JP3847397B2 - Color correction method - Google Patents

Description

と定義すれば、この画素値ＰＰ^＃（ｎ，ｘ，ｙ）は、修正前の画素値ＰＰ（ｎ，ｘ，ｙ）と参照画素値ＰＰ^＊（ｎ，ｘ，ｙ）との、距離Ｄに基づく重みづけをした加重平均になり、この画素値ＰＰ^＃（ｎ，ｘ，ｙ）を修正後の画素値とする修正を行えば、境界線Ｍ，Ｍｐのいずれにおいても、色調が連続した画像が得られることになる。
【００６６】
たとえば、図１４において、Ｄ＝０の位置（境界線Ｍ上）については、ｒ（Ｄ）＝１になるため、ＰＰ^＃（ｎ，ｘ，ｙ）＝ＰＰ^＊（ｎ，ｘ，ｙ）となり、参照画素値ＰＰ^＊（ｎ，ｘ，ｙ）の値が１００％の重みづけをもつことになる。これに対し、Ｄ＝Ｗｐの位置（境界線Ｍｐ上）については、ｒ（Ｄ）＝０になるため、ＰＰ^＃（ｎ，ｘ，ｙ）＝ＰＰ（ｎ，ｘ，ｙ）となり、もとの画素値ＰＰ（ｎ，ｘ，ｙ）の値が１００％の重みづけをもつことになる。このように、距離Ｄの増加とともに、参照画素値ＰＰ^＊（ｎ，ｘ，ｙ）の重みづけは徐々に減少してゆくため、修正後の修正対象領域Ｐｏには、全体的になだらかな色調変化が得られることになり、上述した「フェードアウト処理」が実現できる。
【００６７】
なお、修正対象領域Ｐｏの幅Ｗｐとしては、「フェードアウト処理」による色調変化が肉眼で観察したときに不自然にならない程度の適当な幅を設定すればよい。第１のカラー画像Ｐのもつ本来の色調をできるだけ維持する意味では、幅Ｗｐをできるだけ狭く設定するのが好ましいが、あまり狭く設定すると、「フェードアウト処理」の効果が十分に発揮されず、色調変化が不自然になる。実際には、試行錯誤を何度か繰り返し、最適な幅Ｗｐを設定するようにするのが好ましい。本発明に係る色調修正方法は、幅Ｗｐ，Ｗｑなどのパラメータさえ設定すれば、コンピュータによって自動的に実行することが可能であるため、何度もやり直しを行うことができ、試行錯誤による繰り返し作業を容易に行うことができる。
【００６８】
一方、参照領域Ｑｒの幅Ｗｑとしては、図１４に示す実施形態では、１画素分の幅Ｗｑ＝１を設定している（最も小さな設定）。したがって、この例では、参照領域Ｑｒは、Ｘ座標値ｘ０に位置する画素Ｑ（ｎ，ｘ０，ｙ）を縦１列に並べた画素列によって構成されている。別言すれば、第２のカラー画像Ｑのうち、境界線Ｍに沿った１列の画素のみを参照して、修正対象領域Ｐｏの色調修正を行っていることになる。第２のカラー画像Ｑの境界線Ｍ側の色調分布がほぼ均一であれば、このように、参照領域Ｑｒの幅Ｗｑを１画素分に設定しても実用上は問題はない。
【００６９】
ところで、上述の式(6) は、第１のカラー画像Ｐの第ｎ番目のプレーン上の修正対象領域Ｐｏ内の座標（ｘ，ｙ）に位置する特定の画素Ｐ（ｎ，ｘ，ｙ）の修正後の画素値ＰＰ^＃（ｎ，ｘ，ｙ）を求める式であるが、一般に、修正対象領域Ｐｏ内の修正対象となる画素Ｐについての修正前の画素値をＰＰ、参照画素値をＰＰ^＊、重み関数をｒ（Ｄ）とすれば、修正対象となる画素Ｐについての修正後の画素値ＰＰ^＃は、
ＰＰ^＃＝ＰＰ^＊・ｒ（Ｄ）＋ＰＰ・（１−ｒ（Ｄ）） (7)
なる式で表されることになる。ここで、修正対象となる画素Ｐについてのヒストグラムの平均をＡｐ、分散をＶｐとし、参照領域Ｑｒに関するヒストグラムの平均をＡｑ、分散をＶｑとすれば、修正対象となる画素Ｐについての参照画素値ＰＰ^＊は、前述した式(4) によって、
ＰＰ^＊＝Ａｑ＋（Ｖｑ／Ｖｐ）・（ＰＰ−Ａｐ） (4)
と表されるので、ヒストグラムの平均および分散を用いて相対位置を定量的に定義する手法を用いるのであれば、上述の式(7) に式(4) を代入して、

なる式により、修正後の画素値ＰＰ^＃を演算することができる。
【００７０】
こうして、ステップＳ２４において、修正対象領域Ｐｏ上の全画素について、画素値の修正が行われたら、ステップＳ２５において、この修正後の第１のカラー画像を出力する画像出力段階が行われる。
【００７１】
§５． 空間的な色調分布を考慮に入れた画像接続用の修正方法
上述の§４で述べた方法では、修正対象領域Ｐｏ全体についてのヒストグラムと、参照領域Ｑｒ全体についてのヒストグラムとを利用して、色調修正を行っている。ところが、参照領域Ｑｒの空間的な色調分布が不均一の場合、この§４で述べた修正方法は必ずしも適当ではない。たとえば、図１１に示す例において、参照領域Ｑｒの上方部分はやや赤みがかった色調を帯びているのに対し、下方部分はやや黄色みがかった色調を帯びていた場合、この空間的な色調分布に関する情報は作成されたヒストグラムには何ら残っていないことになり、修正後の修正対象領域Ｐｏの色調は、参照領域Ｑｒの空間的な色調分布を反映したものにはならない。
【００７２】
そこで、空間的な色調分布を考慮に入れた色調修正を行いたい場合には、上述した§４の手法に、§２あるいは§３で述べた手法を加味すればよい。§２で述べた手法の基本的な考え方は、個々の画素ごとに局在的なヒストグラムを定義するということである。たとえば、図１４に示す例の場合、ある１つの画素Ｐ（ｎ，ｘ，ｙ）に着目し、この着目画素について所定の近傍領域を定義し、この近傍領域に関する限りのヒストグラム（別言すれば、この近傍領域内の画素を母集団とするヒストグラム）を作成し、このヒストグラムを着目画素Ｐ（ｎ，ｘ，ｙ）についての固有ヒストグラムと定義するのである。こうして、修正対象領域Ｐｏを構成するＮ組のプレーン上の個々の画素について、それぞれ固有ヒストグラムを定義する。同様に、参照領域Ｑｒを構成するＮ組のプレーン上の個々の画素Ｑ（ｎ，ｘ０，ｙ）についても、それぞれ固有ヒストグラムを定義する。
【００７３】
そして、修正段階では、修正対象領域Ｐｏ内の修正対象となる画素Ｐ（ｎ，ｘ，ｙ）についての修正後の画素値ＰＰ^＃（ｎ，ｘ，ｙ）を求める際に、参照領域Ｑｒ内で、画素Ｐ（ｎ，ｘ，ｙ）と位置的に対応する画素Ｑ（ｎ，ｘ０，ｙ）を参照画素として求め、この参照画素Ｑ（ｎ，ｘ０，ｙ）についてのヒストグラムを用いて、修正後の画素値ＰＰ^＃（ｎ，ｘ，ｙ）を求めるようにすればよい。なお、画素Ｐ（ｎ，ｘ，ｙ）と位置的に対応する画素Ｑ（ｎ，ｘ０，ｙ）としては、Ｙ座標値が同じ画素（すなわち、縦方向に関する位置が同じ画素）を選択するようにすればよい。
【００７４】
このような手法を取り込めば、参照領域Ｑｒの上方部分はやや赤みがかった色調を帯び、下方部分はやや黄色みがかった色調を帯びているという上述の例の場合にも、同じような色調分布が修正後の修正対象領域Ｐｏ側にも得られることになる。
【００７５】
なお、実用上は、§３で述べたブロック分割の考え方を取り込むのが好ましい。図１６は、修正対象領域Ｐｏを８個のブロックに、参照領域Ｑｒを４個のブロックに、それぞれ分割した例を示す図である。このようにブロックを定義したら、各ブロックごとにそれぞれ独立したヒストグラムを求める。ある１つのブロックのヒストグラムは、このブロックに所属する全画素を母集団としたヒストグラムとして定義されることになる。続いて、各ブロック内に代表位置をそれぞれ定義し（図１６の例では、各ブロックの中心点に代表位置が定義されている）、個々のブロックについて求められたヒストグラムの平均および分散を、それぞれ個々のブロックの代表位置についての色調平均および色調分散と定義する。そして、これら代表位置以外の各位置については、近傍に存在する複数の代表位置について定義された色調平均および色調分散に基づく補間により、それぞれ固有の色調平均および色調分散を定義するようにする。
【００７６】
一方、修正段階では、修正対象領域Ｐｏ内の修正対象となる画素Ｐについて、修正に利用する参照領域Ｑｒ内の位置的に対応する画素を参照画素Ｑとして求める。位置的に対応する参照画素Ｑとしては、たとえば、図１４に示す例のように、Ｙ座標値が同じで、Ｘ座標値がｘ０の位置にある画素（境界線Ｍ上の画素）を選ぶようにすればよい。そして、修正対象領域Ｐｏ内の修正対象となる画素Ｐについての修正前の画素値をＰＰ、この画素Ｐの位置について定義された色調平均をＡｐ、色調分散をＶｐとし、参照画素Ｑの位置について定義された色調平均をＡｑ、色調分散をＶｑとし、修正対象領域Ｐｏの幅をＷｐとし、画素Ｐと境界線Ｍとの距離をＤとしたときに、修正対象となる画素Ｐについての修正後の画素値ＰＰ^＃を、

（ただし、ｒ（Ｄ）は、Ｄが０からＷｐへ向かって単調増加するにしたがって、１から０へ単調減少する関数）なる演算によって求めるようにすればよい。
【００７７】
§６． 画像を一部重複させて接続するための色調修正方法
前述した§４および§５では、図１１に示すように、第１のカラー画像Ｐの右辺と第２のカラー画像Ｑの左辺とを、境界線Ｍ上で接続する場合の色調修正方法を述べた。ところが、木目柄パターンや石目柄パターンのような自然界の模様パタ−ンを接続する場合、一部を重複させて接続した方が、より自然な接続が可能になることが多い。図１７は、このような一部重複接続の概念図である。ここに示す例では、第１のパターンが第１の色調で表現された第１のカラー画像Ｐと、第２のパターンが第２の色調で表現された第２のカラー画像Ｐとが用意され、第１のカラー画像Ｐの一部分に第１の重複領域ＰＱ１を定義し、第２のカラー画像Ｑの一部分に第１の重複領域に対して合同な領域からなる第２の重複領域ＰＱ２を定義し、第１の重複領域ＰＱ１と第２の重複領域ＰＱ２とを重複させるようにして、第１のカラー画像Ｐと第２のカラー画像Ｑとの接続が行われる。そして、この接続の際に、色調の急激な変化を緩和するための色調修正が行われることになる。
【００７８】
はじめに、このような画像接続を行う場合の色調修正の基本方針を説明しよう。ここでは、図１７に示すように、第１のカラー画像Ｐ上に第１の重複領域ＰＱ１と第１の非重複領域Ｐｚとを定義し、第２のカラー画像Ｑ上に第２の重複領域ＰＱ２と第２の非重複領域Ｑｚとを定義する。第１の重複領域ＰＱ１と第２の重複領域ＰＱ２とは、平面的な重複が行われる領域であり、両者は幾何学的に合同な領域となる。なお、図１７では、説明の便宜上、第１のカラー画像Ｐと第２のカラー画像Ｑとを立体的に離して示してあるが、ここで述べる接続方法を行う上では、概念的には、両画像は同一平面上に重ねられることになる。また、第１の非重複領域Ｐｚは、第１のカラー画像Ｐ上の第１の重複領域ＰＱ１以外の領域として定義され、第２の非重複領域Ｑｚは、第２のカラー画像Ｑ上の第２の重複領域ＰＱ２以外の領域として定義される。ここでは、第１の重複領域ＰＱ１と第１の非重複領域Ｐｚとの境界を第１の境界線Ｍ１と呼ぶことにし（したがって、第２のカラー画像Ｑの図の左端辺も第１の境界線Ｍ１になる）、第２の重複領域ＰＱ２と第２の非重複領域Ｑｚとの境界を第２の境界線Ｍ２と呼ぶことにする（したがって、第１のカラー画像Ｐの図の右端辺も第２の境界線Ｍ２になる）。更に、第１の非重複領域Ｐｚ内の第１の境界線Ｍ１側の一部分に修正対象領域Ｐｏを定義し、第２の重複領域ＰＱ２内の第１の境界線Ｍ１側の一部分に参照領域Ｑｒを定義する。§４においても述べたように、修正対象領域Ｐｏの幅Ｗｐおよび参照領域Ｑｒの幅Ｗｑは適宜設定すればよい。
【００７９】
ここで述べる色調修正方法は、大きく分けて２つの修正段階から構成される。第１の修正段階は、第１の重複領域ＰＱ１内の画素についての画素値を修正する段階であり、第２の修正段階は、修正対象領域Ｐｏ内の画素についての画素値を修正する段階である。結局、この色調修正方法では、第１の重複領域ＰＱ１と修正対象領域Ｐｏとに対して修正が行われ、最終的には、修正後の第１のカラー画像Ｐに第２の非重複領域Ｑｚを接続した画像が得られることになる。第２の重複領域ＰＱ２上の画像は、修正作業を行う上でお手本として参照されることになるが、最終的に得られる接続画像の構成要素にはならない。別言すれば、最終的に得られる接続画像は、第１の非重複領域Ｐｚ上の画像（修正対象領域Ｐｏの部分について修正がなされている）と、第１の重複領域ＰＱ１上の画像（全領域に対して修正がなされている）と、第２の非重複領域Ｑｚ上の画像（修正はなされていない）と、をこの順に平面上に並べた画像ということになる。
【００８０】
第１の修正段階における修正、すなわち、第１の重複領域ＰＱ１に対する修正には、§１〜§３で述べた手法をそのまま利用すればよい。別言すれば、図１７における第１の重複領域ＰＱ１を図１における第１のカラー画像Ｐと考え、図１７における第２の重複領域ＰＱ２を図１における第２のカラー画像Ｑと考え、§１〜§３で述べた手順を実行すればよい。これにより、第１の重複領域ＰＱ１上の画像については、もとのパターンが表現された状態を維持しつつ、色調が第２の重複領域ＰＱ２上の画像の色調に近付くような色調修正がなされることになる。したがって、第２の境界線Ｍ２における色調の不連続は解消される。ただし、第１の境界線Ｍ１においては、色調の不連続が生じてしまうことになる。
【００８１】
第２の修正段階における修正は、この第１の境界線Ｍ１において新たに生じた色調の不連続を解消するための修正であり、修正対象領域Ｐｏに対する色調修正が、参照領域Ｑｒを参照して行われる。これは、§４，§５で述べた手法をそのまま利用すればよい。
【００８２】
図１８は、上述した基本方針に沿った色調修正方法の手順を示す流れ図である。以下、この流れ図に示す各手順を簡単に説明する。
【００８３】
まず、ステップＳ３１において、各画像を各色成分ごとに取り込む画像入力段階が行われる。すなわち、図３に示したように、第１のカラー画像Ｐおよび第２のカラー画像Ｑを、それぞれＮ組の色成分に分解し、各色成分ごとに、所定の画素値を有する画素の配列からなる単色プレーンを構成し、第１のカラー画像Ｐおよび第２のカラー画像Ｑを、それぞれＮ組の単色プレーンとして取り込む処理が実行される。
【００８４】
続くステップＳ３２では、第１のカラー画像Ｐを構成するＮ組の単色プレーンの第１の重複領域ＰＱ１内の部分について、各プレーンごとに、それぞれ画素値についてのヒストグラムを求める第１の演算段階が実行される。更に、ステップＳ３３では、第２のカラー画像Ｑを構成するＮ組の単色プレーンの第２の重複領域ＰＱ２内の部分について、各プレーンごとに、それぞれ画素値についてのヒストグラムを求める第２の演算段階が実行される。要するに、領域ＰＱ１についてのヒストグラムと、領域ＰＱ２についてのヒストグラムとが求められることになる。
【００８５】
次に、ステップＳ３４において、求めたヒストグラムを利用して、第１の画像Ｐの第１の重複領域ＰＱ１の画素値を修正する第１の修正段階が実行される。すなわち、第１のカラー画像Ｐを構成するＮ組の単色プレーンの第１の重複領域ＰＱ１内の部分について、各プレーンごとに、それぞれ第２のカラー画像Ｑ内の第２の重複領域ＰＱ２内の部分を参照して、個々の画素の画素値を修正する処理が行われることになる。具体的には、修正対象となる画素Ｐが有する画素値ＰＰのヒストグラム上での相対位置を求め、第２のカラー画像Ｑを構成するＮ組の単色プレーンの中から修正対象となる画素と同一の単色プレーンを選択し、この選択した単色プレーンの第２の重複領域ＰＱ２内の部分に関するヒストグラム上で同じ相対位置に存在する画素値ＰＰ^＊を求め、この求めた画素値ＰＰ^＊を修正後の画素値とする修正が行われる。
【００８６】
続くステップＳ３５では、第１のカラー画像Ｐ上の第１の非重複領域Ｐｚ内の、第１の境界線Ｍ１側に、所定の修正対象領域Ｐｏを定義し、第１のカラー画像Ｐを構成するＮ組の単色プレーンの修正対象領域Ｐｏ内の部分について、各プレーンごとに、それぞれ画素値についてのヒストグラムを求める第３の演算段階が実行される。更に、ステップＳ３６では、第２のカラー画像Ｑの第１の境界線Ｍ１側に所定の参照領域Ｑｒを定義し、第２のカラー画像Ｑを構成するＮ組の単色プレーンの参照領域Ｑｒ内の部分について、各プレーンごとに、それぞれ画素値についてのヒストグラムを求める第４の演算段階が実行される。もっとも、実際には、このステップＳ３６における第４の演算段階は、ステップＳ３３における第２の演算段階の演算結果を流用することが可能である。
【００８７】
次に、ステップＳ３７において、求めたヒストグラムを利用して参照画素値を求め、この参照画素値に境界線との距離に基づく重みづけを行った上で、修正対象領域の画素値を修正する第２の修正段階が実行される。すなわち、第１のカラー画像Ｐを構成するＮ組の単色プレーンの修正対象領域Ｐｏ内の部分について、各プレーンごとに、それぞれ第２のカラー画像Ｑ内の参照領域Ｑｒ内の部分を参照して、個々の画素の画素値を修正する処理が行われることになる。具体的には、修正対象となる画素Ｐが有する画素値ＰＰのヒストグラム上での相対位置を求め、第２のカラー画像Ｑを構成するＮ組の単色プレーンの中から修正対象となる画素と同一の単色プレーンを選択し、この選択した単色プレーンの参照領域Ｑｒ内の部分に関するヒストグラム上で同じ相対位置に存在する画素値ＰＰ^＊を参照画素値として求め、修正対象となる画素Ｐの修正前の画素値ＰＰと参照画素値ＰＰ^＊とに基づき、修正対象となる画素Ｐと第１の境界線Ｍ１との距離が小さくなるに従って参照画素値ＰＰ^＊の影響が大きくなるような重みづけをして新たな画素値ＰＰ^＃を決定し、この新たな画素値ＰＰ^＃を修正後の画素値とする修正が行われる。
【００８８】
かくして、第１の修正段階および第２の修正段階が完了すると、ステップＳ３８において、修正後の画素値をもったＮ組の単色プレーンから構成される第１のカラー画像Ｐが、修正後の画像として出力される。この修正後の第１のカラー画像Ｐに、第２のカラー画像Ｑの第２の非重複領域Ｑｚを接続すれば、目的とする接続画像が得られることになる。
【００８９】
なお、空間的な色調分布を考慮した色調修正を行う必要がある場合には、第１の修正段階においては、§２あるいは§３で述べた手法を加味するようにし、第２の修正段階においては、§５で述べた手法を加味するようにすればよい。すなわち、個々の画素について、それぞれ近傍領域についての固有ヒストグラムを定義したり、あるいは、ブロック分割を行って代表位置についての色調平均および色調分散を求め、更に補間法により任意の位置についての色調平均および色調分散を定義したりする手法を取り込むようにすればよい。
【００９０】
§７． 境界線として好ましい形状
最後に、画像接続を行う際の境界線として好ましい形状について述べておく。図１１に示す例では、第１のカラー画像Ｐと第２のカラー画像Ｑとを境界線Ｍを挟んで接続しており、ここでは、直線状の境界線Ｍが用いられている。しかしながら、実用上は、このような直線状の境界線Ｍを用いる代わりに、図１９に示すような波形形状の境界線ＭＭを用いるようにするのが好ましい。特に、第１のカラー画像Ｐ上のパターンと第２のカラー画像Ｑ上のパターンとが不連続な場合、上述した色調修正処理により、色調に関しては境界線における不連続が解消されるが、パターンに関しては境界線における不連続がそのまま残ることになる。一般に、このようなパターンに関する不連続は、色調に関する不連続に比べれば、人間の目では認識されにくい。しかしながら、直線状の境界線Ｍの代わりに、波形形状の境界線ＭＭを用いるようにすれば、人間の目による不連続の認識をより困難にする効果が得られる。
【００９１】
図１７に示したように、画像を一部重複させて接続する場合にも、波形の境界線を用いるようにするのが好ましい。この場合は、図２０に示すように、パターンの不連続が生じる可能性がある第２の境界線として、波形形状の境界線ＭＭ２を用いるようにすればよい。
【００９２】
【発明の効果】
以上のとおり、本発明に係る色調修正方法によれば、ヒストグラム上での相対位置が同じになるように画素値の修正を行うようにしたため、もともとのパターンが表現された状態を維持しつつ、色調だけを修正する処理を容易に行うことができるようになる。
【図面の簡単な説明】
【図１】本発明に係る色調修正方法に用いる２つのカラー画像を示す図である。
【図２】図１に示す第１のカラー画像Ｐに対して、第２のカラー画像Ｑの色調に近付ける色調修正を行うための基本手順を示す流れ図である。
【図３】図２に示す流れ図のステップＳ１１の画像入力段階によって用意された８組の単色プレーンを示す図である。
【図４】図３に示す単色プレーンのうちの１枚について求められたヒストグラムの一例を示すグラフである。
【図５】修正対象となる画素が有する画素値ＰＰについて、第１のカラー画像Ｐに関するヒストグラム上での相対位置を示すグラフである。
【図６】図５に示すヒストグラム上での相対位置と同じ相対位置にある画素値ＰＰ^＊を、第２のカラー画像Ｑに関するヒストグラム上で求めたグラフである。
【図７】各プレーン上の各画素ごとに固有のヒストグラムを求める手法を説明する概念図である。
【図８】各プレーンをブロック分割し、各ブロックごとにヒストグラムを求める手法を説明する概念図である。
【図９】図８に示す各ブロックの代表位置（中心位置）に、それぞれ格子点を定義し、各格子点に色調平均および色調分散を定義した状態を示す図である。
【図１０】図９に示す格子点を用いた補間法により、任意の点についての色調平均および色調分散を定義する手法を示す図である。
【図１１】第１のカラー画像Ｐと第２のカラー画像Ｑとを接続する際に、修正対象領域Ｐｏに対して色調修正を行う手法を示す概念図である。
【図１２】リピータブル模様を有する第１のカラー画像Ｐを多数配列して、大きな模様を構成する例を示す図である。
【図１３】図１１に示す修正対象領域Ｐｏに対する具体的な色調修正方法の手順を示す流れ図である。
【図１４】図１１に示す修正対象領域Ｐｏに対する具体的な色調修正を行う上で、境界線Ｍからの距離Ｄに基づく修正要素の重みづけ分布を示す図である。
【図１５】図１４に示す重みづけ分布に利用できる重み関数ｒ（Ｄ）の一例を示すグラフである。
【図１６】図１１に示す修正対象領域Ｐｏに対する色調修正を行う上でのブロック分割の手法を示す図である。
【図１７】第１のカラー画像Ｐと第２のカラー画像Ｑとを、一部重複させて接続する例を示す概念図である。
【図１８】図１７に示すような一部重複接続を行う際に実行する色調修正方法の手順を示す流れ図である。
【図１９】２つの画像を隣接させて接続する際に、波形形状の輪郭線ＭＭを用いた例を示す図である。
【図２０】２つの画像を一部重複させて接続する際に、波形形状の輪郭線ＭＭ２を用いた例を示す図である。
【符号の説明】
ａ〜ｄ…格子点
ｅ，ｆ…任意の点
Ａ，Ａｐ（ｎ），Ａｑ（ｎ）…色調平均（ヒストグラムの平均）
Ｂ１〜Ｂ２０…ブロック
Ｄ…境界線Ｍからの距離
Ｅｐ，Ｅｑ…近傍領域
Ｈ（ｎ，ｘ，ｙ）…代表位置（ブロックの中心位置）
Ｌｐ…画像Ｐに関するヒストグラム上での色調平均Ａｐ（ｎ）と特定の画素値ＰＰ（ｎ，ｘ，ｙ）との距離
Ｌｑ…画像Ｑに関するヒストグラム上での色調平均Ａｑ（ｎ）と特定の画素値ＰＰ^＊（ｎ，ｘ，ｙ）との距離
Ｍ，Ｍｐ，Ｍ１，Ｍ２…境界線
ＭＭ，ＭＭ２…波形形状の境界線
Ｐ…第１のカラー画像
Ｐｃ，Ｐｍ，Ｐｙ，Ｐｋ…第１のカラー画像Ｐを構成する各単色プレーン
Ｐｏ…修正対象領域
Ｐ（ｎ，ｘ，ｙ）…第１のカラー画像Ｐを構成する第ｎ番目の単色プレーン上の座標（ｘ，ｙ）に位置する画素
ＰＰ（ｎ，ｘ，ｙ）…画素Ｐ（ｎ，ｘ，ｙ）のもつ画素値
ＰＱ１…第１の重複領域
ＰＱ２…第２の重複領域
Ｐｚ…第１の非重複領域
Ｑ…第２のカラー画像
Ｑｃ，Ｑｍ，Ｑｙ，Ｑｋ…第２のカラー画像Ｑを構成する各単色プレーン
Ｑｒ…参照領域
Ｑ（ｎ，ｘ，ｙ）…第２のカラー画像Ｑを構成する第ｎ番目の単色プレーン上の座標（ｘ，ｙ）に位置する画素
ＱＱ（ｎ，ｘ，ｙ）…画素Ｑ（ｎ，ｘ，ｙ）のもつ画素値
Ｑｚ…第２の非重複領域
ｒ（Ｄ）…重み関数
Ｖ，Ｖｐ（ｎ），Ｖｑ（ｎ）…色調分散（ヒストグラムの分散）
Ｗｐ…修正対象領域Ｐｏの幅
Ｗｑ…参照領域Ｑｒの幅[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a color tone correction method, and more particularly to a method of correcting a color tone using a computer for a pattern such as a wood grain pattern prepared as digital image data.
[0002]
[Prior art]
In building materials such as wallpaper, flooring, and joinery, and cabinets for furniture and light electrical appliances, pattern patterns such as wood grain patterns and stone grain patterns are often used as surface decorations. These pattern patterns used as surface decorations are generally captured by a method such as photography from the surface of natural wood or natural stone. Usually, after being taken into a computer in the form of digital image data and subjected to necessary image processing, plate making processing using this digital image data is executed. If color correction is necessary for the captured image, the correction is performed as image processing on the computer.
[0003]
In the field of products such as building materials and furniture, the quality of design is a very important factor, and subtle differences in hues affect sales. Therefore, when a pattern pattern is printed as a surface decoration, it is very important to finely correct the color tone of the pattern pattern in consideration of the purchase layer of the product and the fashion. In addition, when capturing pattern patterns with relatively large areas from natural wood or natural stone, to deal with various technical problems, the pattern pattern can be divided into multiple parts without taking pictures all at once. In many cases, the images are divided and taken independently for each divided portion. In this case, a plurality of divided images captured by photography are connected to form the original large pattern. However, since it is difficult to set the shooting conditions for each time to be exactly the same, there is a case where the hue is slightly different for each photo, and the color tone is discontinuous at the connection portion of the divided images. Even in such a case, it is necessary to correct the color tone of the pattern pattern in order to alleviate the discontinuity of the connection portion. Or, when adopting the technique of constructing one large pattern by arranging many identical images vertically and horizontally, the so-called “repeatable pattern (the leftmost pattern is continuous with the rightmost pattern and the topmost pattern is the bottommost pattern However, even in such a case, it is necessary to correct the color tone in the vicinity of the contour line of the image.
[0004]
Such color correction work is generally performed using application software that can perform retouch processing on an image, and includes correction processing for gamma curves, correction processing for hue, saturation, and brightness, and blurring using a brush. Various methods such as processing and gradation processing are appropriately performed.
[0005]
[Problems to be solved by the invention]
However, the above-described color tone correction work by the retouching process depends on the sensitivity of the operator, and a satisfactory result can only be obtained by a skilled operator. Moreover, even for skilled operators, this color correction work requires a lot of labor and time, which is very expensive and not suitable for mass production. It was. In addition, when a finally satisfied result is not obtained, it is very difficult to redo the work, and it is also difficult to adopt a so-called trial and error process.
[0006]
SUMMARY An advantage of some aspects of the invention is that it provides a color tone correction method capable of easily correcting a desired color tone.
[0007]
[Means for Solving the Problems]
  (1) In the first aspect of the present invention, the first color image in which the first pattern is expressed in the first color tone and the second color image in which the second pattern is expressed in the second color tone In a color tone correction method for correcting the color tone so as to approach the second color tone while maintaining the state in which the first pattern is expressed for the first color image,
  The first color image and the second color image are each divided into N sets of color components, and for each color component, a single-color plane composed of an array of pixels having a predetermined pixel value is formed. And an image input stage that captures each of the second color images as N sets of monochromatic planes;
  For the N sets of monochromatic planes constituting the first color image, a first calculation stage for obtaining a histogram for each pixel value for each plane;
  A second calculation stage for obtaining a histogram for each pixel value for each of the N sets of monochromatic planes constituting the second color image;
  For the N sets of single color planes constituting the first color image, the pixels included in the pixel to be corrected in order to correct the pixel value of each pixel with reference to the second color image for each plane. The relative position of the value on the histogram is obtained, and the same single color plane as the pixel to be corrected is selected from the N sets of single color planes constituting the second color image, and the selected single color plane is displayed on the histogram. A correction stage for obtaining a pixel value existing at the same relative position, and making the obtained pixel value a corrected pixel value,
  An image output stage for outputting a first color image composed of N sets of monochromatic planes having corrected pixel values as a corrected image;
  And
In the first calculation stage, the first color image is divided into a plurality of blocks, and an independent histogram is obtained for each block;
In the second calculation stage, the second color image is divided into a plurality of blocks, and an independent histogram is obtained for each block;
Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
For each position other than the representative position, a unique tone average and tone dispersion are defined by interpolation based on the tone average and tone variance defined for a plurality of representative positions existing in the vicinity,
At the correction stage, the pixel value before correction for the pixel P to be corrected in the first color image is PP, the tone average defined for the position of this pixel P is Ap, and the tone variance is Vp, which is used for correction. The tone average defined for the same position of the second color image is Aq, the tone variance is Vq, and the corrected pixel value PP for the pixel P to be corrected ^* The
PP ^* = Aq + (Vq / Vp). (PP-Ap)
Is obtained by the following calculation.
[0011]
  (2)  Of the present inventionSecondIn the aspect of the present invention, a first color image in which the first pattern is expressed in the first color tone and a second color image in which the second pattern is expressed in the second color tone are prepared, and a predetermined boundary is prepared. In a color correction method for correcting a color tone to alleviate a sudden change in color tone at the boundary line when connecting the first color image and the second color image so as to sandwich the line,
  The first color image and the second color image are each divided into N sets of color components, and for each color component, a single-color plane composed of an array of pixels having a predetermined pixel value is formed. And an image input stage that captures each of the second color images as N sets of monochromatic planes;
  A predetermined correction target area is defined on the boundary line side of the first color image, and the pixel value is determined for each plane for each portion in the correction target area of the N sets of monochromatic planes constituting the first color image. A first computation stage for obtaining a histogram of
  A predetermined reference area is defined on the boundary line side of the second color image, and a histogram of pixel values is obtained for each plane in a portion in the reference area of the N sets of monochromatic planes constituting the second color image. A second computation stage for obtaining
  With respect to the portions in the correction target region of the N sets of single color planes constituting the first color image, each pixel is referred to by referring to the portion in the reference region in the second color image for each plane. In order to correct the value, the relative position on the histogram of the pixel value of the pixel to be corrected is obtained, and the same value as the pixel to be corrected is selected from the N sets of single color planes constituting the second color image. Select a single-color plane, obtain the pixel value that exists at the same relative position on the histogram for the portion in the reference area of the selected single-color plane as the reference pixel value, and the pixel value before correction and the reference pixel of the pixel to be corrected Based on the value, a new pixel value is determined by weighting so that the influence of the reference pixel value increases as the distance between the pixel to be corrected and the boundary line decreases. A modified step of correcting the pixel value of the corrected values,
  An image output stage for outputting a first color image composed of N sets of monochromatic planes having corrected pixel values as a corrected image;
  And
In the first calculation stage, the correction target area in the first color image is divided into a plurality of blocks, and an independent histogram is obtained for each block,
In the second calculation stage, the reference region in the second color image is divided into a plurality of blocks, and an independent histogram is obtained for each block.
Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
For each position other than the representative position, a unique tone average and tone dispersion are defined by interpolation based on the tone average and tone variance defined for a plurality of representative positions existing in the vicinity,
In the correction stage, the pixel value before correction for the pixel P to be corrected in the correction target region is PP, the tone average defined for the position of the pixel P is Ap, and the tone variance is Vp. The pixel corresponding to the position in the area is determined as a reference pixel, the tone average defined for the position of the reference pixel is Aq, the tone variance is Vq, the width of the correction target area is Wp, and the pixel P to be corrected When the distance between the boundary line and the boundary line is D, the corrected pixel value PP for the pixel P to be corrected ^# The
PP ^# = Aq + (Vq / Vp). (PP-Ap) .r (D)
+ PP · (1-r (D))
Here, r (D) is obtained by an operation of D (monotonically decreasing from 1 to 0 as D increases monotonously from 0 to Wp).
[0015]
  (3)  Of the present inventionThirdIn this aspect, a first color image in which the first pattern is expressed in the first color tone and a second color image in which the second pattern is expressed in the second color tone are prepared. Defining a first overlapping region in a part of the color image, defining a second overlapping region consisting of regions congruent to the first overlapping region in a part of the second color image, In a color correction method for correcting a color tone to alleviate a sudden change in color tone when connecting the first color image and the second color image so as to overlap the second overlapping region,
  The first color image and the second color image are each divided into N sets of color components, and for each color component, a single-color plane composed of an array of pixels having a predetermined pixel value is formed. And an image input stage that captures each of the second color images as N sets of monochromatic planes;
  A first calculation step for obtaining a histogram for each pixel value for each plane for the portions in the first overlapping area of the N sets of monochromatic planes constituting the first color image;
  A second calculation step for obtaining a histogram for each pixel value for each plane for the portions in the second overlapping area of the N sets of monochromatic planes constituting the second color image;
  With respect to the portions in the first overlapping region of the N sets of single color planes constituting the first color image, for each plane, refer to the portion in the second overlapping region in the second color image, respectively. In order to correct the pixel value of each pixel, the relative position on the histogram of the pixel value of the pixel to be corrected is obtained, and the correction target is selected from the N sets of single color planes constituting the second color image. The same monochromatic plane as the pixel to be selected is selected, the pixel value existing at the same relative position on the histogram relating to the portion in the second overlapping area of the selected monochromatic plane is obtained, and the pixel value obtained by correcting the obtained pixel value A first correction stage for correcting the value;
  N sets of single colors constituting a first color image by defining a predetermined correction target area on the boundary line side of the second color image in an area other than the first overlapping area on the first color image A third calculation step for obtaining a histogram for each pixel value for each plane for a portion in the correction target area of the plane;
  A predetermined reference area is defined on the boundary line side of the second color image, and a histogram of pixel values is obtained for each plane in a portion in the reference area of the N sets of monochromatic planes constituting the second color image. A fourth computation stage for obtaining
  With respect to the portions in the correction target region of the N sets of single color planes constituting the first color image, each pixel is referred to by referring to the portion in the reference region in the second color image for each plane. In order to correct the value, the relative position on the histogram of the pixel value of the pixel to be corrected is obtained, and the same value as the pixel to be corrected is selected from the N sets of single color planes constituting the second color image. Select a single-color plane, obtain the pixel value that exists at the same relative position on the histogram for the portion in the reference area of the selected single-color plane as the reference pixel value, and the pixel value before correction and the reference pixel of the pixel to be corrected Based on the value, a new pixel value is determined by weighting so that the influence of the reference pixel value increases as the distance between the pixel to be corrected and the boundary line decreases. A second correction step of correcting the pixel value of the corrected values,
  An image output stage for outputting a first color image composed of N sets of monochromatic planes having corrected pixel values as a corrected image;
  Is to do.
[0016]
  (Four)  Of the present invention4thAspects of the aboveThirdIn the color correction method according to the aspect of
  When obtaining histograms in the first computation stage and the second computation stage, a predetermined neighborhood region is defined for each pixel, and a histogram relating to the neighborhood region of a certain pixel of interest is defined as the histogram for this pixel of interest. By defining a unique histogram for each pixel,
  In the first correction stage, when the corrected pixel value for the pixel to be corrected in the first overlap area is obtained, the position corresponding to the pixel to be corrected is positioned in the second overlap area. A pixel is obtained as a reference pixel, and a corrected pixel value for a pixel to be corrected is obtained using a histogram for the reference pixel.
[0017]
  (Five)  Of the present invention5thAspects of the aboveThirdOr4thIn the color correction method according to the aspect of
  In the first correction stage, the pixel value before correction for the pixel P to be corrected in the first overlapping region is set to PP, the average of the histogram for this pixel P is Ap, and the variance is Vp, which are used for correction. The average of the histogram relating to the second overlapping region is Aq, the variance is Vq, and the corrected pixel value PP for the pixel P to be corrected^*The
      PP^*= Aq + (Vq / Vp). (PP-Ap)
Is obtained by the following calculation.
[0018]
  (6)  Of the present invention6thAspects of the aboveThirdIn the color correction method according to the aspect of
  Dividing each of the first overlap region and the second overlap region into a plurality of blocks, and obtaining an independent histogram for each block;
  Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
  For each position other than the representative position, a unique tone average and tone dispersion are defined by interpolation based on the tone average and tone variance defined for a plurality of representative positions existing in the vicinity,
  In the first correction stage, PP is the pixel value before correction for the pixel P to be corrected in the first overlap region, Ap is the tone average defined for the position of this pixel P, and Vp is the tone variance. A tone average defined for the same position of the second overlapping region used for correction is Aq, a tone variance is Vq, and a corrected pixel value PP for the pixel P to be corrected^*The
      PP^*= Aq + (Vq / Vp). (PP-Ap)
Is obtained by the following calculation.
[0019]
  (7)  Of the present invention7thAspects of the aboveThirdIn the color correction method according to the aspect of
  When obtaining histograms in the third calculation stage and the fourth calculation stage, a predetermined neighborhood region is defined for each pixel, and a histogram relating to the neighborhood region of a certain pixel of interest is defined as the histogram for this pixel of interest. By defining a unique histogram for each pixel,
  In the second correction stage, when obtaining a corrected pixel value for a pixel to be corrected in the correction target region, a pixel corresponding to the pixel to be corrected in the reference region is used as a reference pixel. The pixel value after correction for the pixel to be corrected is calculated using the histogram for the reference pixel.
[0020]
  (8)  Of the present invention8thAspects of the aboveThirdOr7thIn the color correction method according to the aspect of
  In the second correction stage, the pixel value before correction for the pixel P to be corrected in the correction target area is PP, the average of the histogram for this pixel P is Ap, and the variance is Vp. When the average of the histogram is Aq, the variance is Vq, the width of the correction target region is Wp, and the distance between the correction target pixel P and the boundary line is D, the corrected pixel P is corrected. Pixel value PP^#The
      PP^#= Aq + (Vq / Vp). (PP-Ap) .r (D)
            + PP · (1-r (D))
Here, r (D) is obtained by an operation of D (monotonically decreasing from 1 to 0 as D increases monotonously from 0 to Wp).
[0021]
  (9)  Of the present invention9thAspects of the aboveThirdIn the color correction method according to the aspect of
  The correction target area in the first color image and the reference area in the second color image are each divided into a plurality of blocks, and an independent histogram is obtained for each block,
  Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
  For each position other than the representative position, a unique tone average and tone dispersion are defined by interpolation based on the tone average and tone variance defined for a plurality of representative positions existing in the vicinity,
  In the second correction stage, the pixel value before correction for the pixel P to be corrected in the correction target region is set to PP, the tone average defined for the position of the pixel P is Ap, and the tone distribution is Vp. The pixel corresponding to the position in the reference area to be used is obtained as a reference pixel, the tone average defined for the position of the reference pixel is Aq, the tone variance is Vq, the width of the correction target area is Wp, and the correction target When the distance between the pixel P and the boundary line is D, the corrected pixel value PP for the pixel P to be corrected^#The
      PP^#= Aq + (Vq / Vp). (PP-Ap) .r (D)
            + PP · (1-r (D))
Here, r (D) is obtained by an operation of D (monotonically decreasing from 1 to 0 as D increases monotonously from 0 to Wp).
[0022]
  (Ten)  Of the present invention10thThe aspect of the above-mentioned3rd to 9thIn the color tone correction method according to the above aspect, the same repeatable image is used as the first color image and the second color image.
[0024]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the present invention will be described based on the illustrated embodiments.
[0025]
§1. Basic concept of color correction method according to the present invention
Here, as shown in FIG. 1, a correction method for correcting the color tone so that the color tone of the first color image P approaches the color tone of the second color image Q will be described. In the first color image P, the first pattern (grain pattern) is expressed in the first color tone, and in the second color image Q, the second pattern (another grain pattern pattern) is represented. ) Is expressed in the second color tone. In this example, each color image represents a wood grain pattern, but the patterns actually expressed are completely different from each other, and the color tone is also different. The object of color tone correction performed here is the first color image P, and the second color image Q is only used as an example of color tone. Therefore, this color tone correction process is a process for replacing the pixel values of the individual pixels constituting the first color image P with new pixel values.
[0026]
The important point here is that the correction is applied to the first color image P, which is the object to be corrected. This correction is only for the purpose of correction related to the color tone, and is related to the pattern pattern. It is not intended for correction. In other words, the first color image P is corrected so that the color tone approaches the second color tone while maintaining the state where the first pattern is expressed. Such correction requests are particularly found in the field of products such as building materials and furniture. For example, consider a case where a first color image P having a first wood grain pattern and a second color image Q having a second wood grain pattern are extracted from different natural trees by photography. In consideration of the taste and fashion of a particular furniture purchaser, it is preferable to use the first pattern appearing in the first color image P as the woodgrain pattern, but as the hue, the second pattern Suppose that the designer has concluded that the second color tone appearing in the color image Q is preferred. In such a case, if the color tone correction method according to the present invention is used, the first color image P can be corrected so that only the color tone approaches the color tone of the second color image Q.
[0027]
FIG. 2 is a flowchart showing the basic procedure of the method for performing the correction as described above. This procedure is actually executed as image processing using a computer. First, in step S11, an image input stage for capturing each image for each color component is performed. That is, the first color image P and the second color image Q shown in FIG. 1 are each divided into N sets of color components, and for each color component, a single color plane composed of an array of pixels having a predetermined pixel value is formed. The first color image P and the second color image Q are respectively captured as N sets of single color planes. In the field of printing, a color image is generally decomposed into four color components of C (cyan), M (magenta), Y (yellow), and K (black). Here, N = 4. An example of decomposition into four sets of color components will be described. Of course, it may be decomposed into three color components of C, M, and Y, or may be decomposed into three color components of R (red), G (green), and B (blue).
[0028]
FIG. 3 is a diagram showing eight sets of monochromatic planes prepared in this image input stage. For the first color image P, four sets of single color planes Pc, Pm, Py, Pk are prepared, and for the second color image Q, four sets of single color planes Qc, Qm, Qy, Qk are prepared. Yes. Here, an array of pixels arranged vertically and horizontally is defined on each single color plane, and each pixel has a specific pixel value (for example, when one pixel value is expressed by 8-bit data). , Values from 0 to 255) are defined. Here, a two-dimensional XY coordinate system is defined on each monochromatic plane, and a pixel defined at the position of the coordinate (x, y) on the nth monochromatic plane for the first color image P is defined as a pixel P ( n, x, y), and the pixel value defined for the pixel P (n, x, y) is called the pixel value PP (n, x, y). Similarly, a pixel defined at the position of the coordinate (x, y) on the n-th monochromatic plane for the second color image Q is referred to as a pixel Q (n, x, y). A pixel value defined for Q (n, x, y) will be referred to as a pixel value QQ (n, x, y).
[0029]
As described above, various methods are already known as specific methods for fetching each color image into the computer as a set of pixel data on N sets of monochrome planes. For example, if an image on a color photograph or an image on a film is captured by a scanner device or directly taken using a digital camera or the like, image data of a plurality of single-color planes as shown in FIG. 3 can be prepared.
[0030]
Subsequently, in step S12, for the N sets of single color planes constituting the first color image P, a first calculation step for obtaining a histogram for each pixel value is executed for each plane. For example, for the nth monochrome plane of the first color image P shown in FIG. 3, a histogram for each pixel value PP (n, x, y) is obtained. FIG. 4 is a graph showing an example of a histogram for one plane obtained in this way. In this example, the horizontal axis indicates the pixel value (for example, 0 to 255), and the vertical axis indicates the frequency (how many pixels having a specific pixel value exist on one plane). If such a histogram is obtained, the average (here, referred to as tone average A) and variance (here, tone variance V) can be obtained by calculation as necessary. Similarly, in step S13, for the N sets of single-color planes constituting the second color image Q, a second calculation step is performed for obtaining a histogram for each pixel value for each plane. As a result, in the example shown here, four histograms are obtained for the first color image P and four histograms are obtained for the second color image Q, for a total of eight histograms.
[0031]
In the next step S14, a correction stage for correcting the pixel value of the first color image P is performed using the obtained histogram. That is, with respect to the four sets of single color planes Pc, Pm, Py, Pk constituting the first color image P, each pixel P (n, x, A process of correcting the pixel value PP (n, x, y) of y) is performed. The main points of this correction processing are as follows. Here, a specific pixel P (n, x, y) on the nth monochromatic plane constituting the first color image P is set as a pixel to be corrected, and a pixel value PP ( n, x, y) as the new pixel value PP^*A specific procedure for correcting to (n, x, y) will be described.
[0032]
First, the relative position on the histogram of the pixel value PP (n, x, y) of the pixel P (n, x, y) to be corrected is obtained. For example, a histogram as shown in FIG. 5 is obtained for the nth plane of the first color image P by the processing in step S12 described above, and the pixel P (n, x, y) to be corrected is obtained. Suppose that the pixel value PP (n, x, y) of is at “a position slightly to the right of the center” on this histogram, as shown in the figure. Of course, in an actual calculation, an ambiguous expression such as “a position slightly to the right of the center” cannot be used to indicate the relative position, so it is necessary to take some quantitative expression. In this embodiment, a method of expressing the relative position quantitatively using the average Ap (n) and the variance Vp (n) of the histogram is adopted. That is, the distance Lp between the pixel value PP (n, x, y) and the average Ap (n) is obtained (Lp = (PP (n, x, y) −Ap (n)), and the variance Vp ( The relative position on the histogram of the pixel value PP (n, x, y) is defined by the ratio “Lp / Vp (n)” to n. The variance Vp (n) is the horizontal direction of the entire histogram. The relative position on the histogram can be defined by defining the ratio of Lp to the variance Vp (n).
[0033]
Next, the same single color plane as the pixel P (n, x, y) to be corrected is selected from the N sets of single color planes constituting the second color image Q, and the histogram on the selected single color plane is displayed. To associate pixel values existing at the same relative position. For example, in the case of the above-described example, the pixel P (n, x, y) to be corrected is a pixel on the nth plane, so that N sets of single color planes constituting the second color image Q are included. The same nth plane is selected from among them. Here, it is assumed that the histogram relating to the selected n-th plane is as shown in FIG. The average Aq (n) and variance Vq (n) of the histogram shown in FIG. 6 are usually different from the average Ap (n) and variance Vp (n) of the histogram shown in FIG. The basic concept of the present invention is to associate pixel values having the same relative position for both histograms. For example, the pixel value PP (n, x, y) in the histogram shown in FIG. 5 is the pixel value PP in the histogram shown in FIG.^*(N, x, y) will be associated. Thus, the pixel value PP^*When (n, x, y) is obtained, the pixel value PP (n, x, y) of the pixel P (n, x, y) to be corrected on the original first color image P is newly set. Pixel value PP^*It is corrected to (n, x, y).
[0034]
If such correction processing is executed for all the pixels on the nth single color plane of the first color image P, and is further executed for all N sets of single color planes, the processing in step S14 is completed. . When such correction processing is performed, it is possible to correct only the color tone to the color tone of the second color image Q while maintaining the first pattern expressed on the first color image P. . That is, even if the pixel value is corrected, the relative position on the histogram of the pixel value for each pixel does not change, so that the relationship of shading between adjacent pixels is maintained as it is. The expression of the pattern is maintained as it is. For example, in the state before correction, when there is a relationship that the pixel P1 is dark (pixel value is large) and the pixel P2 is light (pixel value is small), this light / dark relationship is maintained as it is even after correction. become. Thus, although the relative position on the histogram of each pixel value does not change, since the absolute value of the pixel value changes, only the color tone is corrected.
[0035]
The pixel value PP (n, x, y) before correction shown in FIG. 5 and the pixel value PP after correction shown in FIG.^*Comparing (n, x, y), it can be understood that the absolute values of the pixel values are completely different, but the relative positions on the histogram are the same. Note that, in order to prevent misunderstanding, the pixel value PP (n, x, y) of the pixel P (n, x, y) on the nth plane of the first color image P will be described. Is corrected, the value of the pixel value QQ (n, x, y) of the pixel Q (n, x, z) on the nth plane of the second color image Q is not directly related. In other words, when the pixel value of the pixel located at the coordinate (x, y) of the first color image P is corrected, the pixel located at the same coordinate (x, y) of the second color image Q is corrected. The pixel value is not directly referenced at all. To the last, pixel values at the same relative position on the histogram are referred to, and pixel values at the same spatial position are not referred to.
[0036]
Now, as conceptually described above, in the present invention, since the pixel values at the same relative position on the histogram are obtained, the correction at “a position slightly to the right of the center” on the histogram shown in FIG. With respect to the previous pixel value PP (n, x, y), the pixel value PP at “a position slightly to the right of the center” on the histogram shown in FIG.^*What is necessary is just to obtain | require (n, x, y) as a pixel value after correction. However, in actual calculations, it is necessary to perform more quantitative processing. As described above, in this embodiment, a method of expressing the relative position quantitatively using the average and variance of the histogram is adopted, and the pixel value PP (n, x on the histogram shown in FIG. 5 is used. , Y) is quantitatively expressed by a value “Lp / Vp (n)”. Therefore, the pixel value PP on the histogram shown in FIG.^*The relative position of (n, x, y) may also be a position quantitatively expressed by the same value “Lp / Vp (n)”. After all, the average of the histogram shown in FIG. 6 is Aq (n), the variance is Vq (n), and the pixel value PP^*The distance from the average Aq (n) of (n, x, y) is expressed as Lq (Lq = PP^*(N, x, y) −Aq (n))
Lp / Vp (n) = Lq / Vq (n) (1)
It is only necessary that the distance Lq such that the following formula holds can be obtained. Therefore, the distance Lq is
Lq = (Vq (n) / Vp (n)) · Lp (2)
Can be obtained by the following calculation. Here, since Lp = PP (n, x, y) −Ap (n),
PP^*(N, x, y) = Aq (n) + (Vq (n) / Vp (n))
・ (PP (n, x, y) -Ap (n)) (3)
The corrected pixel value PP^*It can be seen that (n, x, y) can be uniquely determined.
[0037]
This expression (3) is a corrected pixel value PP of a specific pixel P (n, x, y) located at coordinates (x, y) on the nth plane of the first color image P.^*(N, x, y) is an expression for calculating the pixel value before correction for the pixel P to be corrected in the first color image P, and the histogram average for this pixel P is generally Ap. If the variance is Vp, the average of the histogram relating to the second color image Q used for correction is Aq, and the variance is Vq, the corrected pixel value PP for the pixel P to be corrected^*Is
PP^*= Aq + (Vq / Vp). (PP-Ap) (4)
Can be obtained by the following calculation.
[0038]
Thus, when the pixel values are corrected for all pixels on all N sets of planes of the first color image P in step S14, the corrected first color image P is determined in step S15.^*The image output stage for outputting is performed. This image output may be performed in a format suitable for the subsequent process. For example, if an original film is created based on this image data, an image may be output on the film for each of the C, M, Y, and K plates, and a direct printing plate may be created based on this image data. If the data is to be created, the digital data may be output in the data format required by the printing plate creating apparatus.
[0039]
The above-mentioned quantitative treatment using the mean and variance of the pixel values is theoretically based on the assumption that the histogram has a normal distribution, but in practice it is applied to histograms other than the normal distribution. However, no problem occurs. In other words, when color correction by the above-described calculation is performed on a general image that has a histogram other than the normal distribution, the tone relationship between each pixel in the original pattern cannot be strictly maintained. However, practically, when observing with the human eye, a correction result that can be recognized as maintaining almost the same pattern is obtained, so that no problem occurs.
[0040]
§2. Basic correction method taking spatial color distribution into account
In the above §1, one embodiment for describing the basic concept of the color tone correction method according to the present invention has been described. The embodiment of §1 has no problem when the spatial color distribution of the second color image Q is uniform. However, when the spatial color distribution of the second color image Q is non-uniform, there arises a problem that it is not possible to perform color correction that directly incorporates the non-uniform color distribution. Let us explain this phenomenon with a more specific example.
[0041]
For example, in the example shown in FIG. 1, when the color tone correction is performed on the first color image P, the second color image Q is referred to and only the color tone is maintained while maintaining the pattern of the first color image P. A process of approaching the second color image Q is performed. However, in the embodiment described in §1, a histogram is created for the entire second color image Q, and this histogram is used as a reference. Therefore, information on the local tone distribution of the second color image Q is Nothing will be reflected. For example, suppose that the right side portion of the second color image Q has a slightly reddish tone, while the left side portion has a slightly yellowish tone. Thus, even if the spatial tone distribution of the second color image Q is non-uniform, in the embodiment of §1, since the histogram for the entire second color image Q is created, redness is reduced. No information about the spatial tone distribution, such as the tinged and yellowish parts, will remain in the created histogram, which shows the overall orange tone. become. Therefore, in the embodiment described in §1, the corrected first color image P^*The color tone is generally orange, and the overall color tone is close to the color tone of the second color image Q, but the right side is slightly reddish and the left side is slightly yellowish. The spatial tone distribution of the second color image Q, such as being tinged, is not reflected. In other words, the spatial tone distribution is smoothed.
[0042]
Of course, if it is preferable to generate an image having an overall uniform tone distribution ignoring such a spatial tone distribution, the above-described embodiment of §1 may be carried out as it is. However, when it is desired to perform color correction that reflects the spatial color distribution, the embodiment of §1 cannot be implemented as it is. In §2, an embodiment of a color correction method that takes into account such a spatial color distribution will be described. According to this embodiment, the corrected first color image P^*In this case, a non-uniform color distribution in which the right side portion has a slightly reddish tone and the left side portion has a slightly yellowish tone is obtained, and the spatial color tone of the second color image Q is obtained. The distribution will be reflected.
[0043]
The basic idea for performing tone correction in consideration of such tone distribution is to define a local histogram for each pixel. In step S12 or step S13 in the flowchart shown in FIG. 2, one histogram is created for each single color plane. Therefore, the histogram shown in FIG. 5 or 6 is a histogram for all pixels spatially distributed in the entire image region. On the other hand, in the embodiment described here, a predetermined neighborhood region is defined for each pixel, and a histogram as far as the neighborhood region of a certain pixel of interest is defined as a histogram for this pixel of interest. A unique histogram is obtained for each of the pixels.
[0044]
FIG. 7 is a conceptual diagram illustrating a method for obtaining a unique histogram for each pixel in this way. For example, the unique histogram for a specific pixel P (n, x, y) located at the coordinates (x, y) on the nth monochrome plane of the first color image P is obtained as follows. Can do. First, a predetermined neighborhood region Ep (n, x, y) is defined for the pixel P (n, x, y). For example, a circle having a predetermined radius with the coordinate (x, y) as the center may be defined as the neighborhood region Ep (n, x, y). Then, a histogram of pixel values is obtained for a plurality of pixels belonging to the neighborhood region Ep (n, x, y), this is defined as a unique histogram for the pixel P (n, x, y), and this The average and variance for the intrinsic histogram are defined as tone average Ap (n, x, y) and tone variance Vp (n, x, y) for pixel P (n, x, y). If such a definition is made for all the pixels for each plane, an independent unique histogram is defined for each pixel of each plane, and independent tone average and tone dispersion are defined. Exactly the same processing is performed for the second color image Q. That is, for each pixel Q (n, x, y) on the monochromatic plane constituting the second color image Q, a predetermined neighboring region Eq (n, x, y) is defined, and this neighboring region Eq (n, A histogram of pixel values is obtained for a plurality of pixels belonging to x, y), this is defined as the eigen histogram for the pixel Q (n, x, y), and the mean and variance for this eigen histogram are It is defined as a color tone average Aq (n, x, y) and a color tone variance Vq (n, x, y) for the pixel Q (n, x, y).
[0045]
On the other hand, in the correction stage of step S14 in the procedure shown in FIG. 2, the corrected pixel value PP for the pixel P (n, x, y) to be corrected in the first color image P.^*When obtaining (n, x, y), a pixel Q (n, x, y) corresponding to the pixel to be corrected in the second color image Q is obtained as a reference pixel, and Pixel value PP after correction using the inherent histogram for the pixel P (n, x, y) and the inherent histogram for the reference pixel Q (n, x, y)^*(N, x, y) may be obtained.
[0046]
For example, if the pixel P (n, x, y) in the first color image P shown in FIG. 7 is the correction target, the pixel Q (( n, x, y) are reference pixels. Subsequently, for the pixel value PP (n, x, y) of the correction target pixel P (n, x, y), the relative position on the inherent histogram of the pixel P (n, x, y) is obtained. In short, this relative position indicates how much relative position the pixel value PP (n, x, y) occupies when pixels belonging to the neighboring region Ep (n, x, y) are used as a population. It is shown and expressed quantitatively using the tone average Ap (n, x, y) and the tone variance Vp (n, x, y). And the same relative position is shown on the intrinsic | native histogram (Histogram when the pixel which belongs in the vicinity area | region Eq (n, x, y) is made into a population) about reference pixel Q (n, x, y). Pixel value PP^*(N, x, y) is obtained as the corrected pixel.
[0047]
According to the method as described above, it is possible to perform color correction in consideration of even the spatial color distribution, and the spatial color distribution on the second color image Q is the first color after correction. Image P^*It will be reflected above.
[0048]
§3. Practical correction method considering spatial color distribution
According to the above-described §2 embodiment, it is possible to perform color correction in consideration of the spatial color distribution. However, since it is necessary to define a neighborhood region for each individual pixel and obtain a unique histogram, the computational burden on the computer becomes very large. In particular, when an image with a high resolution is used, the total number of pixels is enormous, and it is not always an efficient method if the calculation burden is taken into consideration. Considering that the corrected image is ultimately observed by the human eye, it is not very meaningful to reproduce the spatial tone distribution to the resolution of individual pixels that cannot be recognized by the naked eye. It's not a method. Therefore, here, a practical method using the basic concept described in §2 will be described.
[0049]
Now, as shown in FIG. 8, the first color image P and the second color image Q are each divided into a plurality of blocks. In the illustrated example, the image is divided into four parts in the vertical direction and five parts in the horizontal direction, for a total of 20 parts, and 20 blocks B1 to B20 are defined. In step S12 and step S13 in the procedure of FIG. 2, an independent histogram is obtained for each individual block, and in the correction stage in step S14, the histogram for the block to which the pixel to be corrected belongs is used. Try it. For example, when the pixel belonging to the block B1 on the nth plane of the first color image P is the pixel to be corrected, the histogram for the block B1 (that is, all the pixels belonging to the block B1) The relative position of the pixel value of the pixel to be corrected is obtained on the histogram), and the same relative on the histogram for the same block B1 on the nth plane of the second color image Q. The pixel value indicating the position may be the corrected pixel value.
[0050]
If such a method is adopted, the calculation for obtaining the histogram does not need to be performed for each individual pixel, but may be performed for each individual block. In addition, since different histograms are used for each block, the color distribution is not smoothed over the entire image area, and the local color distribution for each block area is reflected. This method can be said to be an embodiment in which the block region is defined as a common neighborhood region for pixels belonging to the same block in the method described in §2. That is, for any pixel in the block B1, if the block B1 is defined as a neighborhood region, this method is one of the categories described in §2.
[0051]
However, the method as it is causes a new problem that the change in color tone becomes discontinuous at the boundary line of each block. For example, the histogram used for correcting the color tone of the pixels in the block B1 and the histogram used for correcting the color tone of the pixels in the block B2 adjacent thereto are separate histograms. Image P^*In this case, the change in color tone becomes discontinuous at the boundary line between the block B1 and the block B2. Eventually, when observed with the naked eye, the boundary line of each block will be recognized as a discontinuous line of color tone change.
[0052]
In order to avoid such an adverse effect, the present embodiment is devised as follows. First, as shown in FIG. 9, a representative position H is defined in each block. In the illustrated example, the representative position H is defined at the center point of each block. However, the representative position H is not necessarily defined as the center point. For example, the upper left corner of a rectangle that forms the boundary line of the block Can be defined as a representative position H. Then, the average and variance of the histogram obtained for each block are defined as the tone average and tone variance for the representative position H of each block, respectively. For example, assume that the coordinates of the representative position H of the block B18 on the nth plane of the first color image P shown in FIG. 9 are (x, y). In this case, the tone average Ap (n, x, y) and tone variance Vp (n, x, y) defined for this representative position H (n, x, y) are the histogram (ie, block) for the block B18. That is, tone average and tone dispersion for a histogram (having all pixels belonging to B18 as a population). Similarly, for the representative position H (n, x, y) of the block B18 on the nth plane of the second color image Q, the tone average Aq (n, x, y) and the tone distribution Vq (n, x, y) is defined.
[0053]
Eventually, a total of 20 representative positions H are defined for each plane, and for each of the representative positions H, the tone average and the tone dispersion are defined. Now, assuming that grid points are arranged at individual representative positions H, as shown in FIG. 10, grid points are arranged at regular intervals on each plane, and each grid point is unique to each grid point. Color tone average and color tone variance are defined. For example, in the first color image P shown in FIG. 10, each of the lattice points a to d arranged at the representative position H has a specific tone average and tone variance defined, but the point e at an arbitrary position. There is no such definition. Therefore, by using an interpolation method, a unique color tone average and color tone dispersion are defined for any point other than the grid points. That is, for an arbitrary point e, interpolation based on the tone average and tone dispersion defined for a plurality of grid points a to d existing in the vicinity thereof is performed to define the inherent tone average and tone dispersion. As described above, since various methods are known for defining a continuous scalar field based on a discrete scalar field, a detailed description of the specific interpolation method is omitted here. In short, any interpolation method may be used as long as a continuous scalar field whose values gradually change can be defined so that no discontinuity occurs in the scalar values. For example, if the tone average for an arbitrary point e shown in the first color image P of FIG. 10 is defined, the four tone average values defined at the four surrounding grid points a to d are defined. , A weighted average obtained by weighting according to the distance (the closer the grid points, the larger the weighting) may be obtained. The same applies to the interpolation for the color tone dispersion.
[0054]
When such an interpolation method is applied, a unique tone average and tone dispersion value can be obtained as an interpolation value at any point other than the lattice points. Moreover, these interpolated values are spatially continuous values, and no discontinuity occurs over the entire image area. That is, at the stage where the tone average and tone variance are defined by this interpolation method, the tone average and tone variance are no longer discontinuous at the block boundary lines.
[0055]
Thus, when color tone correction is performed on the pixels in the first color image P, the color tone average and the color tone dispersion obtained by the interpolation method may be used. For example, when a pixel located at an arbitrary point e in the first color image P in FIG. 10 is a correction target, first, a point f at the same position in the second color image Q is obtained. Then, the corrected pixel value is obtained using the tone average and tone variance defined for the point e and the tone average and tone variance defined for the point f. This can be expressed as a specific arithmetic expression as follows. That is, the pixel value before correction for the pixel P to be corrected in the first color image P is PP, the tone average defined for the position of this pixel P is Ap, and the tone variance is Vp, which is used for correction. If the tone average defined for the same position of the second color image Q is Aq and the tone variance is Vq, the corrected pixel value PP for the pixel P to be corrected^*Is
PP^*= Aq + (Vq / Vp). (PP-Ap)
Can be obtained by the following calculation.
[0056]
§4. Basic concept of color correction method for connecting images
When capturing a pattern with a relatively large area by photographing, the entire image is divided into a plurality of parts and captured as independent image data, and these are interconnected in a computer to obtain an image of the original large pattern Approach is taken. For example, the example shown in FIG. 11 is an example in which a horizontally long pattern is divided into left and right and separately captured as a first color image P and a second color image Q, respectively. When the two divided images captured in the computer are connected by the boundary line M, it is relatively easy to match the patterns of both images. In other words, if accurate alignment is performed with a resolution in pixel units, a connection that does not cause a pattern shift in the boundary line M is possible. However, matching the color tones of both images is very difficult with conventional techniques. Since it is difficult to set the shooting conditions for capturing the first color image P and the shooting conditions for capturing the second color image Q to exactly the same conditions, there is a subtle difference between the two images. A difference in hue will occur. Therefore, even if both images are directly connected in the computer, discontinuity of the color tone at the boundary line M cannot be avoided.
[0057]
In such a case, the color tone correction method according to the present invention is a very effective method. Specifically, a correction target region Po having a predetermined width Wp is defined on the boundary line M side of the first color image P, and a predetermined width Wq is defined on the boundary line M side of the second color image Q. By defining the reference area Qr and correcting the color tone of the correction target area Po so as to approach the color tone of the reference area Qr, the discontinuity of the color tone at the boundary line M is eliminated. It is also possible to set the predetermined width Wp to the entire horizontal width of the first color image P and set the entire first color image P as the correction target region Po. Similarly, the predetermined width Wq is set to the second color. It is also possible to set the entire width of the image Q so that the entire second color image Q is used as the reference region Qr. However, as the predetermined width Wp is increased, the area to be subjected to color tone correction increases, and the original color tone of the first color image P is impaired. Further, regarding the predetermined width Wq, it is sufficient that the color tone is matched with the area near the boundary line M. Therefore, it is meaningless to increase the predetermined width Wq unnecessarily. Therefore, in practice, it is preferable to set only a partial region on the boundary line M side as the correction target region Po and the reference region Qr.
[0058]
In addition, when using a so-called “repeatable pattern”, it is necessary to perform an image connection process in a computer, resulting in a problem of discontinuity in the color tone at the boundary line. For example, FIG. 12 shows an example in which a large pattern is formed by arranging nine first color images P vertically and horizontally. Such a technique is often used when it is necessary to apply a pattern pattern to a building material with a larger area than the pattern pattern area of a wood grain pattern obtained from natural wood. . In this case, the connection target is the upper side and lower side, and the right side and left side of the first color image P itself. Therefore, the first color image P constitutes a so-called “repeatable pattern” in which the pattern on the upper side and the pattern on the lower side are continuous, and the pattern on the right side and the pattern on the left side are continuous. There must be. In general, the grain pattern itself taken from natural wood or the like does not constitute a “repeatable pattern”, and therefore it is necessary to correct the image so that it becomes a “repeatable pattern” after being taken into a computer. The present invention can also be used for color tone correction in this case (various methods are known for correcting the pattern itself, and are not described here because they are out of the scope of the present invention). ).
[0059]
FIG. 13 shows a predetermined boundary between a first color image P in which the first pattern is expressed in the first color tone and a second color image Q in which the second pattern is expressed in the second color tone. 6 is a flowchart showing a procedure of the present invention for correcting a color tone to alleviate a sudden change in a color tone at a boundary line M when connecting with a line M interposed therebetween. Hereinafter, this procedure will be described with reference to a specific example shown in FIG. As described above, this color tone correction method can also be applied to the color tone correction in the peripheral portion of the so-called “repeatable pattern”. Therefore, in the example illustrated in FIG. 11, for convenience of explanation, the two images to be connected are referred to as the first color image P and the second color image Q, but are distinguished from each other. Are not necessarily separate images, and they may be the same image.
[0060]
The correction process shown in the flowchart of FIG. 13 is a process of correcting the pixel value of the pixel included in the correction target area Po shown in FIG. 11, and the reference area is used as a model for correcting the color tone. The pixel value of the pixel in Qr is used. The principle of color tone correction for the correction target area Po is almost the same as the principle described in §1. That is, a first histogram relating to the pixel value of the pixel in the correction target area Po is obtained, and a second histogram relating to the pixel value of the pixel in the reference area Qr is obtained. Then, a relative position on the first histogram is obtained for the pixel value PP of the specific pixel P in the correction target region Po, and the pixel value PP occupying the same relative position on the second histogram.^*Is used to correct the original pixel value PP. However, if the method described in §1 is applied as it is, inconvenience occurs. The reason is as follows.
[0061]
Now, the correction target area Po in FIG. 11 is considered as the first color image P in FIG. 1, the reference area Qr in FIG. 11 is considered as the second color image Q in FIG. Let's consider the case where is applied as it is. Then, by this correction processing, the correction target area Po is subjected to color correction so that only the color tone approaches the reference area Qr while the original pattern pattern expression is maintained as it is. As a result, the discontinuity of the color tone at the boundary line M is eliminated (that is, the color tone of the corrected region Po is substantially the same as the color tone of the reference region Qr). However, when such color tone correction is performed, a new problem of color tone discontinuity at the boundary line Mp on the left side of the correction target region Po occurs. That is, since the color correction is performed equally on the entire correction target area Po, on the first color image P side, a sudden change in the color tone occurs at the boundary line Mp between the correction target area Po and the other areas. Will occur. In order to prevent this, the color correction is not performed equally on the entire correction target region Po, but a stronger correction is performed on pixels closer to the boundary line M, and a weaker correction is performed on pixels farther from the boundary line M. It is necessary to make corrections in consideration of the weights. In this specification, the tone correction processing taking such weights into account is referred to as “fade-out processing”. In the following, the processing procedure shown in the flowchart of FIG.
[0062]
First, in step S21, an image input stage for capturing each image for each color component is performed. In the embodiment described here, as shown in FIG. 3, N sets of monochrome planes are formed in the computer for each image.
[0063]
In step S22, a correction target area Po having a predetermined width Wp is defined on the boundary line M side of the first color image P, and correction targets in N sets of single color planes constituting the first color image P are defined. For the portion in the region Po, a first calculation step for obtaining a histogram for each pixel value is executed for each plane. Similarly, in step S23, a reference area Qr having a predetermined width Wq is defined on the boundary line M side of the second color image Q, and the reference areas in N sets of monochromatic planes constituting the second color image Q are defined. For the portion in Qr, a second calculation step for obtaining a histogram for each pixel value is executed for each plane.
[0064]
In the next step S24, a correction process using the obtained histogram is executed. Here, the above-described “fade-out process” is executed. Specifically, the following method is adopted. First, on the histogram of the pixel value PP (n, x, y) of the pixel P (n, x, y) to be corrected in the correction target area Po in the nth plane of the first color image P. Find the relative position of. Next, the nth monochrome plane identical to the pixel P (n, x, y) to be corrected is selected from N sets of monochrome planes constituting the second color image Q, and the selected monochrome color is selected. Pixel values PP existing at the same relative position on the histogram relating to the reference region Qr in the plane^*Find (n, x, y). In the correction method described in §1, this pixel value PP^*(N, x, y) is defined as the corrected pixel value, and the original pixel value PP (n, x, y) is defined as the pixel value PP.^*Processing to replace with (n, x, y) was performed. However, in this step S24, this pixel value PP^*(N, x, y) is treated as a reference pixel value, and this reference pixel value PP^*Pixel value PP corresponding to a weighted average of (n, x, y) and original pixel value PP (n, x, y) before correction of the pixel to be corrected^#(N, x, y) is obtained, and this pixel value PP^#The correction is performed with (n, x, y) as a new pixel value of the pixel P (n, x, y) to be corrected. Here, as shown in FIG. 14, the weighting in obtaining the weighted average is the distance D (0 ≦ D ≦ Wp) between the pixel P (n, x, y) to be corrected and the boundary line M. Define the reference pixel value PP as D becomes smaller^*Weighting may be performed so that the influence of (n, x, y) becomes large.
[0065]
In this embodiment, a specific weight function r (D) as shown in the graph of FIG. 15 is defined to perform this weighting. This weight function r (D) is
r (D) = 1/2 · (cos (π · D / Wp) +1) (5)
R (D) is a function that monotonically decreases from 1 to 0 as D increases monotonically from 0 to Wp, as defined in the equation and shown in the graph. Of course, as the weighting function r (D), it is possible to use not only the function shown in the above equation (5) but also another function having the same property. Using such a weight function r (D), the pixel value PP^#(N, x, y)

This pixel value PP^#(N, x, y) is a pixel value PP (n, x, y) before correction and a reference pixel value PP.^*(N, x, y) is a weighted average obtained by weighting based on the distance D, and this pixel value PP^#If correction is performed with (n, x, y) as the corrected pixel value, an image having a continuous color tone can be obtained at both the boundary lines M and Mp.
[0066]
For example, in FIG. 14, for the position of D = 0 (on the boundary line M), r (D) = 1, so PP^#(N, x, y) = PP^*(N, x, y) and the reference pixel value PP^*The value of (n, x, y) will have a weight of 100%. On the other hand, for the position of D = Wp (on the boundary line Mp), r (D) = 0, so PP^#(N, x, y) = PP (n, x, y), and the value of the original pixel value PP (n, x, y) has a weight of 100%. Thus, as the distance D increases, the reference pixel value PP^*Since the weight of (n, x, y) gradually decreases, a gentle color change can be obtained as a whole in the correction target area Po after correction, and the above-described “fade-out process” is performed. realizable.
[0067]
Note that the width Wp of the correction target region Po may be set to an appropriate width that does not cause unnaturalness when the color change due to the “fade-out process” is observed with the naked eye. In order to maintain the original color tone of the first color image P as much as possible, it is preferable to set the width Wp as narrow as possible. However, if the width Wp is set too small, the effect of the “fade-out process” is not sufficiently exhibited and the color tone changes. Becomes unnatural. Actually, it is preferable to repeat trial and error several times to set the optimum width Wp. Since the color tone correction method according to the present invention can be automatically executed by a computer as long as parameters such as widths Wp and Wq are set, it can be performed again and again, and repeated work by trial and error. Can be easily performed.
[0068]
On the other hand, as the width Wq of the reference region Qr, in the embodiment shown in FIG. 14, the width Wq = 1 for one pixel is set (the smallest setting). Therefore, in this example, the reference region Qr is configured by a pixel column in which the pixels Q (n, x0, y) located at the X coordinate value x0 are arranged in one vertical column. In other words, the color correction of the correction target region Po is performed with reference to only one column of pixels along the boundary line M in the second color image Q. If the tone distribution on the boundary line M side of the second color image Q is substantially uniform, there is no practical problem even if the width Wq of the reference region Qr is set to one pixel.
[0069]
By the way, the above-described formula (6) is expressed by the specific pixel P (n, x, y) located at the coordinates (x, y) in the correction target area Po on the nth plane of the first color image P. Pixel value PP after correction^#Although (n, x, y) is an expression, generally, the pixel value before correction and the reference pixel value PP for the pixel P to be corrected in the correction target area Po are PP.^*If the weight function is r (D), the corrected pixel value PP for the pixel P to be corrected^#Is
PP^#= PP^*・ R (D) + PP ・ (1-r (D)) (7)
It will be expressed by the following formula. Here, if the average of the histogram for the pixel P to be corrected is Ap, the variance is Vp, the average of the histogram for the reference region Qr is Aq, and the variance is Vq, the reference pixel value for the pixel P to be corrected PP^*Is given by equation (4) above,
PP^*= Aq + (Vq / Vp). (PP-Ap) (4)
Therefore, if you use the method of quantitatively defining the relative position using the mean and variance of the histogram, substitute equation (4) into equation (7) above,

The corrected pixel value PP^#Can be calculated.
[0070]
Thus, when the pixel values are corrected for all the pixels on the correction target area Po in step S24, an image output stage for outputting the corrected first color image is performed in step S25.
[0071]
§5. Correction method for image connection taking into account spatial color distribution
In the method described in §4 above, color correction is performed using the histogram for the entire correction target region Po and the histogram for the entire reference region Qr. However, when the spatial tone distribution of the reference area Qr is not uniform, the correction method described in section 4 is not necessarily appropriate. For example, in the example shown in FIG. 11, when the upper part of the reference region Qr has a slightly reddish color tone, the lower part has a slightly yellowish color tone. No information remains in the created histogram, and the color tone of the correction target area Po after correction does not reflect the spatial color distribution of the reference area Qr.
[0072]
Therefore, when it is desired to perform color correction in consideration of the spatial color distribution, the method described in §2 or §3 may be added to the method described in §4. The basic idea of the method described in §2 is that a local histogram is defined for each pixel. For example, in the example shown in FIG. 14, focusing on one pixel P (n, x, y), a predetermined neighborhood region is defined for this pixel of interest, and a histogram (in other words, as far as this neighborhood region is concerned). A histogram with the pixels in this neighboring region as a population is created, and this histogram is defined as a unique histogram for the pixel of interest P (n, x, y). In this way, a unique histogram is defined for each pixel on the N sets of planes constituting the correction target region Po. Similarly, a unique histogram is defined for each pixel Q (n, x0, y) on the N sets of planes constituting the reference region Qr.
[0073]
In the correction stage, the corrected pixel value PP for the pixel P (n, x, y) to be corrected in the correction target area Po.^#When obtaining (n, x, y), the pixel Q (n, x0, y) corresponding to the position of the pixel P (n, x, y) in the reference region Qr is obtained as a reference pixel. The corrected pixel value PP using the histogram for the pixel Q (n, x0, y)^#(N, x, y) may be obtained. Note that pixels having the same Y coordinate value (that is, pixels having the same position in the vertical direction) are selected as the pixels Q (n, x0, y) that correspond in position to the pixel P (n, x, y). You can do it.
[0074]
If such a method is taken in, the above-described example in which the upper part of the reference region Qr has a slightly reddish tone and the lower part has a slightly yellowish tone has a similar tone distribution. It is also obtained on the correction target area Po side after correction.
[0075]
In practice, it is preferable to incorporate the concept of block division described in §3. FIG. 16 is a diagram illustrating an example in which the correction target area Po is divided into 8 blocks and the reference area Qr is divided into 4 blocks. When blocks are defined in this way, an independent histogram is obtained for each block. The histogram of a certain block is defined as a histogram with all pixels belonging to this block as a population. Subsequently, a representative position is defined in each block (in the example of FIG. 16, a representative position is defined at the center point of each block), and the average and variance of the histogram obtained for each block are respectively expressed. It is defined as tone average and tone variance for the representative position of each block. For each of the positions other than the representative positions, unique tone average and tone dispersion are defined by interpolation based on the tone average and tone dispersion defined for a plurality of representative positions existing in the vicinity.
[0076]
On the other hand, at the correction stage, the pixel corresponding to the position in the reference area Qr used for correction is obtained as the reference pixel Q for the pixel P to be corrected in the correction target area Po. As the reference pixel Q corresponding to the position, for example, as in the example shown in FIG. 14, a pixel (pixel on the boundary line M) having the same Y coordinate value and an X coordinate value of x0 is selected. You can do it. The pixel value before correction for the pixel P to be corrected in the correction target region Po is PP, the tone average defined for the position of the pixel P is Ap, the tone variance is Vp, and the position of the reference pixel Q is When the defined tone average is Aq, the tone variance is Vq, the width of the correction target region Po is Wp, and the distance between the pixel P and the boundary line M is D, the corrected pixel P is corrected Pixel value PP of^#The

(However, r (D) may be obtained by an operation of D (monotonically decreasing from 1 to 0 as D increases monotonously from 0 to Wp).
[0077]
§6. Color tone correction method to connect with overlapping images partly
In the above-mentioned §4 and §5, as shown in FIG. 11, a tone correction method in the case where the right side of the first color image P and the left side of the second color image Q are connected on the boundary line M is described. It was. However, when connecting a natural pattern such as a wood grain pattern or a stone pattern, it is often possible to make a more natural connection by overlapping a part of the patterns. FIG. 17 is a conceptual diagram of such partially overlapping connection. In the example shown here, a first color image P in which the first pattern is expressed in the first color tone and a second color image P in which the second pattern is expressed in the second color tone are prepared. The first overlapping area PQ1 is defined in a part of the first color image P, and the second overlapping area PQ2 is defined in a part of the second color image Q. The second overlapping area PQ2 is an area congruent to the first overlapping area. Then, the first color image P and the second color image Q are connected so that the first overlap region PQ1 and the second overlap region PQ2 overlap each other. At the time of this connection, color correction is performed to alleviate a sudden change in color tone.
[0078]
First, let's explain the basic policy for color correction when making such image connections. Here, as shown in FIG. 17, a first overlapping area PQ1 and a first non-overlapping area Pz are defined on the first color image P, and a second overlapping area is defined on the second color image Q. PQ2 and the second non-overlapping region Qz are defined. The first overlapping region PQ1 and the second overlapping region PQ2 are regions where planar overlap is performed, and both are geometrically congruent regions. In FIG. 17, for convenience of explanation, the first color image P and the second color image Q are illustrated as being three-dimensionally separated. However, in performing the connection method described here, conceptually, Both images are superimposed on the same plane. The first non-overlapping region Pz is defined as a region other than the first overlapping region PQ1 on the first color image P, and the second non-overlapping region Qz is the second non-overlapping region Qz on the second color image Q. 2 is defined as an area other than the overlapping area PQ2. Here, the boundary between the first overlapping region PQ1 and the first non-overlapping region Pz is referred to as a first boundary line M1 (therefore, the left end side of the second color image Q is also the first boundary). The boundary between the second overlapping region PQ2 and the second non-overlapping region Qz will be referred to as a second boundary line M2 (therefore also the right edge of the first color image P in the figure). The second boundary line M2). Further, the correction target region Po is defined in a part on the first boundary line M1 side in the first non-overlapping region Pz, and the reference region Qr is defined in a part on the first boundary line M1 side in the second overlapping region PQ2. Define As described in §4, the width Wp of the correction target region Po and the width Wq of the reference region Qr may be set as appropriate.
[0079]
The color tone correction method described here is roughly divided into two correction stages. The first correction stage is a stage for correcting the pixel values for the pixels in the first overlapping area PQ1, and the second correction stage is a stage for correcting the pixel values for the pixels in the correction target area Po. is there. After all, in this color tone correction method, the first overlapping area PQ1 and the correction target area Po are corrected, and finally the second non-overlapping area Qz is added to the corrected first color image P. Will be obtained. The image on the second overlapping area PQ2 is referred to as a model for performing the correction work, but does not become a constituent element of the finally obtained connection image. In other words, the connection image finally obtained is an image on the first non-overlapping region Pz (a portion of the correction target region Po has been corrected) and an image on the first overlapping region PQ1 ( That is, all the areas are corrected) and the image on the second non-overlapping area Qz (not corrected) are arranged in this order on a plane.
[0080]
For the correction in the first correction stage, that is, the correction for the first overlapping area PQ1, the method described in §1 to §3 may be used as it is. In other words, the first overlapping area PQ1 in FIG. 17 is considered as the first color image P in FIG. 1, the second overlapping area PQ2 in FIG. 17 is considered as the second color image Q in FIG. What is necessary is just to perform the procedure described in 1-3. As a result, for the image on the first overlapping area PQ1, the color tone is corrected so that the color tone approaches the color tone of the image on the second overlapping area PQ2, while maintaining the state in which the original pattern is expressed. Will be. Therefore, the discontinuity of the color tone at the second boundary line M2 is eliminated. However, in the first boundary line M1, discontinuity of the color tone occurs.
[0081]
The correction in the second correction stage is a correction for eliminating the discontinuity of the color tone newly generated in the first boundary line M1, and the color correction for the correction target region Po is performed with reference to the reference region Qr. Done. For this purpose, the method described in §4 and §5 may be used as it is.
[0082]
FIG. 18 is a flowchart showing the procedure of the color tone correction method according to the basic policy described above. Hereinafter, each procedure shown in this flowchart will be briefly described.
[0083]
First, in step S31, an image input stage for capturing each image for each color component is performed. That is, as shown in FIG. 3, the first color image P and the second color image Q are each divided into N sets of color components, and for each color component, from an array of pixels having a predetermined pixel value. A process for capturing the first color image P and the second color image Q as N sets of single color planes is executed.
[0084]
In the subsequent step S32, a first calculation step for obtaining a histogram for each pixel value for each plane in the first overlapping area PQ1 of the N sets of single color planes constituting the first color image P is performed. Executed. Further, in step S33, a second calculation stage for obtaining a histogram for each pixel value for each plane in the second overlapping area PQ2 of the N sets of single color planes constituting the second color image Q. Is executed. In short, a histogram for the region PQ1 and a histogram for the region PQ2 are obtained.
[0085]
Next, in step S34, a first correction stage for correcting the pixel value of the first overlapping region PQ1 of the first image P using the obtained histogram is executed. That is, for each portion in the first overlapping region PQ1 of the N sets of single color planes constituting the first color image P, each of the planes in the second overlapping region PQ2 in the second color image Q is included. A process of correcting the pixel value of each pixel is performed with reference to the portion. Specifically, the relative position on the histogram of the pixel value PP of the pixel P to be corrected is obtained, and is the same as the pixel to be corrected from among N sets of single color planes constituting the second color image Q. Pixel values PP existing at the same relative position on the histogram relating to the portion in the second overlapping area PQ2 of the selected monochrome plane.^*And the obtained pixel value PP^*The correction is performed with the corrected pixel value.
[0086]
In the subsequent step S35, a predetermined correction target region Po is defined on the first boundary line M1 side in the first non-overlapping region Pz on the first color image P, and the first color image P is formed. For the portion in the correction target area Po of the N sets of single color planes to be performed, a third calculation step for obtaining a histogram for each pixel value is executed for each plane. Further, in step S36, a predetermined reference area Qr is defined on the first boundary line M1 side of the second color image Q, and the N sets of monochrome planes constituting the second color image Q are within the reference area Qr. For the portion, a fourth calculation step is performed for obtaining a histogram for each pixel value for each plane. However, in practice, the fourth calculation stage in step S36 can use the calculation result of the second calculation stage in step S33.
[0087]
Next, in step S37, a reference pixel value is obtained using the obtained histogram, and the reference pixel value is weighted based on the distance from the boundary line, and then the pixel value of the correction target region is corrected. Two correction steps are performed. That is, with respect to the portions in the correction target region Po of the N sets of single color planes constituting the first color image P, the respective portions in the reference region Qr in the second color image Q are referred to for each plane. A process of correcting the pixel value of each pixel is performed. Specifically, the relative position on the histogram of the pixel value PP of the pixel P to be corrected is obtained, and is the same as the pixel to be corrected from among N sets of single color planes constituting the second color image Q. Pixel values PP existing at the same relative position on the histogram relating to the portion in the reference area Qr of the selected monochrome plane.^*As a reference pixel value, and the pixel value PP before correction and the reference pixel value PP of the pixel P to be corrected^*And the reference pixel value PP as the distance between the pixel P to be corrected and the first boundary line M1 decreases.^*The new pixel value PP is weighted so that the influence of^#And determine the new pixel value PP^#The correction is performed with the corrected pixel value.
[0088]
Thus, when the first correction stage and the second correction stage are completed, in step S38, the first color image P composed of N sets of single color planes having the corrected pixel values is converted into the corrected image. Is output as If the second non-overlapping area Qz of the second color image Q is connected to the corrected first color image P, a target connection image can be obtained.
[0089]
In addition, when it is necessary to perform color correction in consideration of spatial color distribution, in the first correction stage, the method described in §2 or §3 is added, and in the second correction stage. Should take into account the method described in §5. That is, for each pixel, a unique histogram for each neighboring region is defined, or block division is performed to obtain a tone average and tone variance for the representative position, and further, a tone average and a tone average for an arbitrary position are obtained by interpolation. A method for defining tone dispersion may be incorporated.
[0090]
§7. Preferred shape as a boundary line
Finally, a preferable shape as a boundary line when performing image connection will be described. In the example shown in FIG. 11, the first color image P and the second color image Q are connected with the boundary line M interposed therebetween, and here, the linear boundary line M is used. However, in practice, it is preferable to use a waveform-shaped boundary line MM as shown in FIG. 19 instead of using such a linear boundary line M. In particular, when the pattern on the first color image P and the pattern on the second color image Q are discontinuous, the discontinuity at the boundary line is eliminated with respect to the color tone by the above-described color tone correction processing. In this case, the discontinuity at the boundary line remains as it is. In general, the discontinuity related to such a pattern is less likely to be recognized by the human eye than the discontinuity related to the color tone. However, if the waveform-shaped boundary line MM is used instead of the linear boundary line M, an effect of making it difficult to recognize discontinuities by the human eye can be obtained.
[0091]
As shown in FIG. 17, it is preferable to use the boundary line of the waveform even when the images are partially overlapped and connected. In this case, as shown in FIG. 20, a waveform-shaped boundary line MM2 may be used as the second boundary line that may cause pattern discontinuity.
[0092]
【The invention's effect】
As described above, according to the color tone correction method according to the present invention, since the pixel value is corrected so that the relative positions on the histogram are the same, while maintaining the state in which the original pattern is expressed, A process of correcting only the color tone can be easily performed.
[Brief description of the drawings]
FIG. 1 is a diagram showing two color images used in a color tone correction method according to the present invention.
FIG. 2 is a flowchart showing a basic procedure for correcting a color tone that approaches the color tone of a second color image Q for the first color image P shown in FIG. 1;
FIG. 3 is a diagram showing eight sets of monochromatic planes prepared by the image input stage in step S11 of the flowchart shown in FIG. 2;
4 is a graph showing an example of a histogram obtained for one of the single-color planes shown in FIG. 3. FIG.
FIG. 5 is a graph showing a relative position on a histogram related to a first color image P with respect to a pixel value PP of a pixel to be corrected.
6 is a pixel value PP at the same relative position as the relative position on the histogram shown in FIG.^*Is a graph obtained on the histogram relating to the second color image Q.
FIG. 7 is a conceptual diagram illustrating a technique for obtaining a unique histogram for each pixel on each plane.
FIG. 8 is a conceptual diagram illustrating a method of dividing each plane into blocks and obtaining a histogram for each block.
9 is a diagram showing a state in which lattice points are defined at the representative positions (center positions) of the blocks shown in FIG. 8 and tone average and tone dispersion are defined at each lattice point. FIG.
10 is a diagram showing a method of defining tone average and tone dispersion for an arbitrary point by an interpolation method using grid points shown in FIG.
FIG. 11 is a conceptual diagram illustrating a technique for correcting a color tone on a correction target region Po when connecting a first color image P and a second color image Q.
FIG. 12 is a diagram illustrating an example in which a large pattern is configured by arranging a large number of first color images having a repeatable pattern.
13 is a flowchart showing the procedure of a specific color correction method for the correction target area Po shown in FIG.
14 is a diagram showing a weighting distribution of correction elements based on a distance D from a boundary line M when performing a specific color correction on the correction target region Po shown in FIG.
15 is a graph showing an example of a weight function r (D) that can be used for the weight distribution shown in FIG.
FIG. 16 is a diagram illustrating a block division technique for performing color correction on the correction target area Po shown in FIG. 11;
FIG. 17 is a conceptual diagram illustrating an example in which a first color image P and a second color image Q are connected in a partially overlapping manner.
FIG. 18 is a flowchart showing a procedure of a color correction method executed when a partially overlapping connection as shown in FIG. 17 is performed.
FIG. 19 is a diagram illustrating an example in which a waveform-shaped contour line MM is used when two images are connected adjacent to each other.
FIG. 20 is a diagram illustrating an example in which a contour line MM2 having a waveform shape is used when two images are partially overlapped and connected.
[Explanation of symbols]
a to d: lattice points
e, f ... arbitrary points
A, Ap (n), Aq (n) ... Tone average (average of histogram)
B1-B20 ... Block
D: Distance from boundary line M
Ep, Eq ... Neighborhood
H (n, x, y) ... representative position (center position of block)
Lp: Distance between the tone average Ap (n) on the histogram relating to the image P and the specific pixel value PP (n, x, y)
Lq: tone average Aq (n) on the histogram relating to image Q and specific pixel value PP^*Distance to (n, x, y)
M, Mp, M1, M2 ... boundary lines
MM, MM2 ... Waveform boundary
P ... First color image
Pc, Pm, Py, Pk... Monochromatic planes constituting the first color image P
Po ... Correction target area
P (n, x, y)... Pixel located at the coordinates (x, y) on the nth monochrome plane constituting the first color image P
PP (n, x, y) ... Pixel value of the pixel P (n, x, y)
PQ1 ... 1st overlap area
PQ2 ... second overlap area
Pz: First non-overlapping area
Q ... Second color image
Qc, Qm, Qy, Qk... Monochromatic planes constituting the second color image Q
Qr: Reference area
Q (n, x, y) ... pixel located at coordinates (x, y) on the n-th monochromatic plane constituting the second color image Q
QQ (n, x, y) ... Pixel value of pixel Q (n, x, y)
Qz: second non-overlapping region
r (D) ... Weight function
V, Vp (n), Vq (n) ... Tonal variance (histogram variance)
Wp: width of the correction target area Po
Wq: width of the reference area Qr

Claims (10)

A first color image in which a first pattern is expressed in a first color tone and a second color image in which a second pattern is expressed in a second color tone are prepared, and the first color image On the other hand, a color tone correction method for correcting the color tone so as to approach the second color tone while maintaining the state where the first pattern is expressed,
The first color image and the second color image are each divided into N sets of color components, and a single color plane including an array of pixels having a predetermined pixel value is configured for each color component, and the first color image An image input stage for capturing the image and the second color image as N sets of monochromatic planes, respectively;
For the N sets of single color planes constituting the first color image, a first calculation step for obtaining a histogram for each pixel value for each plane;
A second calculation step for obtaining a histogram for each pixel value for each of the N sets of monochromatic planes constituting the second color image;
For the N sets of single color planes constituting the first color image, for each plane, the pixel value of each pixel is corrected in order to correct the pixel value of each pixel with reference to the second color image. The relative position on the histogram of the pixel value is obtained, and the same single color plane as the pixel to be corrected is selected from the N sets of single color planes constituting the second color image, and the selected single color plane A correction step of obtaining a pixel value existing at the relative position on the histogram and correcting the obtained pixel value as a corrected pixel value;
An image output stage for outputting a first color image composed of N sets of monochromatic planes having corrected pixel values as a corrected image;
Have
In the first calculation step, the first color image is divided into a plurality of blocks, and an independent histogram is obtained for each block;
In the second calculation step, the second color image is divided into a plurality of blocks, and an independent histogram is obtained for each block;
Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
For each position other than the representative position, by defining the tone average and tone dispersion specific to each by interpolation based on the tone average and tone dispersion defined for a plurality of representative positions existing in the vicinity,
In the correction step, the pixel value before correction for the pixel P to be corrected in the first color image is PP, the tone average defined for the position of the pixel P is Ap, and the tone variance is Vp. The tone value defined for the same position of the second color image to be used is Aq, the tone variance is Vq, and the corrected pixel value PP ^* for the pixel P to be corrected is
PP ^* = Aq + (Vq / Vp). (PP-Ap)
A color tone correction method characterized in that the color tone correction method is obtained by the following calculation. A first color image in which the first pattern is expressed in the first color tone and a second color image in which the second pattern is expressed in the second color tone are prepared so as to sandwich a predetermined boundary line A color correction method for correcting a color tone to alleviate a sudden change in color tone at the boundary line when connecting the first color image and the second color image,
The first color image and the second color image are each divided into N sets of color components, and a single color plane including an array of pixels having a predetermined pixel value is configured for each color component, and the first color image An image input stage for capturing the image and the second color image as N sets of monochromatic planes, respectively;
A predetermined correction target area is defined on the boundary line side of the first color image, and a portion in the correction target area of the N sets of single color planes constituting the first color image is determined for each plane. A first calculation step for obtaining a histogram for each pixel value;
A predetermined reference area is defined on the boundary line side of the second color image, and a pixel in each of the planes of the N sets of monochrome planes constituting the second color image is provided for each plane. A second computation stage for obtaining a histogram for the values;
With respect to the portions in the correction target area of the N sets of monochromatic planes constituting the first color image, each plane is individually referred to the portion in the reference area in the second color image. In order to correct the pixel value of the pixel, the relative position on the histogram of the pixel value of the pixel to be corrected is obtained, and the correction target is selected from the N sets of single color planes constituting the second color image. The same monochrome plane as the pixel to be selected is selected, the pixel value existing at the relative position on the histogram relating to the portion in the reference area of the selected monochrome plane is obtained as the reference pixel value, and the correction of the pixel to be corrected Weighting based on the previous pixel value and the reference pixel value so that the influence of the reference pixel value increases as the distance between the pixel to be corrected and the boundary line decreases. To determine a new pixel value, and modifying steps of correcting the pixel value after correction the new pixel values,
An image output stage for outputting a first color image composed of N sets of monochromatic planes having corrected pixel values as a corrected image;
Have
In the first calculation step, the correction target area in the first color image is divided into a plurality of blocks, and an independent histogram is obtained for each block,
In the second calculation step, the reference region in the second color image is divided into a plurality of blocks, and an independent histogram is obtained for each block,
Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
For each position other than the representative position, by defining the tone average and tone dispersion specific to each by interpolation based on the tone average and tone dispersion defined for a plurality of representative positions existing in the vicinity,
In the correction stage, PP is the pixel value before correction for the pixel P to be corrected in the correction target area, Ap is the tone average defined for the position of the pixel P, and Vp is the tone variance, which is used for correction. The pixel corresponding to the position in the reference area is determined as a reference pixel, the tone average defined for the position of the reference pixel is Aq, the tone variance is Vq, the width of the correction target area is Wp, and the correction target When the distance between the pixel P and the boundary line is D, the corrected pixel value PP ^# for the pixel P to be corrected is
PP ^# = Aq + (Vq / Vp). (PP-Ap) .r (D)
+ PP · (1-r (D))
A color tone correction method characterized in that r (D) is obtained by an operation of (where D is monotonically increasing from 0 to Wp and monotonically decreasing from 1 to 0). A first color image in which a first pattern is expressed in a first color tone and a second color image in which a second pattern is expressed in a second color tone are prepared, and the first color image of the first color image is prepared. Defining a first overlapping region in a part, defining a second overlapping region consisting of regions congruent to the first overlapping region in a part of the second color image, and A color correction for correcting a color tone to alleviate a sudden change in a color tone when connecting the first color image and the second color image so as to overlap the second overlap area A method,
The first color image and the second color image are each divided into N sets of color components, and a single color plane including an array of pixels having a predetermined pixel value is configured for each color component, and the first color image An image input stage for capturing the image and the second color image as N sets of monochromatic planes, respectively;
A first calculation step of obtaining a histogram for each pixel value for each plane for the portions in the first overlapping area of the N sets of single color planes constituting the first color image;
A second calculation step of obtaining a histogram for each pixel value for each plane for the portions in the second overlapping area of the N sets of monochromatic planes constituting the second color image;
For the portions in the first overlapping region of the N sets of single color planes constituting the first color image, the portions in the second overlapping region in the second color image are respectively determined for each plane. Referring to this, in order to correct the pixel value of each pixel, the relative position on the histogram of the pixel value of the pixel to be corrected is obtained, and the N sets of single color planes constituting the second color image are obtained. The same monochromatic plane as the pixel to be corrected is selected from among the pixels, and the pixel value existing at the relative position on the histogram relating to the portion in the second overlapping area of the selected monochromatic plane is obtained, and the obtained A first correction stage for correcting the pixel value to the corrected pixel value;
A predetermined correction target area is defined on a boundary line side of the second color image in an area other than the first overlapping area on the first color image, and the first color image is configured. A third calculation step for obtaining a histogram for each pixel value for each plane for the portion of the N sets of monochromatic planes within the correction target region;
A predetermined reference area is defined on the boundary line side of the second color image, and a pixel in each of the planes of the N sets of monochrome planes constituting the second color image is provided for each plane. A fourth computation stage for obtaining a histogram for the values;
With respect to the portions in the correction target area of the N sets of monochromatic planes constituting the first color image, each plane is individually referred to the portion in the reference area in the second color image. In order to correct the pixel value of the pixel, the relative position on the histogram of the pixel value of the pixel to be corrected is obtained, and the correction target is selected from the N sets of single color planes constituting the second color image. The same monochrome plane as the pixel to be selected is selected, the pixel value existing at the relative position on the histogram relating to the portion in the reference area of the selected monochrome plane is obtained as the reference pixel value, and the correction of the pixel to be corrected Weighting based on the previous pixel value and the reference pixel value so that the influence of the reference pixel value increases as the distance between the pixel to be corrected and the boundary line decreases. To determine a new pixel value, the second correction step of correcting the pixel value after correction the new pixel values,
An image output stage for outputting a first color image composed of N sets of monochromatic planes having corrected pixel values as a corrected image;
A color tone correction method comprising: The color tone correction method according to claim 3 ,
When obtaining histograms in the first computation stage and the second computation stage, a predetermined neighborhood region is defined for each pixel, and a histogram relating to the neighborhood region of a certain pixel of interest is defined as the histogram for this pixel of interest. By defining a unique histogram for each pixel,
When obtaining a corrected pixel value for a pixel to be corrected in the first overlap area in the first correction stage, it corresponds to the position of the pixel to be corrected in the second overlap area A tone correction method, wherein a pixel to be corrected is calculated as a reference pixel, and a corrected pixel value for the pixel to be corrected is calculated using a histogram for the reference pixel. In the color tone correction method according to claim 3 or 4 ,
In the first correction stage, the pixel value before correction for the pixel P to be corrected in the first overlap region is set to PP, the average of the histogram for this pixel P is Ap, and the variance is Vp, which are used for correction. The average of the histogram relating to the second overlapping region is Aq, the variance is Vq, and the corrected pixel value PP ^* for the pixel P to be corrected is
PP ^* = Aq + (Vq / Vp) · (PP−Ap)
A tone correction method characterized by being obtained by the following calculation. The color tone correction method according to claim 3 ,
Dividing each of the first overlap region and the second overlap region into a plurality of blocks, and obtaining an independent histogram for each block;
Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
For each position other than the representative position, by defining the tone average and tone dispersion specific to each by interpolation based on the tone average and tone dispersion defined for a plurality of representative positions existing in the vicinity,
In the first correction stage, PP is the pixel value before correction for the pixel P to be corrected in the first overlap region, Ap is the tone average defined for the position of this pixel P, and Vp is the tone variance. A tone average defined for the same position of the second overlapping region used for correction is Aq, a tone variance is Vq, and the corrected pixel value PP ^* for the pixel P to be corrected is
PP ^* = Aq + (Vq / Vp) · (PP−Ap)
A tone correction method characterized by being obtained by the following calculation. The color tone correction method according to claim 3 ,
When obtaining histograms in the third calculation stage and the fourth calculation stage, a predetermined neighborhood region is defined for each pixel, and a histogram relating to the neighborhood region of a certain pixel of interest is defined as the histogram for this pixel of interest. By defining a unique histogram for each pixel,
In the second correction stage, when obtaining a corrected pixel value for a pixel to be corrected in the correction target area, a pixel corresponding to the pixel to be corrected in the reference area is referred to as a reference pixel. And a corrected pixel value for the pixel to be corrected is calculated using a histogram for the reference pixel. In the color correction method according to claim 3 or 7 ,
In the second correction stage, the pixel value before correction for the pixel P to be corrected in the correction target area is PP, the average of the histogram for this pixel P is Ap, and the variance is Vp. When the average of the histogram is Aq, the variance is Vq, the width of the correction target region is Wp, and the distance between the correction target pixel P and the boundary line is D, the correction target pixel P The corrected pixel value PP ^# is
PP ^# = Aq + (Vq / Vp). (PP-Ap) .r (D)
+ PP · (1-r (D))
A color tone correction method, wherein r (D) is obtained by an operation of (where D is a function that monotonously decreases from 1 to 0 as D increases monotonously from 0 to Wp). The color tone correction method according to claim 3 ,
The correction target area in the first color image and the reference area in the second color image are each divided into a plurality of blocks, and an independent histogram is obtained for each block,
Define a representative position in each block, and define the average and variance of the histogram obtained for each block as the tone average and tone variance for the representative position of each block,
For each position other than the representative position, by defining the tone average and tone dispersion specific to each by interpolation based on the tone average and tone dispersion defined for a plurality of representative positions existing in the vicinity,
In the second correction stage, the pixel value before correction for the pixel P to be corrected in the correction target area is set to PP, the tone average defined for the position of the pixel P is Ap, and the tone variance is set to Vp. The pixel corresponding to the position in the reference region used for the reference pixel is obtained as a reference pixel, the tone average defined for the position of the reference pixel is Aq, the tone variance is Vq, the width of the correction target region is Wp, When the distance between the pixel P to be corrected and the boundary line is D, the corrected pixel value PP ^# for the pixel P to be corrected is
PP ^# = Aq + (Vq / Vp). (PP-Ap) .r (D)
+ PP · (1-r (D))
A color tone correction method, wherein r (D) is obtained by an operation of (where D is a function that monotonously decreases from 1 to 0 as D increases monotonously from 0 to Wp). In the color tone correction method according to any one of claims 3 to 9 ,
A tone correction method, wherein the same repeatable image is used as the first color image and the second color image.

Images