in which each dot is spatially separated is generated. For this reason, dot adjacency is unlikely to occur. Further, since the dot pattern has irregularity as well as regularity, a regular pattern such as Bayer and a warm pattern with a medium density such as a blue noise mask do not occur. Therefore, in the present image display device, preferable halftone expression can be achieved, and the image quality can be improved.
[0029]
According to another aspect of the present invention, in the image processing apparatus, a storage unit that stores the dot dispersion mask, an input unit that receives input image data, and the input image data using the dot dispersion mask. Image processing means for performing image processing on the image processing apparatus.
[0030]
In a similar aspect, the image processing method includes a step of receiving input image data and a step of performing image processing on the input image data using the dot dispersion mask.
[0031]
According to the above-described image processing apparatus or image processing method, for example, input image data is received from a scanner, an imaging unit, or other external image data source, and image processing is performed using the above-described dot dispersion type mask. Preferred halftone expression and the like are possible.
[0032]
In a similar aspect, an image processing program executed on a computer receives input image data, and performs image processing on the input image data using the dot dispersion mask. Are executed by the computer. By executing this image processing program on a computer, the above-described image processing apparatus can be realized.
[0033]
In another aspect of the present invention, in a dot dispersion type mask creation method, a dot dispersion type mask for converting the gradation value of each pixel in continuous tone image data to binary or multivalued is provided. A method of creating a mask having an array of threshold values that makes regularity appear in the dot pattern obtained by conversion, and so that irregularity appears in the dot pattern obtained by conversion with respect to the created mask And a step of changing the array of threshold values.
[0034]
As described above, since the threshold value array is changed so that irregularity appears based on the mask having the threshold value arrangement such that regularity appears in the dot pattern, the threshold value such that regularity appears in the dot pattern is changed. It is possible to alleviate the disadvantages while leaving the advantages of the mask having an array (for example, Bayer mask), and it is possible to prevent the occurrence of a warm pattern in the blue noise mask containing only irregular components. As a result, the image quality can be improved.
[0035]
In another aspect of the present invention, a method for creating a dot-dispersed mask for converting the tone value of each pixel in continuous tone image data to binary or multivalue is based on the size of the mask to be created. Developing a basic threshold pattern in which thresholds are arranged with a predetermined regularity in a region of a primary matrix having a large size; and relative coordinates of the thresholds in the primary matrix by a predetermined rotation angle on the region of the primary matrix And a step of arranging the threshold values of the primary matrix so as not to overlap at positions corresponding to the coordinates after the rotation.
[0036]
According to the above method for creating a dot dispersion mask, a basic threshold pattern in which thresholds are arranged with a predetermined regularity is developed in a primary matrix area, and each coordinate is set at a predetermined rotation angle on the primary matrix area. Rotate and place a primary matrix threshold at each rotated coordinate. When the threshold value is arranged, a portion where the regularity of the basic threshold pattern is retained and a portion where the regular threshold pattern is not retained are generated by rotation. As a result, the created dot dispersion mask has a portion having regularity and a portion having irregularity. Will be mixed.
[0037]
In one aspect of the method for creating the dot dispersion mask, when the coordinates by the rotation are mapped to the same position as other coordinates that have been rotated, and another threshold that has already been rotated is arranged at the position. The method includes a step of searching for coordinates where the threshold value is not arranged and existing in the vicinity of the position and arranging the threshold value. The rotation as described above may cause multiple points to be mapped to the same position. In such a case, if the threshold value of the primary mask has already been input to the mask, it exists in the vicinity of that position. The coordinates where the threshold is not arranged are searched and the threshold is arranged. Thereby, it is possible to avoid duplication of threshold values due to rotation of coordinates.
[0038]
In another aspect of the present invention, a dot-dispersed mask creation method for converting the tone value of each pixel in continuous tone image data to binary or multivalue is a size larger than the size of the mask to be created. Developing a basic threshold pattern in which threshold values are arranged with predetermined regularity in a region of the primary matrix, rotating one reference line segment in the primary matrix by a predetermined rotation angle, and rotating the reference line It should be created by repeatedly storing each threshold value on the primary matrix corresponding to the position of the minute and acquiring and storing the threshold value from the primary matrix along the reference line segment after the rotation. Determining a threshold value of the mask.
[0039]
According to the above method for creating a dot dispersion matrix, a basic threshold pattern in which thresholds are arranged with a predetermined regularity is developed on a primary matrix, and then a reference line segment is rotated by a predetermined rotation angle, and after the rotation A dot dispersion mask is created by acquiring and storing a threshold value along the reference line segment. Therefore, a suitable dot dispersion matrix can be created by the rotation process and the process of determining the threshold value from the primary matrix along with the rotation process.
[0040]
In one aspect of the method for creating the dot dispersion matrix, the threshold acquired for the specific threshold position along the reference line segment is a first line segment perpendicular to the rotated reference line segment, It is acquired based on the distance from the intersection with the second straight line passing through the specific threshold position perpendicular to the first line segment to the specific threshold position. As a result, it is possible to create a dot dispersion matrix by acquiring a threshold value at a correct position according to the rotation angle.
[0041]
In one aspect of the method for creating the dot dispersion matrix, the predetermined rotation angle is in the range of 5 to 20 degrees or 70 to 85 degrees. The inventor has found that the psychological table value for image quality is the highest by rotating within a range of 5 to 20 degrees or 70 to 85 degrees. This makes it possible to create a mask that can improve the image quality.
[0042]
In another aspect of the present invention, a dot-dispersed mask creation method for converting the tone value of each pixel in continuous tone image data into binary or multivalue is a size larger than the size of the mask to be created. Developing a basic threshold pattern in which threshold values are arranged with a predetermined regularity in a region of the primary matrix, setting a shift amount for shifting the threshold value array, and according to the shift amount on the primary matrix Shifting the threshold array.
[0043]
According to the above method for creating a dot dispersion mask, a basic threshold pattern in which thresholds are arranged with a predetermined regularity is developed on a primary matrix, and a shift amount is set for the basic threshold pattern to shift the threshold arrangement. Thus, a dot dispersion mask having both regularity and irregularity can be created.
[0044]
In one aspect of the method for creating a dot dispersion mask, the shift amount is set by setting a predetermined number of adjacent thresholds as a group, and the predetermined number set for groups adjacent to each other is different. And This introduces moderate irregularity into the created dot dispersion matrix.
[0045]
In one aspect of the method for creating the dot dispersion mask, a difference in shift amount between the adjacent threshold values is within ± 1. As a result, since the basic threshold pattern is not suddenly changed in the dot dispersion mask, irregularities can be introduced while the regularity of the basic threshold pattern is appropriately preserved.
[0046]
In one aspect of the method for manufacturing a dot dispersion mask, the angle generated by the shift amount is in the range of 5 to 20 degrees or 70 to 85 degrees. This makes it possible to create a mask that can improve the image quality.
[0047]
In one aspect of the above method for creating a dot dispersion mask, the difference in the shift amount in the horizontal direction of each threshold value at the upper and lower ends of the mask to be created, and the vertical direction of each threshold value at the left and right edges of the mask to be created. The difference in shift amount is within ± 1. Thereby, when a plurality of dot dispersion masks are connected and used, it is possible to prevent the regularity and irregularity from changing suddenly at the boundary portion.
[0048]
In one aspect of the method for creating a dot dispersion mask, the basic threshold pattern is a Bayer-type dither matrix. This makes it possible to eliminate the disadvantages with a simple operation while retaining the advantages of the Bayer-type dither matrix.
[0049]
In another aspect of the present invention, it is possible to provide a dot dispersion mask creation program that, when executed on a computer, causes the computer to execute each step of the dot dispersion mask creation method.
[0050]
Furthermore, it is possible to provide a recording medium on which the dot dispersion mask creation program is recorded. By reading the recording medium into a computer and executing it, the above dot dispersion mask creation method can be executed.
[0051]
DETAILED DESCRIPTION OF THE INVENTION
[Dither method]
Before describing the embodiment, a multi-level systematic dither method for a color image will be described. The dither method is a method of making the intermediate gray level of the display gray level appear in a pseudo manner in a device with a small number of display gray levels. This is called multi-value dither. Multi-level dither is composed of a combination of binary dithers.
[0052]
The binary dither will be described. First, a dither matrix indicating the order of threshold values shown in FIG. 16C is determined, and a threshold matrix shown in FIG. 16B is calculated from the dither matrix. Next, between the input image shown in (a) and the threshold matrix shown in (b), threshold processing is performed to output 255 if it is equal to or greater than the threshold, and 0 if it is less than 0, 255 shown in (d). A pseudo halftone image (output image) consisting of the above binary values is output. The average value of each threshold value of the pseudo halftone image shown in (d) is 128, and the halftone 128 is expressed by a binary pattern of 0,255.
[0053]
Multi-level dither is performed by dividing the range of input gradations and performing binary dither within each range. When the multi-value is four values, the output gradation is four values of 0, 85, 170, 255, and the input gradation is (1) 0-85, (2) 86-170, (3) 171-255. Divided into three ranges.
[0054]
Depending on which gradation range the input image belongs to, threshold matrixes shown in (b) to (d) are generated from the dither matrix shown in FIG. When the input image is the image shown in FIG. 16A, the gradation 128 is included in the range {circle around (2)}. Therefore, the threshold matrix shown in FIG. 16C is generated and binary dither using gradations 85 and 170 is generated. And a pseudo halftone image as shown in (e) is output.
[0055]
The above is a case where dithering is performed on an image in which one pixel is composed of one color (gray). However, when dithering is performed on a color image, one pixel is composed of three colors of R, G, and B. Therefore, one pixel is divided into R, G, and B color components, and each pixel is regarded as an image composed of one color and dithered. At this time, the same matrix may be used for each color of R, G, and B, or separate matrices may be used. Further, when different dither matrices are associated with each color, as shown in FIG. 18, the R, G, and B color matrices are individually held in the memories 1201 to 1203, and the dither processing means 1204 to 1206 performs dither processing. do it.
[0056]
Hereinafter, embodiments of the present invention will be specifically described.
[0057]
FIG. 1 is a block diagram showing a configuration of an image display apparatus according to an embodiment of the present invention, and FIG. 2 is a diagram showing a function of a dither threshold mask. The image display apparatus 100 includes an input unit 101, a dither processing unit 102, and an LCD panel 103 that performs color display.
[0058]
The input unit 101 captures an image obtained by a scanner or an imaging unit as input image data. The dither processing unit 102 divides the input image data captured by the input unit 101 into planes indicating images for each of R, G, and B, and sets the R, G, and B planes based on the dither matrix. Pseudo halftone processing using a multi-valued systematic dither method is performed in association with the threshold matrix (dither threshold mask shown in FIG. 2). This dither threshold mask is a feature of the present embodiment.
[0059]
Then, the image data after the pseudo halftone process are collectively output to the LCD panel 103. The LCD panel 103 includes a screen in which R, G, and B primary color filters each having a rectangular shape are arranged in a stripe pattern, and a full-color image is displayed on the screen by mixing the color filters. That is, as shown in FIG. 2, the LCD panel 103 displays a binary or multi-valued image by changing the transmittance of the liquid crystal by turning on / off the driving voltage to the liquid crystal cells arranged in a matrix. To do.
[0060]
The dither threshold mask of the present embodiment has the frequency characteristics shown in FIGS. In each of FIGS. 3 to 6, (a) shows a dot pattern converted into a binary value by the dither threshold mask of the present embodiment. The density of the dot pattern has four levels of 1/8, 2/8, 3/8, and 4/8, which are shown in FIGS. Each figure (b) shows the amplitude of a two-dimensional FFT obtained by Fourier transforming these dot patterns, and each figure (c) shows a three-dimensional plot of the FFT. As described above, in the dither threshold mask according to the present embodiment, there are frequency components that exhibit regularity similar to the Bayer mask, and frequency components that exhibit irregularity are scattered in the high-frequency region. Furthermore, the frequency component showing regularity has a larger amplitude value or energy than the frequency component showing irregularity. For this reason, irregularity appears in the dot pattern as well as regularity.
[0061]
As shown in FIG. 9, the dither threshold mask of the present embodiment does not generate a regular pattern like the Bayer mask. Also, a warm pattern due to being too irregular like a blue noise mask is not generated. That is, a pattern in which dots are irregularly connected, such as a blue noise mask, is avoided by a pattern in which a cross and an X character remain. For this reason, image quality is improved.
[0062]
The inventor rotates the position of each threshold based on the Bayer-type mask, and spreads the threshold so that there is no missing threshold when rotated, so that the dither threshold mask having the above frequency characteristics is obtained. I found out that it can be created. That is, the regular frequency component based on the Bayer type mask can be left in a rotated state with regularity. Then, irregular frequency components can be added by spreading the threshold value so that the threshold value is not lost when rotated. As a result, it is possible to alleviate the disadvantages while leaving the advantages of the Bayer mask, and to prevent the occurrence of warm patterns in the blue noise mask containing only irregular components. As a result, the image quality can be improved. Furthermore, since the blue noise mask needs to have only irregular components, that is, have all frequency components evenly, the mask size is large, such as 128 × 128 or 256 × 256. On the other hand, since the dither threshold mask according to the present embodiment is a relatively small mask such as 4 × 4 or 16 × 16, the memory capacity can be reduced. As a result, it is suitable for a small LCD such as a mobile phone.
[0063]
[How to create a dither threshold mask]
(First creation method)
Next, a first method for creating a dither threshold mask according to the present embodiment will be described with reference to the flowchart shown in FIG. In the following description, the dither threshold mask is expressed as a matrix. Here, for example, a Bayer type (4 × 4) is used for the input dither matrix, and the dither matrix is created. In the flowchart shown in FIG. 10, the dither matrix is read (step S1), and the areas of the output matrix and the primary matrix are secured (step S2). Next, the threshold values of the dither matrix are tiled in the primary matrix (step S3). Further, the coordinates of the output matrix are converted using a rotation matrix (step S4).
[0064]
Next, it is determined whether or not the threshold value of the corresponding primary matrix coordinate has been selected (step S5). If the threshold value of the corresponding primary matrix coordinate has been selected, from the periphery of the coordinate of the primary matrix An unselected threshold value in the vicinity is searched (step S6), and the process proceeds to step S7. On the other hand, if the threshold value of the corresponding primary matrix coordinate has not been selected in step S5, the process proceeds to step S7.
[0065]
Next, the threshold value is stored in the output matrix (step S7), and it is determined whether the threshold value is stored in the entire output matrix (step S8). If the threshold value is not stored in the entire output matrix, the process proceeds to step S4. If the threshold value is stored in the entire output matrix, the process ends.
[0066]
The output matrix rotation process in steps S1 to S8 includes the following three processes. That is, (1) pre-processing of dither matrix rotation, (2) conversion of coordinates using a rotation matrix, and (3) processing of searching for nearby threshold values when the converted points are mapped to the same position. .
[0067]
(1) The preprocessing of the dither matrix rotation is performed as follows. That is, the input dither matrix is read to secure an output matrix area of a specified size. Next, when the output matrix is rotated, a primary matrix region having a sufficiently large area so as not to protrude is secured. The dither matrix input to the area is laid out in a tile shape as shown in FIG.
[0068]
(2) The process of converting the coordinates by the rotation matrix is performed as follows. That is, when the coordinates of the primary matrix are (x, y), the coordinates of the output matrix are (X, Y), and each rotation is θ, the conversion formula for rotating the dither matrix is as shown in the following formula (1). .
[0069]
[Expression 1]

[0070]
However, when mapping is performed from the input side to the output side, that is, from the primary matrix side to the output matrix side, when all the rotations are completed, a portion having no threshold appears in the output matrix. For this reason, a method for obtaining coordinates by inverse transformation with the output side as a reference is adopted. The conversion formula of this inverse conversion is as shown in the following formula (2). Further, the state of the inverse transformation is as shown in FIG.
[0071]
[Expression 2]

[0072]
When the coordinates of the conversion destination are obtained as described above, the points after conversion from the output matrix side to the primary matrix side may be mapped to the same position. Therefore, in the present embodiment, it is assumed that the threshold value of the primary matrix can be selected only once. If it has already been selected, a threshold value that has not yet been selected is searched from the vicinity of the position.
[0073]
(3) When the converted points are mapped at the same position, the process of searching for a nearby threshold is performed as follows. That is, when the converted points are mapped to the same position, the thresholds not yet selected from the vicinity of the position are the thresholds not yet selected from among the eight neighbors as shown in FIG. Explore. Next, a threshold value that minimizes the Euclidean distance is selected from the corresponding ones (however, here, the distance from the floating-point type coordinates obtained by Expression (2)) is selected. If it is not found in the vicinity of 8, the search is performed in the same manner by expanding the range to the vicinity of 16 and 24.
[0074]
If the matrix is created by the above method, the regular components of the basic dither matrix can remain in a rotated form while maintaining regularity. Also, irregular components can be added by searching and spreading the thresholds so that missing and overlapping thresholds do not occur when rotated.
[0075]
Comparing the strengths of these regular and irregular components, the underlying dither matrix is regular and many of its components are held in a rotated form, so the regular components are not strong. It becomes stronger than the strength of the regular component. The magnitude of the intensity of these components can be expressed as the magnitude of the amplitude in the frequency domain or the square of the amplitude as described above with reference to FIGS.
[0076]
Next, the result of investigating the number of searches for each rotation angle in the rotation conversion will be described. As shown in FIG. 13, it can be confirmed that the number of searches is small near 0 degrees and 90 degrees, and mapping to the same position is rare.
[0077]
Further, a dither matrix (dot dispersion mask) in which dot patterns are dispersed can be obtained by the method using the rotation matrix as described above. Here, as described above, a Bayer type (4 × 4) is used for the input dither matrix, and a dither matrix is created in each size of 4 × 4 and 132 × 162 (a typical size of a 2-inch LCD panel). did. FIG. 14 shows a dot pattern in which only dots whose threshold value is 0 are turned on. FIGS. 14B and 14C show dot patterns having a size of 4 × 4 and rotation angles of 10 degrees and 30 degrees, respectively. FIGS. 9D and 9E show dot patterns with a size of 132 × 162 and rotation angles of 10 degrees and 30 degrees. FIG. 5A shows a dot pattern obtained by using a rotation angle of 0 °, that is, a Bayer type (4 × 4) as it is. However, any pattern shows a pattern in which dots having a threshold value of 0 are laid out in a tile shape to be 132 × 162.
[0078]
As can be seen from FIGS. 14B and 14C, when the image is rotated at a size of 4 × 4, the dot pattern hardly rotates even if the rotation operation is performed by designating a predetermined rotation angle. This means that when Bayer type 4 × 4 is used for the input dither matrix, a certain range of spread is necessary to enable the rotation operation. Here, the spread range refers to the size of the output matrix in the description of the flowchart of FIG. The inventor has empirically found that when a size of 4 × 4 is used as the size of the basic input dither matrix, the rotation operation works effectively if the size of the output matrix is about 16 × 16 or more. I have confirmed. It is also confirmed that a pattern almost the same as the 132 × 162 pattern shown in FIGS. 14D and 14E can be generated for any size larger than 16 × 16. is doing.
[0079]
Further, looking at the characteristics of the dot pattern due to the difference in rotation angle, from the result of 132 × 162, it was shown that the dot pattern was more disordered when the rotation angle was 30 degrees than when it was 10 degrees. This is because the point is often mapped to the same position at 30 degrees, and in FIG. 13, the number of searches is large at 30 degrees.
[0080]
Next, sensory evaluation experiments were performed to determine the best rotation angle. This is an experiment performed by observing a dot pattern. Specifically, a dither matrix having different rotation angles in the range of 0 to 45 degrees was created, and the image quality by the obtained dither matrix was evaluated by a paired comparison method to determine the best rotation angle. As shown in FIG. 15, the evaluation result showed the highest psychological evaluation value when the rotation angle was 10 degrees. The psychological evaluation value on the vertical axis in FIG. 15 indicates that the larger the value, the better the image quality compared to the others. That is, when the rotation range is considered as 0 to 90 degrees, 10 or 80 degrees is a desirable angle. For this reason, in the present embodiment, the rotation angle is rotated within a range of 5 to 20 degrees or 70 to 85 degrees. More preferably, it rotates at 10 degrees or 80 degrees. As described above, the present inventor has found that the psychological value for the image quality is the highest by rotating within the range of 5 to 20 degrees or 70 to 85 degrees. This makes it possible to create a mask that can improve the image quality.
[0081]
(Second creation method)
Next, a second method for creating a dither threshold mask according to the first embodiment will be described. In the first creation method described above, a Bayer-type dither matrix is used as the input matrix, and a dither threshold value matrix having both regularity and irregularity is created by rotating this. The rotation process was performed by converting the threshold value of the input matrix using the rotation matrix.
[0082]
In contrast, the second creation method uses a Bayer-type dither matrix as an input matrix, and creating a dither threshold matrix having both regularity and irregularity by rotation processing is the same as the first creation method. Similar to the first generation method, the rotation process uses a digital line segment and boundary tracking algorithm instead of using a rotation matrix.
[0083]
FIG. 19 shows a flowchart of the second generation method of the dither threshold mask. Again, the dither threshold mask is called a matrix. First, a Bayer-type dither matrix is read as an input matrix (step S11), and areas of a primary matrix and an output matrix having a predetermined size are secured (step S12). Similar to the first creation method, the primary matrix can be considered as a work area used for the mask creation work. Referring to FIG. 20, a Bayer type dither matrix schematically shown in FIG. 20A is prepared. Then, as shown in FIG. 20 (b), a primary matrix area of a predetermined size is secured on a certain coordinate system (here, the XY coordinate system), and a plurality of dither matrices are spread over the primary matrix area. A matrix region is formed. In addition, a storage area corresponding to the size of the output matrix is secured on a memory or the like.
[0084]
Next, assuming that the rotation angle of the rotation process is θ, as shown in FIG. 20C, it passes through the origin O set in the coordinate system of the primary matrix and passes through one axis (for example, the X axis) on the coordinate system. And a line segment having an angle θ (reference line segment 72 after rotation) is drawn, and the threshold value of each point existing on the line segment in the primary matrix is stored as a threshold value in the output matrix (step S13). ). As a result, a threshold value on the post-rotation reference line segment 72 in the output matrix obtained by the rotation process of the angle θ is obtained.
[0085]
Next, another threshold value in the output matrix is obtained by boundary tracking processing using the post-rotation reference line segment 72 in the output matrix thus obtained as a reference. Specifically, first, the boundary tracking start position is determined and its threshold value is obtained (step S14), and the threshold value is input into the input matrix along the reference line after rotation obtained in step S13 with the start position as the starting point. And stored as an output matrix threshold value (step S15).
[0086]
In this way, the threshold value of the output matrix having a predetermined size is sequentially determined by sequentially determining the threshold value along the boundary with the first obtained reference line segment after rotation as a reference. Then, when all the threshold values of the output matrix are determined and stored (step S16; Yes), the process ends.
[0087]
FIG. 21 shows an example of the boundary tracking process performed in step S14. The boundary tracking process is a process of determining a boundary point adjacent to the line segment along the already obtained line segment and determining a threshold value for the boundary point. FIG. 21A shows two line segments (line segment 90 and line segment 91) having different inclinations. Assume that a line segment 90 is obtained as the post-rotation reference line segment 72 in step S13. FIG. 21A shows boundary points already determined adjacent to the line segment, and boundary candidate points determined based on these boundary points. In the case of a line segment rising to the right, by using the boundary candidate matrix shown in FIG. 21B and applying this to one boundary point, boundary candidate points can be obtained around the boundary candidate matrix. The boundary candidate matrix illustrated in FIG. 21B is a matrix having one boundary point at the lower left and the other three adjacent points as boundary candidate points.
[0088]
Among the boundary candidate points obtained using the boundary candidate matrix, a point adjacent to the line segment, not a point constituting the line segment itself, is determined as the boundary point, and a threshold corresponding to the point on the primary matrix Is stored as a threshold adjacent to the rotated reference line segment 72 in the output matrix. A boundary point is determined along the post-rotation reference line segment 72, and by storing the threshold value, the threshold value for one line is stored along the post-rotation reference line segment 72.
[0089]
Next, the threshold value in the output matrix is determined for each line in order from the post-rotation reference line segment 72 by performing the same boundary tracking process on the line segment constituted by the boundary point thus determined. Eventually, all the threshold values in the output matrix are stored, and the output matrix is completed. FIG. 21C shows an example in which boundary tracking processing for two lines is performed in the Y direction of the coordinate system based on the reference line segment 72 after rotation, and boundary points are determined.
[0090]
Next, a method for acquiring the threshold value of the pixel determined as the boundary point in the boundary tracking process described above will be described with reference to FIG. Now, assume that a post-rotation reference line segment 72 is obtained at a rotation angle θ as shown in FIG. 22, and the threshold value of the output matrix is determined by boundary tracking processing using the post-rotation reference line segment 72 as a reference. In the boundary tracking process, the threshold is determined line by line in the direction of the Yr axis (axis obtained by rotating the Y axis by the rotation angle θ) parallel to the reference line segment 72 after rotation. Assume that the threshold in the output matrix of points 78 along a parallel line 73 parallel to the line segment 72 is determined.
[0091]
A processing flowchart in this case is shown in FIG. First, a parallel straight line 73 parallel to the post-rotation reference line segment 72 is obtained (step S21), and then a straight line (Yr axis) perpendicular to the parallel straight line 73 is obtained (step S22). Then, the intersection 74 is obtained (step S23), the distance from the intersection to the point 78 that is the threshold determination target is obtained (step S24), and the distance is converted into the number of pixels (step S25). In the example of FIG. 22, the number from the intersection 74 to the point 78 is n. Then, the threshold value of the point 77 existing at n distances in the X-axis direction from the intersection 74 on the primary matrix is stored as the threshold value of the point 78 (step S26). Thus, the threshold value of the boundary point is determined in the boundary tracking process after rotation. As described above, the threshold is determined based on the distance from the intersection 74 in order to obtain a threshold that correctly corresponds to the rotation angle θ. For example, in the above example, if the threshold corresponding to the point 78 in the primary matrix is stored at the position corresponding to the point 78 in the output matrix, the threshold after rotation of the rotation angle θ cannot be stored. . If the threshold is acquired from the primary matrix so that the distance (number of pixels) from the point 74 on the Yr axis rotated by the rotation angle θ as described above is equal, the threshold after correctly rotating by θ is acquired. Can do.
[0092]
In the above example, the reference line segment is taken in the X-axis direction of the coordinate system of the primary matrix (that is, the column direction of the primary matrix), and the threshold value of the output matrix is obtained sequentially in the row direction with reference to the rotated reference line segment 72. ing. Instead, the reference line segment may be taken in the Y-axis direction of the primary matrix coordinate system (that is, the row direction of the primary matrix), and the threshold value of the output matrix may be obtained sequentially in the column direction with the post-rotation reference line segment as a reference.
[0093]
(Third creation method)
Next, a third method for creating a dither threshold mask according to the present invention will be described. In the first and second creation methods described above, the dither threshold mask is created by performing rotation processing on an input dither matrix such as a Bayer type, but in this embodiment, shift processing is performed on the input dither matrix. To create a dither threshold mask. Also by this shift processing, a dither threshold mask having both regularity and irregularity, which is a feature of the present invention, can be created.
[0094]
A method of creating a dither threshold mask by shift processing will be described with reference to FIGS. In this method, the Bayer-type dither matrix shown in FIG. 25A is used as the input matrix. Further, in order to perform shift processing by expanding the input matrix, a work area having a size larger than that of the output matrix is used.
[0095]
First, the input matrix is developed in the work area (step S31). In this example, as shown in FIG. 25 (b), it is assumed that the output matrix has a size of 8 × 8 dots, and four Bayer type input dither matrices shown in FIG. An 8-dot matrix is developed in the work area. If the output matrix is larger than this, a larger number of input dither matrices are developed and shift processing is performed.
[0096]
Next, a shift amount in the shift process is set (step S32). In the shift process, the matrix developed in step S31 is shifted by a predetermined amount in the column (X-axis) direction and the row (Y-axis) direction to deform the matrix. Therefore, as the shift amount, an X-direction shift amount is set for each row of the expanded matrix, and a Y-direction shift amount is set for each column of the expanded matrix. FIG. 25C schematically shows the result of shift processing with the X-direction shift amount 0 from the 0th row to the 3rd row and the X-direction shift amount 1 from the 4th row to the 7th row. FIG. 25D schematically shows the result of shift processing with the Y-direction shift amount 0 from the 0th column to the 3rd column and the Y-direction shift amount 1 from the 4th column to the 7th column. . Note that both the X-direction shift amount and the Y-direction shift amount can be arbitrarily set, and the X-direction shift amount and the Y-direction shift amount may be different.
[0097]
After the shift amount is set in this way, a shift operation is next executed according to the shift amount (step S33). Specifically, as shown in FIG. 25E, first, an X-direction shift operation is performed, and each dot in the fourth to seventh rows is shifted one by one in the X direction. Note that the dot row protruding to the X direction side by the shift is moved in the negative direction of the X axis as shown by an arrow in FIG. 25E, and finally becomes a square matrix of 8 × 8 dots. The
[0098]
Next, a Y-direction shift operation is performed on the matrix in a state where the X-direction shift operation has been completed. That is, as shown in FIG. 25 (f), each dot in the fourth row to the seventh row is shifted downward by one in the Y direction, and the protruding portion similarly moves upward in the Y axis as indicated by an arrow. A square matrix is created. In this way, a dither threshold matrix is obtained on the work area as shown in FIG. 25 (g), each threshold is stored (step S34), and creation of the dither threshold matrix is completed.
[0099]
In the above example, the X-direction shift operation is performed first and then the Y-direction shift operation. However, the Y-direction operation may be performed first, and then the X-direction shift operation may be performed thereafter. . By changing this order, a different dither threshold matrix is obtained. 40 to 43 show the frequency characteristics of the dither threshold mask created by the shift process.
[0100]
Next, a suitable shift amount will be described. In the dither threshold matrix of the present invention, a Bayer type having regularity is used as an input matrix, and irregularity is introduced by performing rotation processing or shift processing on the Bayer type. In the present embodiment, irregularities are introduced by the shift processing, so that the setting of the shift amount greatly affects the performance of the dither threshold mask to be created.
[0101]
FIG. 27A shows an example in which irregularities are introduced into the input matrix by shift processing. In this example, the Bayer-type dither matrix is connected in the horizontal direction, and two dots every three dots are shifted by “−1” in the Y direction with respect to the dots in the 0th row. In this way, the regular threshold value array pattern of the input matrix is broken by the shift process, and an irregular threshold value array pattern is born.
[0102]
FIGS. 27B to 27E show dot shift examples when the shift amount is changed. FIG. 27B is an example in which “−1” is shifted in the Y direction in units of one dot, and FIG. 27C is an example in which “−1” is shifted in the Y direction in units of two dots. FIG. 27D shows an example of shifting by “−1” in the Y direction in units of 3 dots, and FIG. 27E shows an example of shifting by “−1” in the Y direction in units of 4 dots. As described above, when the shift amount is adjusted, the angle α by which the threshold row is moved is determined by the shift operation. In the example of FIG. 27B, the angle α is 45 degrees, and in the example of FIG. 27E, the angle α is about 14 degrees. FIG. 26 shows changes in the angle α when the widths in FIGS. 27B to 27E are changed. As can be seen from FIG. 26, the greater the width value, the smaller the value of the angle α at which the threshold moves.
[0103]
Here, in the creation of the dither threshold matrix by the rotation processing described above, a preferable rotation angle for creating a dither threshold matrix having a desired irregularity as a result of sensory evaluation or the like is 5 to 20 degrees or 70 to 70. It has been found that it is 85 degrees, and in particular, around 10 degrees or 80 degrees is suitable. From this point of view, it can be seen that the value of the width that takes the rotation angle α around 10 degrees or 80 degrees is also good when the dither threshold value matrix is created by the shift processing. Specifically, it can be seen from FIG. 26 that if the shift amount is determined so that the width is between 3 and 5, the rotation angle α is approximately 10 degrees, which is preferable. Actually, the width between widths 3 to 5 is combined so that the same width does not continue in the X direction or the Y direction, and an array corresponding to a rotation angle of about 10 degrees as a whole dither threshold matrix is created. More preferably.
[0104]
As described above, different shift amounts can be set for each dot position in each of the X direction and the Y direction within one dither threshold mask. For example, in the example of FIG. 25C, the X-direction shift amount from the 0th row to the 3rd row is 0, the X-direction shift amount from the 4th row to the 7th matrix is 1, and they are different. Yes. However, even when such different shift amounts are set, it is preferable that the shift amounts in the adjacent portions are within a range of ± 1. This is because when the absolute value of the shift amount in the adjacent portion is 2 or more, the regularity of the input matrix is rapidly changed by the shift process. On the other hand, if the shift amount is set so that the shift amount in the adjacent portion is within the range of ± 1, the predetermined portion of the input matrix having the regularity is gradually shifted in the X direction or the Y direction. This is because the ratio and the positional relationship between the regular part and the irregular part in the dither threshold matrix are appropriate.
[0105]
Further, when the created dither threshold matrix is viewed, the shift amount between the left end and the right end is preferably within ± 1, and the shift amount between the upper end and the lower end is also preferably within ± 1. If the shift amounts in the X direction and the Y direction in the shift process are set in this way, when using a plurality of created dither threshold matrixes arranged adjacent to each other, a portion having irregularity and irregularity in those boundary regions are used. The portion having the property is appropriately arranged.
[0106]
[Application example]
The dither threshold mask of this embodiment is suitable for area gradation expression by dithering on an LCD panel with a small number of display gradations (256 colors, 4096 colors). In other words, if gradation representation is performed simply by driving a liquid crystal cell in an LCD, power consumption by driving increases and the driving circuit becomes complicated. For this reason, in an LCD panel with a small number of display gradations, when the dither mask of this embodiment is used to supplement the number of expressible gradations, an LCD panel with low power consumption and no complicated drive circuit is provided. Is possible.
[0107]
Further, the dither threshold mask of this embodiment is preferably used in an LCD panel in which one element has a certain pixel area. That is, when the pixel area is reduced and the resolution is increased in the LCD panel, the ratio of the area of the drive element and the wiring in the pixel is increased, and the ratio of the area of the display portion in the pixel is decreased. Becomes darker. For this reason, when used for an LCD having a resolution of, for example, about 100 to 150 dpi, it is effective for ensuring the brightness of the display unit. At this time, when the resolution is low, the human visual characteristic is likely to recognize a regular pattern or a warm pattern in the dot pattern, but the dither threshold mask of the present embodiment does not generate such a pattern. The improvement is effective.
[0108]
Needless to say, the dither threshold mask according to the present embodiment can be applied not only to the LCD but also to an organic EL display and other display devices, and further to a printing unit of a printer.
[0109]
Further, the essence of the present invention is characterized in the characteristics of the dither mask having a plurality of threshold values. Therefore, dither mask is used in various image processing that has been proposed, such as dither application processing using dither processing as part of the processing, or adding irregular noise components using dither mask. The dither mask of the present invention can be applied to various processes.
[0110]
【The invention's effect】
As described above, the dot dispersion type mask of the present invention is a dot dispersion type mask for converting the gradation value of each pixel in continuous tone image data to binary or multivalue, and obtained by conversion. The dot pattern is characterized by having an array of threshold values such that regularity and irregularity appear.
[0111]
Thus, since the dot dispersion type is adopted, a dot pattern in which each dot is spatially separated is generated. For this reason, dot adjacency is unlikely to occur. Further, since the dot pattern has irregularity as well as regularity, a regular pattern such as Bayer and a warm pattern with a medium density such as a blue noise mask do not occur. Therefore, preferable halftone expression is possible, and the image quality can be improved.
[Brief description of the drawings]
FIG. 1 is a diagram showing a schematic configuration of an image display apparatus according to an embodiment.
FIG. 2 is a diagram illustrating the function of a dither threshold mask.
FIG. 3 is a diagram showing frequency characteristics of a dither threshold mask according to the present embodiment.
FIG. 4 is a diagram showing frequency characteristics of a dither threshold mask according to the present embodiment.
FIG. 5 is a diagram showing frequency characteristics of a dither threshold mask according to the present embodiment.
FIG. 6 is a diagram showing frequency characteristics of a dither threshold mask according to the present embodiment.
FIG. 7 is a diagram illustrating a halftone result using a Bayer-type mask.
FIG. 8 is a diagram illustrating a halftone result using a blue noise mask.
FIG. 9 is a diagram showing a halftone result by a dither threshold mask according to the present embodiment.
FIG. 10 is a flowchart of a first creation method of a dither threshold matrix according to the present embodiment.
FIG. 11 is a diagram showing a state of rotation conversion.
FIG. 12 is a diagram illustrating a threshold search order.
FIG. 13 is a diagram illustrating a relationship between a rotation angle and the number of searches.
FIG. 14A is a diagram showing a dot pattern based on a dither threshold mask when the rotation angle is 0 degree.
(B) It is a figure which shows the dot pattern by the dither threshold value mask in case a size is 4x4 and a rotation angle is 10 degree | times.
(C) It is a figure which shows the dot pattern by a dither threshold value mask in case a size is 4x4 and a rotation angle is 30 degree | times.
(D) It is a figure which shows the dot pattern by the dither threshold value mask in case a size is 132x162 and a rotation angle is 10 degree | times.
(E) It is a figure which shows the dot pattern by the dither threshold value mask in case a size is 132x162 and a rotation angle is 30 degree | times.
FIG. 15 is a diagram showing the results of a sensory evaluation experiment on a dot pattern for each rotation angle.
FIG. 16 is a matrix diagram for explaining binary dither.
FIG. 17 is a matrix diagram for explaining multilevel dither.
FIG. 18 is a block diagram of an apparatus for associating different dither matrices with R, G, and B colors.
FIG. 19 is a flowchart of a second generation method of a dither threshold mask.
FIG. 20 is a diagram illustrating rotation processing and boundary tracking processing in the second method for creating a dither threshold mask.
FIG. 21 is a diagram schematically illustrating a boundary point determination method in boundary tracking processing.
FIG. 22 is a diagram for explaining a threshold determination / storing method in boundary tracking processing;
FIG. 23 is a flowchart of a threshold determination / storing method in boundary tracking processing;
FIG. 24 is a flowchart of a third generation method of a dither threshold mask.
FIG. 25 is a diagram illustrating an example of a shift process.
FIG. 26 is a chart showing a relationship between a shift width and a corresponding rotation angle in the shift process.
FIG. 27 is a diagram illustrating a shift example in the shift process.
FIG. 28 is a diagram illustrating frequency characteristics of a Bayer-type mask.
FIG. 29 is a diagram illustrating frequency characteristics of a Bayer-type mask.
FIG. 30 is a diagram illustrating frequency characteristics of a Bayer-type mask.
FIG. 31 is a diagram illustrating frequency characteristics of a Bayer-type mask.
FIG. 32 is a diagram illustrating frequency characteristics of a blue noise mask.
FIG. 33 is a diagram illustrating frequency characteristics of a blue noise mask.
FIG. 34 is a diagram illustrating frequency characteristics of a blue noise mask.
FIG. 35 is a diagram illustrating frequency characteristics of a blue noise mask.
FIG. 36 is a diagram showing frequency characteristics of a mask disclosed in Japanese Patent Laid-Open No. 6-22123.
FIG. 37 is a diagram showing frequency characteristics of a mask disclosed in Japanese Patent Laid-Open No. 6-22123.
FIG. 38 is a diagram showing frequency characteristics of a mask disclosed in Japanese Patent Laid-Open No. 6-22123.
FIG. 39 is a diagram showing frequency characteristics of a mask disclosed in Japanese Patent Laid-Open No. 6-22123.
FIG. 40 is a diagram illustrating frequency characteristics of a dither threshold mask created using shift processing.
FIG. 41 is a diagram illustrating frequency characteristics of a dither threshold mask created using shift processing.
FIG. 42 is a diagram illustrating frequency characteristics of a dither threshold mask created using shift processing.
FIG. 43 is a diagram illustrating frequency characteristics of a dither threshold mask created using shift processing.
[Explanation of symbols]
100 Image processing apparatus
101 Input means
102 Dither processing means
103 LCD panel

Claims (7)

A method of creating a dot dispersion mask for converting a gradation value of each pixel in continuous tone image data into binary or multi-value,
Developing a basic threshold pattern in which thresholds are arranged with a predetermined regularity in a region of a primary matrix larger than the size of a mask to be created;
Relatively rotating each coordinate of the threshold value in the primary matrix by a predetermined rotation angle on the area of the primary matrix;
Placing the primary matrix thresholds at positions corresponding to the coordinates after the rotation so as not to overlap;
If the coordinates by the rotation are mapped to the same position as other coordinates that have been rotated, and another threshold that has already been rotated has been placed at the position, the coordinates of the thresholds that have not been placed in the vicinity of the position are And a step of searching and arranging the threshold value. A method of creating a dot dispersion mask for converting a gradation value of each pixel in continuous tone image data into binary or multi-value,
Developing a basic threshold pattern in which thresholds are arranged with a predetermined regularity in a region of a primary matrix larger than the size of a mask to be created;
Rotating one reference line segment in the primary matrix by a predetermined rotation angle, and storing each threshold on the primary matrix corresponding to the position of the reference line segment after rotation;
Determining the threshold value of the mask to be created by repeatedly acquiring and storing the threshold value from the primary matrix along the reference line after rotation, and determining a threshold value of the mask to be created. How to create   The threshold value acquired for the specific threshold position along the reference line segment includes a first line segment perpendicular to the rotated reference line segment, and the specific threshold value perpendicular to the first line segment. The dot dispersion mask creating method according to claim 2, wherein the dot dispersion mask is acquired based on a distance from an intersection with a second straight line passing through the position to the specific threshold position.   4. The method of creating a dot dispersion mask according to claim 1, wherein the predetermined rotation angle is in a range of 5 degrees to 20 degrees or 70 degrees to 85 degrees. A program for creating a dot-dispersed mask that is executed on a computer and converts the gradation value of each pixel in continuous-tone image data into binary or multivalued data,
Developing a basic threshold pattern in which thresholds are arranged with a predetermined regularity in a region of a primary matrix larger than the size of a mask to be created;
Relatively rotating each coordinate of the threshold value in the primary matrix by a predetermined rotation angle on the area of the primary matrix;
Placing the primary matrix thresholds at positions corresponding to the coordinates after the rotation so as not to overlap;
If the coordinates by the rotation are mapped to the same position as other coordinates that have been rotated and another threshold value that has already been rotated has been arranged at the position, the coordinates of the threshold value unallocated existing in the vicinity of the position are A program for creating a dot dispersion mask, which causes the computer to execute a step of searching and arranging the threshold value. A method of creating a dot-dispersed mask that is executed on a computer and converts a gradation value of each pixel in continuous-tone image data to binary or multi-value,
Developing a basic threshold pattern in which thresholds are arranged with a predetermined regularity in a region of a primary matrix larger than the size of a mask to be created;
Rotating one reference line segment in the primary matrix by a predetermined rotation angle, and storing each threshold on the primary matrix corresponding to the position of the reference line segment after rotation;
Determining the threshold value of the mask to be created by repeatedly acquiring and storing the threshold value from the primary matrix along the reference line segment after rotation, and causing the computer to execute Dot dispersion mask creation program. A computer-readable recording medium characterized by recording a creation program dot dispersion type mask according to claim 5 or 6.

Images