JP2004118244A - Image processing method, image processor, image forming apparatus, computer program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processing method, an image processor, an image forming apparatus, a computer program and a recording medium, for extracting characteristic amount of an image by reflecting the information of an image detailed part and having a memory with a small storage capacity. <P>SOLUTION: This image processor includes: a reducing means 5a for obtaining a reduced image data according to the value of whole pixels of an original image data, a characteristic amount extract means 5b for extracting the characteristic amount from the obtained reduced image data, a mapping means 6a for mapping the extracted characteristic amount and the respective pixels of original image data and assigning the characteristic amount to the pixels; and a processed image data generating means 6c for generating processed image data processed according to the assigned characteristic amount. <P>COPYRIGHT: (C)2004,JPO

Description

【００５０】
なお、黒生成時に特徴量抽出処理結果を利用する場合は、特徴量抽出結果に応じてＵＣＲ率を変更したり、ｆ（ｘ）の特性カーブを変更したりする。
【００５１】
黒生成下色除去部１０９から出力されたＣＭＹＫ信号の処理画像データは、出力階調補正部１１０で出力階調補正処理を施され、階調再現処理部１１１で、最終的に画像を画素に分離して夫々の階調を再現できるように処理する階調再現処理（中間調生成処理）が施されて、画像メモリ１３へ入力される。
出力階調補正処理は、黒生成下色除去部１０９から出力されたＣＭＹＫ信号の処理画像データを、プリンタ部１４の特性値である網点面積率に変換する処理である。
【００５２】
階調再現処理部１１１では、例えば、特徴量抽出処理部１０５にて検出されたエッジ強度が高い領域は、特に黒文字又は色文字の再現性を高めるために、高域周波数の再現に適した高解像度のスクリーンでの二値化または多値化処理が選択される。また、特徴量抽出処理部１０５にてエッジ検出結果が少ない領域に関しては、階調再現性を重視したスクリーンでの二値化または多値化処理が行われる。
【００５３】
カラー画像処理部１２（階調再現処理部１１１）から出力された処理画像データは、画像メモリ１３に記憶され、所定のタイミングで画像メモリ１３から読み出されてプリンタ部１４へ入力される。
【００５４】
なお、カラー画像処理部１２は、原画像データが、入力階調補正部１０３へ入力された後に特徴量抽出処理部１０５へ入力され、次いで、空間フィルタ処理部１０６及び変倍処理部１０７等へ入力されるよう構成してあるが、このような構成に限定されるものではない。
【００５５】
例えば、空間フィルタ処理部１０６、変倍処理部１０７、及び黒生成下色除去部１０９で特徴量抽出結果を用いず、階調再現処理部１１１でのみ特徴量抽出結果を用いる（階調再現処理部１１１のみ前記生成手段を備える）場合、原画像データが、入力階調補正部１０３及び色空間変換部１０４へ入力された後に、空間フィルタ処理部１０６、変倍処理部１０７、黒生成下色除去部１０９、及び出力階調補正部１１０へ入力され、次いで、特徴量抽出処理部１０５へ入力され、最後に、階調再現処理部１１１へ入力されるよう構成しても良い。即ち、ＲＧＢ信号の原画像データが、入力階調補正部１０３にてＬＣＣ信号の原画像データに変換された後、変換された原画像データに対し、特徴量抽出結果を用いる画像処理を施す前に、縮小処理及び特徴量抽出処理を施すよう、カラー画像処理部１２の各部を構成すれば良い。
【００５６】
図３は、特徴量抽出処理部１０５及び空間フィルタ処理部１０６の構成を示すブロック図である。
特徴量抽出処理部１０５は、縮小手段５ａ及び特徴量抽出手段５ｂを備える。
また、空間フィルタ処理部１０６は、マッピング手段６ａ、ＲＯＭ１５のエッジ強調フィルタ記憶部１５ａからエッジ強調フィルタを読み込み、マッピング手段６ａで求めた新たな特徴量に基づいてエッジ強調フィルタを選択するフィルタ選択手段６ｂ、及び処理画像データ生成手段６ｃを備える。
【００５７】
色空間変換部１０４から出力されたＬＣＣ信号の原画像データは、１画素毎（１ライン毎、又は１プレーン毎等）に、特徴量抽出処理部１０５の縮小手段５ａへ入力され、また、特徴量抽出処理部１０５を介して、空間フィルタ処理部１０６の処理画像データ生成手段６ｃへ入力される。
縮小手段５ａへ入力された原画像データは、縮小処理を施される。この場合、所定の個数の画素が有するデータ（以下、画素値と言う）が縮小手段５ａへ入力されるまでは、原画像データの縮小手段５ａへの入力を続行する。このとき、所定の個数の画素値として、Ｎ×Ｎ個の方形の画素が有するデータを用い、該画素に対応する画素値が入力された時点で、縮小処理を開始する。
【００５８】
縮小手段５ａでは、画像の細部の再現性（以下、細線再現性を言う）が高い縮小処理を行なう。以下では、画素値の平均値を求める縮小処理と、線形補間法を用いる縮小処理とを例示する。
図４は、縮小手段５ａで用いられている縮小処理の説明図である。２×２の方形に配置された各矩形は、原画像データの画素を示し、各矩形が有する数値は、各画素の画素値を示している。また、中央部の斜線で示す矩形は、２×２画素に対応する縮小画像データの画素を示している。
以下では、原画像データが８ビットデータであり、各画素の画素値０〜２５５の値を取る場合を例示する。また、本実施の形態では、原画像データにおける２×２の方形の画素の画素値に基づいて、縮小画像データの画素の画素値を求める。
【００５９】
例えば、原画像データのサイズが６４０×４８０である場合、縦横夫々１／２縮小を行ない、３２０×２４０のサイズを有する縮小画像データを求めるとき、画素値の平均値を求める縮小処理では、原画像データの縦横のサイズ比を保ちつつ、２×２画素毎に画素値の平均値を求める。図４の場合（画素値２００、１００、２５５、及び２００）、該平均値（縮小画像データの画素値）は１８９となる。
以上のような縮小処理は簡単であり、しかも、原画像データの全ての画素値を用いて縮小画像データを求めることになるため、各画素に含まれた細線情報が欠落することを防止できる。
【００６０】
なお、Ｎ×Ｎ画素毎に画素値の平均値を求める場合、Ｎの値が過大（画素数が過多）であると、Ｎ×Ｎ画素全体の中で数画素程度しか存在しない細線情報は、他の画素の影響で、前記平均値に反映されにくい。このため、Ｎの値を設定する場合、該値が過大にならないよう注意する必要がある。
【００６１】
図５は、縮小手段５ａで用いられる線形補間法による縮小処理の説明図である。２×２の方形に配置された各矩形は、原画像データの画素を示している。また、前記画素の内、左上の画素の画素座標を（ｘ，ｙ）、画素値をＰ（ｘ，ｙ）とする。更に、中央部の斜線で示す矩形は、原画像データの２×２画素に対応する縮小画像データの画素を示しており、該画素の画素座標を（ｕ，ｖ）、画素値をＰ（ｕ，ｖ）とする。この場合、ｕ，ｖは、ｘ＜ｕ＜ｘ＋１，ｙ＜ｖ＜ｙ＋１の値とする。
【００６２】
原画像データにおける２×２の方形の画素の画素値に基づいて、縮小画像データの画素の画素値を求める場合、線形補間法を用いる縮小処理では、周囲４点を用いて線形補間を行なうバイリニア補間法を用いる。この場合、縮小画像データの画素の画素値は、該縮小画像データの画素と原画像データにおける４個の画素との距離（前記縮小画像データの画素の画素座標と前記４個の画素の画素座標との差）に比例した重み係数を前記４個の画素の画素値に乗じることによって算出される。
このため、縮小画像データの画素が有すべき画素値Ｐ（ｕ，ｖ）は、次式で求められる。
【００６３】

【００６４】
例えば、（ｘ，ｙ）の画素座標が（１００，１００）の場合、（ｕ，ｖ）の画素座標を（１００．５，１００．５）とし、式（２）に、各画素座標と図中の画素値とを代入して、Ｐ（ｕ，ｖ）を小数点第１位四捨五入で算出すると、

となる。
【００６５】
図６は、従来のバイリニア法による縮小画像データと、縮小手段５ａで用いられるバイリニア法による縮小画像データとを示す模式図である。図中（ａ）は原画像データ、（ｂ），（ｃ）は該原画像データをバイリニア法で縮小した縮小画像データである。ただし、図６（ｂ）の各画素値は、式（２）にｕ＝ｘ，ｖ＝ｙを代入して算出してあり、図６（ｃ）の各画素値は、式（２）にｕ＝ｘ＋０．５，ｖ＝ｙ＋０．５を代入して算出してある。
【００６６】
図６（ｂ）においては、縮小画像データの各画素の画素座標は原画像データの各２×２の画素の内、左上の画素の画素座標と一致している。このため、縮小画像データは、原画像データの画素を単純にサブサンプリングして縮小した場合の縮小画像データに等しく、２×２の画素の内、左上の画素以外の画素が有する画素値が縮小画像データの画素値に反映されていない。
図６（ｃ）においては、縮小画像データの各画素の画素座標は、原画像データの２×２の画素の画素座標と異なる。このため、該２×２の画素が有する画素値が縮小画像データの画素値に反映される。
なお、縮小画像データの各画素の画素座標と、原画像データのＮ×Ｎ画素の画素座標との差は、０．５画素分以下とするのが好ましい。これにより、細線情報の欠落を防止して、縮小処理を行ない、縮小画像データを得ることができる。
【００６７】
図７は、原画像データの一例を示す模式図、図８は、縮小手段５ａで、図７に示される原画像データを縮小してなる縮小画像データを示す模式図である。
本実施の形態においては、縮小手段５ａで、原画像データに画素値の平均値を求める縮小処理を施した場合であっても、バイリニア法を用いる縮小処理を施した場合であっても、計算結果は等しくなる。このため、図７の８×８画素の原画像データを縮小してなる縮小画像データは、図８の中心部分４×４画素の縮小画像データとなる。
なお、図８の外側部分の２０個の画素は、８×８画素の原画像データの外側の図示しない画素分の原画像データを用いて求めた縮小画像データとする。即ち、図８の全体は、１２×１２画素の原画像データを縮小した縮小画像に相当する。
【００６８】
以上のようにして、縮小手段５ａで、原画像データから縮小画像データを求めた後、特徴量抽出手段５ｂで、縮小画像データから特徴量を抽出する。
この場合、特徴量抽出手段５ｂでの特徴量抽出処理は、縮小画像データの全画素分の画素値を求めてから開始しても良く、特徴量抽出処理に必要な個数の画素値が求められた時点で開始しても良い。
以下では、ラプラシアンフィルタを用いたエッジ強度抽出処理を例示する。
【００６９】
図９は、特徴量抽出手段５ｂで用いられるラプラシアンフィルタの模式図である。該ラプラシアンフィルタは、３×３のカーネルサイズのフィルタであり、中心が８で周辺が−１のフィルタである。
前記ラプラシアンフィルタを用いて、特徴量抽出手段５ｂで縮小画像データにフィルタ処理を施し、エッジ強度を抽出した場合、抽出されるエッジ強度夫々の値（以下、エッジ強度値と言う）は−２０４０〜２０４０（１０進数）となる。次いで、エッジ強度値を、４ビット右シフトする。このとき、エッジ強度値は−１２８〜１２７（１０進数）となる。このため、エッジ強度を８ビットの値で表わすことができる。
また、各エッジ強度値は、縮小画像データの各画素の画素座標に対応している。
【００７０】
エッジ部分は、エッジ強度値の符号が切り替わっているところであり、更に、符号が切り替わっているエッジ強度値の差を、新たなエッジ強度として求める。
そこで、各エッジ強度値と、該エッジ強度値に対応する画素座標の８近傍の画素座標に夫々対応するエッジ強度値との符合をチェックし、符号が切り替わっているところに関して新たなエッジ強度を求め、求めた新たなエッジ強度値が最大のところをエッジ強度抽出結果の中心位置とする。また、８近傍で符号が変わっていないところは、抽出結果を０とし、エッジと判定された部分は周辺のエッジ強度値の絶対値の和を求める。絶対値の和が２５５を超える場合は、２５５に丸める。
【００７１】
図１０は、特徴量抽出手段５ｂで、図８に示される縮小画像データから特徴量を抽出してなる特徴量抽出結果を示す模式図である。
図中（ａ）は、前記縮小画素データの中心部分４×４画素に対し、特徴量抽出手段５ｂで、図９に示すラプラシアンフィルタ用いてフィルタ処理を行なった結果を示し、（ｂ）は、前記フィルタ処理を行なった結果を、４ビット右シフトした結果を示している。また、（ｃ）は、特徴量抽出手段５ｂでの処理を終えて、該特徴量抽出手段５ｂから画像メモリ１３へ入力されるエッジ強度抽出結果を示している。
【００７２】
図１０（ｂ）に示されるエッジ強度抽出結果において、各エッジ強度抽出結果を、８近傍の抽出結果と比較する。例えば、左上の画素座標を（ｓｘ，ｓｙ）とすると、（ｓｘ＋１，ｓｙ）については、（ｓｘ＋２，ｓｙ）と（ｓｘ＋２，ｓｙ＋１）との間で符号が逆転しており、その部分にエッジが存在することになる。また、エッジ強度抽出結果の中心部分は、（ｓｘ＋２，ｓｙ）と（ｓｘ＋１，ｓｙ）との差分値の絶対値と、（ｓｘ＋２，ｓｙ＋１）と（ｓｘ＋１，ｓｙ）との差分値の絶対値との大きな方、つまり、（ｓｘ＋２，ｓｙ）の方向になる。
次いで、エッジ部分として抽出された部分は、近傍８画素との絶対値の和を取り、また、エッジ部分以外のところは値を０として、各画素に対応するエッジ強度を求める。
【００７３】
以上のようにして求められたエッジ強度抽出結果と、該エッジ強度抽出結果の中心位置のデータとを、画像メモリ１３へ一旦退避させる。次いで、空間フィルタ処理部１０６のマッピング手段６ａでマッピング処理を行なう。
本実施の形態においては、エッジ強度抽出結果を各部へ入力する場合、画像メモリ１３に記憶してあるエッジ強度抽出結果を用いる（画像メモリ１３を介してエッジ強度抽出結果を各部へ入力する）よう構成してあるが、エッジ強度抽出結果を使用する処理が１箇所だけであり、特徴量抽出処理の次に続いているときは、エッジ強度抽出結果と中心位置のデータとを画像メモリ１３へ一旦退避させる必要はなく、カラー画像処理部１２が備えるレジスタ内に格納されているエッジ強度抽出結果と中心位置のデータとを使用して、前記処理を行なうよう構成しても良い。この場合、画像処理を高速化することができる。
【００７４】
特徴量抽出結果に施されるマッピング処理は、画像メモリ１３に一時退避させてある特徴量抽出結果、又はカラー画像処理部１２が備えるレジスタ内に格納されている特徴量抽出結果と、原画像データとを対応させる。
原画像データを縮小率１／２で縮小してなる縮小画像データから抽出された特徴量抽出結果の１画素分の特徴量は、原画像データの２×２画素分の特徴量に対応する。このため、特徴量抽出結果の特徴量の画素座標から、原画像データにおいて前記画素座標に対応する２×２画素分の画素座標を算出し、算出した画素座標に対し、前記１画素分の特徴量を振り当てる。
【００７５】
特徴量抽出結果の特徴量の画素座標を（ｓｘ，ｓｙ）とすると、対応する２×２画素分の画素座標（ｘ，ｙ）は、次の式（４），（５）を用いて夫々求める。
ｘ＝ｓｘ×２＋ａ　　　　　　　　　　　　　　　　　　　　　　　…（４）
ｙ＝ｓｙ×２＋ｂ　　　　　　　　　　　　　　　　　　　　　　　…（５）
ただし、ａ，ｂは０≦ａ，ｂ＜（縮小率１／２の逆数）を満たす整数であり、この場合、ａ，ｂ＝０，１である。
このとき、（ｓｘ，ｓｙ）に対応する４個の画素座標は、（ｓｘ×２，ｓｙ×２）、（ｓｘ×２＋１，ｓｙ×２）、（ｓｘ×２，ｓｙ×２＋１）、（ｓｘ×２＋１，ｓｙ×２＋１）となる。
【００７６】
次に、マッピングされた画素座標に、縮小手段５ａで用いた縮小処理の方法に応じて、特徴量を振り当てる。この場合、エッジ強度抽出結果の中心位置に対応して振り当てられる特徴量が最も高く、前記中心位置の周囲の画素座標に対応して振り当てられる特徴量が、前記中心位置に対応して振り当てられる特徴量より低くなるよう、夫々特徴量を振り当てる。この場合、線形補間法を用いる縮小処理を施したときは、ガウシアン分布的に中心位置だけ値が高く、周囲に行くにつれ、値が急激に下がるような特徴を持つ関数分布に従って、夫々特徴量を振り当てる。また、画素値の平均値を求める縮小処理を施したときは、中心位置の値が高く、周囲に行くにつれ、値が線形分布的に下がるような特徴を持つ関数分布に従って、夫々特徴量を振り当てる。
【００７７】
なお、この場合、振り当てられるべき特徴量が、ガウシアン分布的又は線形的に減少するような係数を有するフィルタを用いても良く、前記関数分布に対応する計算式を用いても良く、更に、前記係数をテーブルとしてＲＯＭ１５に記憶してあっても良い。
【００７８】
図１１は、マッピング手段６ａにて、図１０に示される特徴量抽出結果をマッピングして特徴量を振り当ててなる特徴量抽出結果を示す模式図である。図中（ａ）は特徴量抽出結果を示し、（ｂ）はマッピング時の重み係数を示し、（ｃ）は特徴量に重み付けして各画素に振り当てた特徴量を示し、（ｄ）は該特徴量に対応する原画像データを示している。
マッピング手段６ａでは、抽出したエッジ強度に対応する画素座標と、原画像データの各画素の画素座標とのマッピングを行なう。この場合、前述したように、エッジ強度抽出結果の画素座標（ｓｘ，ｓｙ）に対して、原画像データの４画素分の画素座標（ｓｘ×２，ｓｙ×２）、（ｓｘ×２＋１，ｓｙ×２）、（ｓｘ×２，ｓｙ×２＋１）、及び（ｓｘ×２＋１，ｓｙ×２＋１）が対応する。
【００７９】
また、エッジ強度抽出結果の中心位置に応じて、マッピング時の重み係数を求める。例えば、図１１（ｂ）に示すように、（ｓｘ＋１，ｓｙ）に対するマッピング時の重み係数は、エッジ強度抽出結果の中心位置が（ｓｘ＋２，ｓｙ）方向であるため、右側の（ｓｘ＋１，ｓｙ）と（ｓｘ＋１，ｓｙ＋１）が中心位置に近い画素座標となり、該画素座標に対応する重み係数として０．７５を用い、左側の（ｓｘ，ｓｙ）と（ｓｘ，ｓｙ＋１）が中心位置から遠い画素座標となり、該画素座標に対応する重み係数として０．２５を用いる。
【００８０】
この場合、バイリニア法を用いる縮小処理を施した場合はガウシアン分布的な重み係数を用い、画素値の平均値を求める縮小処理を施した場合は線形的な重み係数を用いるのが好ましいが、本実施の形態では、どちらの縮小処理を施した場合であっても同じ縮小結果になるため、重み係数を簡単に算出できる線形的な重み係数を例示する。即ち、該重み係数は、特徴量抽出結果の中心位置から２×２画素分の中央を点対称とする反対側に向かって値が小さくなるように設定される。
【００８１】
図１１に示すように、特徴量抽出結果とマッピング時の重み係数とを掛け合わせ、原画像データの各画素に対する特徴量抽出結果を求める。次いで、その値と、あらかじめ設定された閾値とを比較することにより、適応的フィルタ処理を行なう。
なお、図１１では、マッピング結果をわかりやすくするために、マッピングされた特徴量抽出結果を示している。このため、特徴量抽出結果を拡大処理しているように見えるが、実際の処理では、各エッジ強度値にマッピング時の重み係数を掛けた結果を求め、その値に基づいてフィルタ係数を選択し、フィルタ処理を施すため、原画像データと同じサイズの特徴量抽出結果を保持しているわけではない。
【００８２】
次に、マッピングされた画素座標に振り当てられた特徴量抽出結果を用いて、適応的な画像処理を行なう。以下では、エッジ強度抽出結果を用いた適応的なフィルタ処理を行なう場合を例示する。
まず、エッジ強度抽出結果の符号が切り替わっている部分での抽出結果の大きさを複数の閾値で分類し、各閾値以上の場合に、それに対応したフィルタ係数でエッジ強調処理を行なうようにする。例えば、エッジの強調度合いを変えた３つのエッジ強調フィルタをエッジ強調フィルタ記憶部１５ａに記憶しておき、フィルタ選択手段６ｂへ、前記エッジ強調フィルタを読み込む。
【００８３】
図１２は、エッジ強調フィルタ記憶部１５ａに記憶してあるエッジ強調フィルタの一例を示す模式図である。
図中（ａ）は一番強いエッジ強調フィルタであり、（ｂ）は二番目に強いエッジ強調フィルタであり、（ｃ）は三番目に強いエッジ強調フィルタである。
フィルタ選択手段６ｂでは、１９２，１２８，６４の３つの閾値を用いて、０〜２５５を略均等に４個の範囲に分類する。
フィルタ選択手段６ｂにおいては、エッジ強度値が１９２以上２５５以下のときは、一番強いエッジ強調フィルタ（ａ）を選択する。また、１２８以上１９１以下のときは、二番目に強いエッジ強調フィルタ（ｂ）を選択し、６４以上１２７以下のときは、三番目に強いエッジ強調フィルタ（ｃ）を選択する。更に、０以上６３以下のときは、エッジ強調フィルタを選択しない。
【００８４】
次いで、処理画像データ生成手段６ｃにおいて、エッジ強度値が６４以上２５５以下の場合は、フィルタ選択手段６ｂで選択したエッジ強調フィルタを用いて、原画像データに対し、フィルタ処理を行ない、処理画像データを生成して、変倍処理部１０７へ出力する。０以上６３以下の場合は、フィルタ処理を行なわず、原画像データを処理画像データとして、変倍処理部１０７へ出力する。
以上のようにして、抽出したエッジ強度に応じた適応的なエッジ強調フィルタ処理を行なう。
【００８５】
図１３は、処理画像データ生成手段６ｃで、原画像データに対し、エッジ強調フィルタを用いてフィルタ処理を施してなる処理画像データを示す模式図である。
以上のような画像処理を、注目画素を１画素ずつずらしながら、縮小画像データ１ライン分の画像処理を行なう。該１ライン分の処理が終了した場合、次の１ライン分の画像処理を行ない、最終的に、全画素に対して画像処理を行なう。
【００８６】
図１４は、特徴量抽出処理部１０５の画像処理（特徴量抽出処理）の手順を示すフローチャートである。
色空間変換部１０４から出力された原画像データの入力を受け付ける。（Ｓ１１）。この場合、前記原画像データは、特徴量抽出処理部１０５の縮小手段５ａへ入力され、また、特徴量抽出処理部１０５を介して遅延され、所定のタイミングで空間フィルタ処理部１０６の処理画像データ生成手段６ｃへ入力される（後述するＳ３５参照）。
次に、入力された原画像データに対し、縮小手段５ａにて、縮小率１／２で、２×２画素の平均値を求める（又はバイリニア法を用いる）縮小処理を施し、縮小画像データを生成する（Ｓ１２）。
【００８７】
生成した縮小画像データに対し、ラプラシアンフィルタを用いたフィルタ処理を施し（Ｓ１３）、算出したエッジ強度に対して４ビット右シフト演算を行ない（Ｓ１４）、各エッジ強度と、該エッジ強度に対応する画素座標の８近傍の画素座標に夫々対応するエッジ強度の符合をチェックし、符号が切り替わっているところに関して、エッジ強度値の差を新たなエッジ強度として抽出し（Ｓ１５）、求めた新たなエッジ強度が最大のところをエッジ強度抽出結果の中心位置とする（Ｓ１６）。
【００８８】
求めたエッジ強度抽出結果と、エッジ強度抽出結果の中心位置のデータとを、画像メモリ１３へ出力し（Ｓ１７）、最後に、原画像データの全画素が特徴量抽出処理部１０５へ入力され、Ｓ１２〜Ｓ１７の処理が終了したか否かを判定し（Ｓ１８）、終了していない場合は（Ｓ１８でＮＯ）、処理をＳ１１へ戻して原画像データの入力を受け付ける。前記処理が終了した場合は（Ｓ１８でＹＥＳ）、特徴量抽出処理を終了する。
【００８９】
図１５は、空間フィルタ処理部１０６の画像処理（処理画像データ生成処理）の手順を示すフローチャートである。
まず、３個のエッジ強調フィルタを読み込む（Ｓ３１）。
画像メモリ１３から出力されたエッジ強度の入力を１画素分ずつ受け付ける（Ｓ３２）。
１画素分のエッジ強度をマッピングして２×２画素分に振り当て（Ｓ３３）、振り当てた各エッジ強度を閾値と比較し、比較結果に基づいてエッジ強調フィルタを選択する（Ｓ３４）。
【００９０】
次いで、特徴量抽出処理部１０５を介した原画像データの入力を受け付ける（Ｓ３５）。入力された原画像データの前記２×２画素に対応する各画素に対し、Ｓ３４で選択したエッジ強調フィルタを用いてフィルタ処理を行ない、処理画像データを生成する（Ｓ３６）。
最後に、生成した処理画像データを変倍処理部１０７へ出力し（Ｓ３７）、特徴量抽出結果の全画素に対して、Ｓ３２〜Ｓ３７の処理が終了したか否かを判定する（Ｓ３８）。前記処理が終了していない場合は（Ｓ３８でＮＯ）、処理をＳ３２へ戻してエッジ強度の入力を受け付け、終了した場合は（Ｓ３８でＹＥＳ）、処理画像データ生成処理を終了する。
【００９１】
以上のような画像形成装置１は、原画像データを縦横１／２に縮小してなる縮小画像データを用いて特徴量を抽出するため、特徴量を抽出する際に扱うべきデータ量が１／４になる。このため、特徴量を抽出する処理に要する時間が短縮され、画像メモリ１３が要すべき記憶容量も１／４になる。
また、全方向のエッジ強度を抽出する場合に、ラプラシアンフィルタを用いてエッジ強度を抽出するため、例えばソーベルフィルタを用いてエッジ強度を抽出する場合のように、複数のフィルタを用いる必要がない。
【００９２】
なお、本実施の形態においては、画像形成装置として、内蔵のスキャナ部で原稿を読み取ってプリンタ部で印刷するコピー装置を例示したが、外部から送信された原画像データを受信し、受信した原画像データから処理画像データを生成してプリンタ部で印刷するコピー装置であっても良い。また、ファクシミリ装置、又はコピー装置とファクシミリ装置との複合機等であっても良い。
【００９３】
また、本実施の形態においては、処理画像データを生成する場合に、特徴量に基づいてフィルタ処理（この場合、エッジ強度に基づいたエッジ強調フィルタ処理）を原画像データに施す場合を例示したが、特徴量に基づいて色補正又は解像度変換等を行なうよう構成しても良い。
【００９４】
スキャナ部で画像を読み取ってなるＲＧＢデータを、プリンタ部で印刷すべくＣＭＹＫデータに変換するような色補正を行なう場合、原画像データのデータ（色情報）を各画素の色度の値として、ａ＊　又はｂ＊　（国際照明委員会ＣＩＥ１９７６Ｌ＊　ａ＊　ｂ＊　空間で定められる色度）等を特徴量として抽出し、抽出結果に基づいて色域圧縮し、色補正を行なう。
【００９５】
また、人の肌、空の青、又は草木の緑等を、人間が見て好ましく感じられるように、元の画像に忠実な色で再現せず、人間が記憶している色に近い色（記憶色）で再現するような色補正を行なう場合、抽出結果に基づいて記憶色であるか否か判定し、記憶色である場合は所定の記憶色処理（肌色については彩度を低下させ、肌色以外の色については彩度と明度とを上昇させる処理）を行ない、記憶色でない場合は通常の色補正処理を行なう。
【００９６】
解像度変換を行なう場合、例えば、エッジ、画素の色、又はランレングス（同じ濃度の画素が連続する数）等を文字の特徴量として抽出し、文字の部分だけ解像度が高くなるよう抽出結果に基づいて解像度変換を行なう。この場合、文字のエッジ再現性を向上させることができる。
また、複数のエッジ強調フィルタのみならず、他のフィルタ又は色補正若しくは解像度変換等の処理手段の内、適宜の処理手段を、抽出した特徴量に基づいて選択し、選択した処理手段を原画像データに施して処理画像データを生成するよう構成しても良い。
【００９７】
また、本実施の形態では、特徴抽出処理として近傍８画素を用いたエッジ強度抽出処理としているが、これ以外の周辺画素数を用いても構わない。更に、特徴量抽出結果を用いた適応的画像処理としてエッジ強調フィルタ処理を挙げているが、それ以外の処理に適応できることは言うまでもない。
更に、本実施の形態では、処理画像データを用いて画像を形成する手段として、用紙に画像を形成するプリンタ部を例示したが、ディスプレイに画像を形成する表示部であっても良い。
【００９８】
実施の形態　２．
図１６は、本発明の実施の形態２に係る画像処理装置２の構成を示すブロック図である。
図中２０は、画像処理装置２の制御部であり、該制御部２０は、ＣＲＴ又は液晶ディスプレイ等を用いてなる表示部２１、及び、マウス又はキーボード等を用いてなる操作部２２等の装置各部に、バスを介して接続されている。
制御部２０は、ハードディスクを用いてなる補助記憶部２６、又はＲＯＭ２４に記憶されたプログラムに従って装置各部を制御し、各種処理を実行する。
【００９９】
外部記憶部２７は、例えばＣＤ−ＲＯＭドライブであり、制御部２０に制御されて、ＣＤ−ＲＯＭのような可搬性を有する記録媒体（例えば本実施の形態に係るプログラム、並びに該プログラムを実行する場合に用いるエッジ強調フィルタ及びデータ等が記録されている記録媒体３）から情報を読み込んで補助記憶部２６に記憶させる。
ＬＡＮインターフェイス２３は、スキャナ装置４１及びプリンタ装置４２が接続されている構内回線Ｌに接続してあり、制御部２０は、構内回線Ｌ及びＬＡＮインターフェイス２３を介して、スキャナ装置４１及びプリンタ装置４２に対する画像データの送受信を行なう。
【０１００】
スキャナ装置４１からＬＡＮインターフェイス２３を介して画像データを受信した場合、制御部２０は、受信した画像データをＲＡＭ２５又は補助記憶部２６に記憶させる。また、ユーザの指示に応じて、ＲＡＭ２５に記憶させた画像データに対し、前記プログラムの画像処理手順に従って画像処理を施し、処理画像データを生成する。更に、ユーザの指示に応じて、生成した処理画像データを補助記憶部２６に記憶させたり、プリンタ装置４２へ送信したりする。
【０１０１】
図１７は、画像処理装置２の画像処理手順を示すフローチャートである。
制御部２０は、ＬＡＮインターフェイス２３を介して受信した原画像データ、又は補助記憶部２６から読み出した原画像データをＲＡＭ２５に記憶させる（Ｓ５１）。
ＲＡＭ２５に記憶させた原画像データに対し、縮小率１／２で、２×２画素の平均値を求める（又はバイリニア法を用いる）縮小処理を施し、縮小画像データを生成する（Ｓ５２）。この場合、前記原画像データを加工／消去しないようにして縮小画像データを生成する。なお、前記縮小処理を行なう前に、所定の画像処理（例えば色空間変換処理）を行っても良い。
【０１０２】
生成した縮小画像データに対し、ラプラシアンフィルタを用いたフィルタ処理を施し（Ｓ５３）、算出したエッジ強度に対して４ビット右シフト演算を行ない（Ｓ５４）、各エッジ強度と、該エッジ強度に対応する画素座標の８近傍の画素座標に夫々対応するエッジ強度の符合をチェックし、符号が切り替わっているところに関して、エッジ強度値の差を新たなエッジ強度として抽出し（Ｓ５５）、求めた新たなエッジ強度が最大のところをエッジ強度抽出結果の中心位置とする（Ｓ５６）。エッジ強度抽出結果が求まった後は、前記縮小画像データを消去しても良い。
次いで、補助記憶部２６から、３個のエッジ強調フィルタをＲＡＭ２５へ読み込む（Ｓ５７）。
【０１０３】
各エッジ強度をマッピングして２×２画素分に振り当て（Ｓ５８）、振り当てた各エッジ強度を閾値と比較し、比較結果に基づいてエッジ強調フィルタを選択する（Ｓ５９）。
ＲＡＭ２５に記憶させた原画像データの前記２×２画素に対応する各画素に対し、Ｓ５９で選択したエッジ強調フィルタを用いてフィルタ処理を行ない、処理画像データを生成する（Ｓ６０）。この場合、前記原画像データを加工／消去しても良い。また、生成した処理画像データに、所定の画像処理（例えば変倍処理、又は色補正処理等）を行っても良い。
最後に、生成した処理画像データを出力する（Ｓ６１）。この場合、前記処理画像データをプリンタ装置４２へ送信したり、補助記憶部２６に記憶させたり、表示部２１に表示させたりする。
【０１０４】
以上のような画像処理装置２は、実施の形態１の画像形成装置１と同様にして、原画像データから処理画像データを生成することができる。
また、本実施の形態では、コンピュータに実行させるためのプログラムを記録したコンピュータ読み取り可能な記録媒体に、画像処理方法を記録するものとなっている。この結果、特徴抽出方法を行なうプログラムを記録した記録媒体を持ち運び自在に提供することができる。
【０１０５】
なお、本実施の形態においては、画像処理装置として、外付けのスキャナ装置で原稿を読み取って生成された原画像データを受信し、受信した原画像データから処理画像データを生成してプリンタ装置へ送信するパーソナルコンピュータを例示したが、このような構成に限られるものではなく、例えばデジタルカメラから出力された原画像データを受信するよう構成しても良い。また、内蔵の補助記憶部にインストールされたペイントソフトを用いて生成された原画像データから処理画像データを生成して、表示部に出力する（表示させる）よう構成してあってもよい。
また、プログラムを補助記憶部２６にインストールして実行するだけでなく、記録媒体３から直接的に読み込んで実行する構成であっても良い。
【０１０６】
また、記録媒体としては、磁気テープ若しくはカセットテープ等のテープ系、フレキシブルディスク若しくはハードディスク等の磁気ディスク、ＣＤーＲＯＭ／ＭＯ／ミニディスク／ＤＶＤ等の光ディスクのディスク系、ＩＣカード（メモリカードを含む）／光カード等のカード系、又はマスクＲＯＭ、ＥＰＲＯＭ（Ｅｒａｓａｂｌｅ　Ｐｒｏｇｒａｍｍａｂｌｅ　Ｒｅａｄ　Ｏｎｌｙ　Ｍｅｍｏｒｙ）、ＥＥＰＲＯＭ（Ｅｌｅｃｔｒｉｃａｌｌｙ　Ｅｒａｓａｂｌｅ　Ｐｒｏｇｒａｍｍａｂｌｅ　Ｒｅａｄ　Ｏｎｌｙ　Ｍｅｍｏｒｙ　）、フラッシュメモリ等による半導体メモリを含めた固定的にプログラムを担持する媒体等であっても良い。
【０１０７】
また、本実施の形態においては、インターネットを含む通信ネットワークを接続可能なシステム構成も可能であることから、通信ネットワークからプログラムをダウンロードするように流動的にプログラムを担持する媒体であっても良い。なお、このように通信ネットワークからプログラムをダウンロードする場合には、そのダウンロード用のプログラムは予め本体装置に格納しておくか、あるいは別な記録媒体からインストールされるものであっても良い。
【０１０８】
【発明の効果】
本発明の画像処理方法、画像処理装置、該画像処理装置を備える画像形成装置、コンピュータを前記画像処理装置として用いるためのコンピュータプログラム、及び該コンピュータプログラムを記録する記録媒体によれば、特徴量を抽出する処理に要する時間を短縮することができるため、画像処理全体に要する時間も短縮することができる。また、記憶容量が小さいメモリを用いて特徴量を抽出する処理を行なうことができるため、メモリコストが低減した装置を構成することができる。例えばデータ量が１／４の縮小画像データを用いる場合、必要な記憶容量は１／４であるため、メモリコストを１／４に削減することができる。更に、全体的なパフォーマンスを向上することができる。
【０１０９】
また、本発明は、特にエッジ強調に好適である。
また、前記コンピュータプログラムは、従来に比べて全体的なパフォーマンスが向上した画像処理方法をコンピュータに実行させることができ、前記画像処理方法を汎用的なものとすることができる。
更に、前記記録媒体は、前記コンピュータプログラムをコンピュータに容易に供給することができる等、本発明は優れた効果を奏する。
【図面の簡単な説明】
【図１】本発明の実施の形態１に係る画像形成装置の構成を示すブロック図である。
【図２】本発明の実施の形態１に係る画像形成装置が備えるカラー画像処理部の構成を示すブロック図である。
【図３】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部及び空間フィルタ処理部の構成を示すブロック図である。
【図４】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部が有する縮小手段で用いられている縮小処理の説明図である。
【図５】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部が有する縮小手段で用いられる線形補間法による縮小処理の説明図である。
【図６】従来のバイリニア法による縮小画像データと、本発明の実施の形態１に係る画像処理方法で用いられるバイリニア法による縮小画像データとを示す模式図である。
【図７】原画像データの一例を示す模式図である。
【図８】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部が有する縮小手段で原画像データを縮小してなる縮小画像データを示す模式図である。
【図９】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部が有する特徴量抽出手段で用いられるラプラシアンフィルタの模式図である。
【図１０】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部が有する特徴量抽出手段で縮小画像データから特徴量を抽出してなる特徴量抽出結果を示す模式図である。
【図１１】本発明の実施の形態１に係る画像処理装置が備える空間フィルタ処理部が有するマッピング手段にて、特徴量抽出手段で抽出された特徴量抽出結果をマッピングして特徴量を振り当ててなる特徴量抽出結果を示す模式図である。
【図１２】本発明の実施の形態１に係る画像形成装置が備えるエッジ強調フィルタ記憶部に記憶してあるエッジ強調フィルタの一例を示す模式図である。
【図１３】本発明の実施の形態１に係る画像処理装置が備える空間フィルタ処理部が有する処理画像データ生成手段で、原画像データに対し、エッジ強調フィルタを用いてフィルタ処理を施してなる処理画像データを示す模式図である。
【図１４】本発明の実施の形態１に係る画像処理装置が備える特徴量抽出処理部の画像処理手順を示すフローチャートである。
【図１５】本発明の実施の形態１に係る画像処理装置が備える空間フィルタ処理部の画像処理手順を示すフローチャートである。
【図１６】本発明の実施の形態２に係る画像処理装置の構成を示すブロック図である。
【図１７】本発明の実施の形態２に係る画像処理装置の画像処理手順を示すフローチャートである。
【符号の説明】
１１　スキャナ部
１２　カラー画像処理部
１３　画像メモリ
１４　プリンタ部
１５ａ　エッジ強調フィルタ記憶部
５ａ　縮小手段
５ｂ　特徴量抽出手段
６ａ　マッピング手段
６ｃ　処理画像データ生成手段[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image processing method, an image processing apparatus, and an image forming apparatus including the image processing apparatus for extracting a quantity or a value (hereinafter, referred to as a feature quantity) representing a characteristic property of an image from image data of the image. The present invention relates to a computer program for using a computer as the image processing device, and a recording medium for recording the computer program.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, when a feature amount of an image is extracted and an image is processed based on the extracted feature amount, image data (digital data) of an image from which the feature amount is to be extracted is used, and data included in each pixel of the image data is used. The processing of extracting the feature amount has been performed for each of them.
[0003]
Hereinafter, a case will be exemplified where a standard deviation value in a block of 5 × 5 pixels is used as a feature amount for an image of 512 × 512 pixels. In this case, using the image data of the image, the standard deviation value in the block is calculated for all 512 × 512 pixels. That is, for each pixel in the first row of the image data, a standard deviation value is calculated using the block in order from the first column to the 512th column. The standard deviation values are calculated using the blocks in order up to the 512th column. Such arithmetic processing is repeated in order up to the 512th line, and a feature amount (512 × 512 feature amounts) corresponding to each pixel of the image data is extracted.
Finally, based on the extracted feature amount (for example, using a filter adapted to the extracted feature amount), a process of generating required processed image data (image enhancement, color correction, etc.) is performed.
[0004]
Further, in the conventional image feature amount extraction method, a process of extracting a feature amount is performed for each of data included in some pixels of the image data. Next, the generated feature amount is enlarged to obtain a feature amount corresponding to each pixel of the image data (for example, see Patent Document 1).
[0005]
For example, a standard deviation value is calculated for each pixel of an odd column (1, 3,..., 511) in the first row of the image data using the block in order, and then the odd column in the third row is calculated. A standard deviation value is calculated for each of the pixels of the eye in order using the blocks. Such calculation processing is sequentially performed on the odd-numbered rows (1, 3,..., 511 rows), and the feature amount (512 × 512 ÷) corresponding to 画素 of the pixels included in the image data. (Four feature amounts) are extracted. Next, the feature amount corresponding to the pixel in the n-th row and the n-th column (n: an odd number equal to or more than 512) is calculated by using the pixel in the n-th row and the n-th column, the pixel in the n-th row and the n + 1-th column, And n + 1 rows and n + 1 columns.
[0006]
[Patent Document 1]
JP-A-61-180383
[0007]
[Problems to be solved by the invention]
However, when the process of extracting the feature amount is performed for all the pixels, there is a problem that the process requires a long time. In addition, since data possessed by all pixels is used, calculated data is temporarily saved during the above-described processing, or extracted feature amounts are stored until processing for generating processed image data is performed. There is also a problem that it is necessary to use a memory having a large storage capacity as the memory to perform.
Further, the image feature amount extraction method disclosed in Patent Document 1 can reduce the time required for the process of extracting the feature amount. However, when performing the process, a skip operation that simply skips an appropriate number of pixels is performed. Therefore, when the skipped pixels include data of image details (for example, thin lines), there is a problem that the extracted feature amounts do not reflect the characteristics of the image details.
[0008]
The present invention has been made to solve such a problem, and based on data of all pixels of image data, generates reduced image data obtained by reducing the image data, and generates the reduced image data. By performing the process of extracting the feature amount for each data of each pixel, the time required for the process of extracting the feature amount can be reduced, the storage capacity of the memory required for the process can be reduced, and Image processing method, image processing device, image forming apparatus using the image processing device, and computer program for using a computer as the image processing device, and computer It is an object to provide a recording medium for recording a program.
[0009]
[Means for Solving the Problems]
An image processing method according to the present invention is an image processing method for extracting a feature amount from image data and generating processed image data processed based on the extracted feature amount. To obtain the data, the data that the pixels of the reduced image data should have is obtained based on the data of the plurality of pixels of the image data, and the obtained reduced image data is made to correspond to the pixel coordinates of each pixel of the reduced image data. The feature amount is extracted, and the pixel coordinates of the extracted feature amount and the pixel coordinates of each pixel of the image data are mapped to obtain a new feature amount in which the feature amount is assigned to the pixel coordinates. Generating processed image data based on the new feature amount.
[0010]
In the image processing method according to the present invention, the pixel coordinates of the N × N pixels (N: a natural number of 2 or more) of the image data are different from the pixel coordinates of the pixels of the reduced image data. And determining the data that the pixel should have based on the data of the N × N pixels and the difference between the pixel coordinates of the pixel and the pixel coordinates of the pixel of the reduced image data. I do.
The image processing method according to the present invention is characterized in that the data that the pixels of the reduced image data should have is determined by averaging the data of the N × N pixels of the image data.
[0011]
In the image processing method according to the present invention, when the feature amount is edge strength, a plurality of edge enhancement filters having different intensities are prepared, and pixel coordinates of edge strength extracted from the reduced image data; The pixel coordinates of each pixel of the data are mapped to determine the new edge strength obtained by assigning the edge strength to the pixel coordinates, and the edge selected based on the new edge strength is selected from among the edge enhancement filters. It is characterized in that processed image data is obtained using an enhancement filter.
The image processing method according to the present invention obtains an edge intensity obtained by performing a shift operation on the edge intensity extracted from the reduced image data, and maps pixel coordinates of the obtained edge intensity to pixel coordinates of each pixel of the image data. It is characterized by doing.
[0012]
The image processing method according to the present invention maps pixel coordinates of a feature amount corresponding to a pixel of the reduced image data to pixel coordinates of the N × N pixels, and allocates the feature amount to the pixel coordinates. In this case, pixel coordinates at which the assigned feature amount should be maximum are calculated based on the extracted feature amounts, and pixel coordinates near the calculated pixel coordinates are assigned to pixel coordinates separated from the pixel coordinate vicinity. The feature amount is assigned to a feature amount larger than the assigned feature amount.
The image processing method according to the present invention is characterized in that the feature amount assigned to each pixel coordinate is reduced in a Gaussian distribution from pixel coordinates near the pixel coordinates to pixel coordinates separated from the pixel coordinates. It is characterized in that the amount is applied.
The image processing method according to the present invention is characterized in that the feature amount assigned to each pixel coordinate is linearly reduced from the pixel coordinates in the vicinity of the pixel coordinates to the pixel coordinates separated from the vicinity of the pixel coordinates. Is characterized by
[0013]
An image processing apparatus according to the present invention is configured to extract reduced image data obtained by reducing image data in an image processing apparatus that extracts a feature amount from image data and generates processed image data that is processed based on the extracted feature amount. Reducing means for obtaining data that the pixels of the reduced image data should have, based on data of a plurality of pixels of the image data, and pixel coordinates of each pixel of the reduced image data from the obtained reduced image data. A feature value extraction unit that extracts a feature value in correspondence with the above, mapping the pixel coordinates of the extracted feature value and the pixel coordinates of each pixel of the image data, and assigning the feature value to the pixel coordinates. It is characterized by comprising mapping means for obtaining a new feature amount, and generating means for generating processed image data processed based on the obtained new feature amount.
[0014]
An image forming apparatus according to the present invention includes: the image processing apparatus; and a unit configured to form an image using processed image data generated by the image processing apparatus.
[0015]
A computer program according to the present invention is a computer program for causing a computer to extract a feature amount from image data and to generate processed image data processed based on the extracted feature amount. Determining the data that the pixels of the reduced image data should have, based on the data of the plurality of pixels of the image data, and obtaining the reduced image data from the obtained reduced image data. Extracting a feature amount corresponding to the pixel coordinates of each pixel of the reduced image data; mapping the pixel coordinates of the extracted feature amount to the pixel coordinates of each pixel of the image data; A step of obtaining a new feature amount to which the feature amount is assigned; and a process based on the obtained new feature amount. Characterized in that to perform the step of generating the processed image data.
[0016]
The recording medium according to the present invention has a computer that records a computer program for causing a computer to extract a feature amount from image data and generate processed image data processed based on the extracted feature amount. In a possible recording medium, based on data of a plurality of pixels of the image data, data required by the pixels of the reduced image data is obtained in order to cause the computer to obtain reduced image data obtained by reducing the image data. And extracting a feature amount from the obtained reduced image data in correspondence with the pixel coordinates of each pixel of the reduced image data. Pixel coordinates of the extracted feature amount and each pixel of the image data Mapping the pixel coordinates of, and determining a new feature amount by assigning the feature amount to the pixel coordinates, Characterized by recording a computer program for executing the step of generating the processed image data processed based on the new feature quantity was fit.
[0017]
According to the present invention, reduced image data obtained by reducing image data is obtained before performing a process of extracting a characteristic amount without directly extracting a characteristic amount from the image data. In this case, reduced image data having fewer pixels than original image data that has not been reduced (hereinafter referred to as original image data) is obtained. At this time, data to be possessed by each pixel of the reduced image data is obtained based on the data of all the pixels without skipping some pixels of the original image data.
Next, a feature amount is extracted from the obtained reduced image data in correspondence with the pixel coordinates of each pixel of the reduced image data. The reduced image data has a smaller number of pixels to be handled when extracting the feature amount than the original image data, so that the time required for the process of extracting the feature amount can be reduced, and a memory having a small storage capacity is used. The above-described processing can be performed.
[0018]
Further, the reduced image data is not obtained by simply sub-sampling (decimating) the pixels of the original image data, and the data of the pixels of the reduced image data should be based on the data of the plurality of pixels of the original image data. Is obtained, the feature of the details of the image included in the original image data can be included in the reduced image data. Therefore, the features of the details of the image are reflected in the feature amount extracted from the reduced image data.
Further, the feature quantity extracted from the reduced image data corresponds to the pixel coordinates of each pixel of the reduced image data, and the feature quantity is mapped by mapping the pixel coordinates and the pixel coordinates of each pixel of the original image data. A plurality of pixel coordinates to which is to be assigned are obtained. Next, a new feature value corresponding to the pixel coordinates is obtained by assigning the feature value to the pixel coordinates. By performing such processing, the feature amount extracted from the reduced image data can be handled in the same manner as the feature amount extracted from the original image.
[0019]
Further, processed image data processed based on the obtained new feature amount is generated. The processed image data may be processed image data using the new feature amount, or processed image data obtained by processing (eg, filtering) the original image data based on the feature amount. Is also good.
Further, the new feature amount is not a feature amount obtained by simply enlarging the feature amount and corresponding to the pixel coordinates, but a feature amount obtained by assigning appropriate weights to the pixel coordinates. Therefore, in the original image data, a large amount of the extracted feature amount is assigned to a pixel having a feature amount corresponding to the extracted feature amount, and the extracted feature amount is assigned to a pixel other than the pixel. Can be allocated less. In other words, it is possible to prevent each pixel from being over- or under-performed.
[0020]
The image processing apparatus that performs image processing as described above can be configured to be provided in an image forming apparatus such as a digital copying machine, a printer, or a facsimile machine. Further, the present invention can be configured by using the program of the present invention stored in the recording medium of the present invention and installing the program on a computer.
[0021]
In the present invention, two or more natural numbers N are determined in accordance with a required reduction ratio, and one of the reduced image data is determined based on data of N × N square pixels of the original image data. The data that the (or less than N × N) pixels should have is determined. That is, the N × N square pixels of the image are reduced to one (or less than N × N) pixels. In this case, reduced image data can be easily obtained while maintaining the vertical / horizontal size ratio of the image.
[0022]
In order to obtain reduced image data, the reduced image data of the N × N pixels of the original image data and the difference (distance) between the pixel coordinates of the pixel and the pixel coordinates of the pixel of the reduced image data are determined. When obtaining data that a pixel should have, for example, a linear interpolation method (in the case of N = 2, a bilinear method of performing four-point interpolation by performing linear interpolation between two points twice) is used. At this time, if the pixel coordinates of the pixels of the reduced image data are determined to be equal to any one of the pixel coordinates of the N × N pixels, the pixel data of the reduced image data is The data becomes equal to the data of the pixel located at the pixel coordinates, and the data of the pixels other than the pixel are not included in the data of the pixels of the reduced image data, and only reduced image data similar to that obtained when sub-sampling is obtained. Therefore, by determining the pixel coordinates of the pixels of the reduced image data so as to be different from the pixel coordinates of the N × N pixels, it is possible to obtain reduced image data in which the data of all the pixels is reflected. That is, the features of the details of the image are reflected in the feature amount extracted from the obtained reduced image data.
[0023]
Also, when averaging the data of the N × N pixels of the original image data to determine the data that the pixels of the reduced image data should have, one or more of the N × N pixels are sub-sampled. Thus, reduced image data reflecting the data of all the pixels can be easily obtained.
[0024]
In the present invention, the edge intensity is extracted as a feature using a Laplacian filter or a Sobel filter. Further, an edge enhancement filter is selected according to the extracted edge strength, and, for example, original image data is processed using the selected edge enhancement filter to obtain processed image data. That is, filter processing is performed while switching the degree of edge enhancement. For this reason, it is possible to emphasize the edge according to the intensity of the edge so that the portion having a strong edge is sharpened and the portion where the edge is not strong is not sharpened, so that the blurred image can be sharpened. (The sharpness can be improved).
[0025]
Further, for example, an edge intensity is extracted using a Laplacian filter, and an edge intensity obtained by performing a shift operation on the extracted edge intensity is obtained. The extracted edge strength is multiplied by a constant by a shift operation. In this case, by shifting right by n bits, the edge strength becomes 2 ^-N Multiply. When the edge enhancement filter is switched based on a predetermined threshold, when the extracted edge strength is used, the difference between the edge strengths may be more detailed than necessary. For example, the edge of 2040, 2035, 2030, and 2025 may be used. It is difficult to set a threshold for the intensity (it is necessary to set the threshold in a pinpoint manner, which is very severe).
[0026]
However, when a right shift of 4 bits is performed for each edge strength, edge strengths 127, 127, 126, and 126 are calculated, and the setting of the threshold value becomes easy (so that the setting of the threshold value can be made somewhat rough). . For this reason, the edge strength can be easily used as a criterion for selecting an edge enhancement filter, and the accuracy of the filter processing can be improved.
[0027]
Further, in the present invention, the pixel coordinates of the feature amount corresponding to the pixels of the reduced image data and the pixel coordinates of the N × N pixels of the original image data are mapped, and the reduced image is mapped to the pixel coordinates. The feature amount corresponding to the pixel of the data is allocated. In this case, a pixel coordinate at which the assigned feature amount should be the maximum is obtained based on the feature amount extracted from the reduced image data. Then, the pixel coordinates are set as a center position, and the pixel coordinates around the center position are appropriately weighted so as to reduce the feature amount to be assigned as the distance from the center position increases, and the reduction is performed. The feature amount corresponding to the pixel of the image data is assigned.
[0028]
The feature amount corresponding to the pixel of the reduced image data is a result of extracting the feature amount of N × N pixels of the original image data. The feature amount corresponding to the pixel of the reduced image data is represented by N × N pixels. Since all the pixels do not necessarily have the same, the feature amount corresponding to the pixel of the reduced image data is assigned the largest to the pixel corresponding to the pixel coordinate of the center position, and the periphery of the center position is determined. The feature amount corresponding to the pixel of the reduced image data is allotted to a pixel corresponding to the pixel coordinates of. For this reason, it is possible to prevent an excessively large or small feature amount from being assigned to each pixel.
[0029]
Further, when data to be included in each pixel of the reduced image data is obtained by, for example, a linear interpolation method, the data is determined based on the weight of the distance of the data of the N × N pixels of the original image data. , The data is not used equally. For this reason, the feature amount corresponding to the pixel of the reduced image data is assigned the largest to the pixel corresponding to the pixel coordinate of the center position, and the pixel corresponding to the pixel coordinate around the center position is When the feature amount corresponding to the pixel of the reduced image data is assigned smaller, the feature amount to be assigned is determined according to a Gaussian distribution having a feature that rapidly decreases from the center position. That is, most of the extracted feature amount is assigned to a pixel corresponding to the pixel coordinate of the center position, and the feature amount is hardly assigned to pixels around the pixel, thereby This prevents the feature amount from being excessively assigned to the pixel.
[0030]
Furthermore, when the data which the pixels of the reduced image data should have is obtained by averaging the data of the N × N pixels of the original image data, the data is obtained by the N × N pixels of the original image data. Determined using data equally. For this reason, the feature amount corresponding to the pixel of the reduced image data is assigned the largest to the pixel corresponding to the pixel coordinate of the center position, and the pixel corresponding to the pixel coordinate around the center position is When the feature amount corresponding to the pixel of the reduced image data is assigned smaller, the feature amount to be assigned so as to linearly decrease radially from the center position is determined. That is, the extracted feature amount can be assigned little by little to each pixel.
[0031]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, the present invention will be described in detail with reference to the drawings showing the embodiments.
The present embodiment is a feature amount extraction method for extracting features of input image data and performing adaptive processing based on the extraction result, wherein the input image data is reduced to a predetermined size. When performing a desired feature extraction process on the reduced image data and performing an adaptive process on the original image data, a mapping between the target pixel and the extracted feature extraction result is performed. An example of a feature amount extraction method characterized in that an optimum process is selected and performed according to a mapped extraction result.
[0032]
When the image is reduced to the predetermined size, the pixel coordinates of the reduced image are set so that the pixel positions do not overlap with each other, based on the pixel coordinates of the original image according to the reduction ratio, and linear interpolation is performed from that position. It is characterized in that a reduced image is obtained. Further, when the image is reduced to the predetermined size, a reduced image is obtained by obtaining an average value of pixel values for every N × N pixels while maintaining the vertical and horizontal size ratio of the input image. .
[0033]
In addition, an edge detection process is performed as the desired feature amount extraction process, and one of a plurality of edge enhancement filters having different intensities is selected in order to perform an adaptive filter process according to the edge extraction result. I do. Further, a final edge strength is extracted by performing a shift operation on the filter result.
Further, when performing the mapping, in the N × N pixel size of the original image corresponding to one pixel of the reduced image, the value of the feature amount extraction processing result is changed from the central portion of the N × N pixel to the periphery. It is characterized by being lowered. Further, when the feature amount extraction result is reduced, when the linear interpolation method is used as the reduction method, the value of the extraction result is reduced in a Gaussian distribution from a central portion. Furthermore, when the feature value extraction result is lowered, when the average value of N × N pixels is used as a reduction method, the value of the extraction result is linearly lowered from the central portion. .
[0034]
Further, in the present embodiment, in an image processing apparatus including a feature amount extracting unit for extracting features of input image data and performing adaptive processing based on the extraction result, the feature amount extracting unit includes at least a feature amount extracting unit. A reducing means for reducing input image data to a predetermined size; a feature extracting means for performing desired feature extracting processing on the reduced image data; and an adaptive processing for the original image data. And a mapping unit that performs mapping between the target pixel and the extracted feature amount extraction result when performing the image processing. Further, an image forming apparatus provided with the image processing apparatus will be exemplified.
Further, a program for causing a computer to execute the feature amount extracting method, and a recording medium for storing the program in a computer-readable manner will be exemplified.
[0035]
Embodiment 1.
FIG. 1 is a block diagram illustrating a configuration of an image forming apparatus 1 according to Embodiment 1 of the present invention. The image forming apparatus 1 is a color copying apparatus, and includes a color image processing unit 12 as an image processing apparatus according to Embodiment 1 of the present invention, and a printer unit 14 as a unit that forms an image included in the image processing apparatus.
[0036]
In the figure, reference numeral 10 denotes a control unit of the image forming apparatus 1, which is connected to each unit of the apparatus such as a ROM 15 and a RAM 16 via a bus. The control unit 10 controls each unit of the apparatus according to a control program stored in the ROM 15, temporarily stores data generated at this time or data input from the outside in the RAM 16, and executes various processes.
The ROM 15 is a storage medium storing the program of the present invention, and a part of the storage area of the ROM 15 includes a plurality of edge enhancement filters (each edge enhancement filter) used when performing edge enhancement filter processing on original image data. An edge enhancement filter storage unit 15a that stores filter coefficients constituting a filter).
[0037]
The display unit 17 is controlled by the control unit 10 to display an operation state of the apparatus or an input instruction to the user, and the operation unit 18 includes a numeric keypad and a numeric keypad for the user to input numbers or symbols while viewing the display unit 17. It has a start key for inputting a formation start instruction, various function keys, and the like.
The scanner unit 11 includes a light source, a reflecting mirror, an image forming lens, a CCD (Charge Coupled Device), and the like, and converts a reflected light image from a document set in the scanner unit 11 into RGB (R: red, G: green). , B: blue) to generate original image data, and input the generated original image data to the color image processing unit 12.
[0038]
The color image processing unit 12 includes an A / D (analog / digital) conversion unit 101 and a shading correction unit 102, which will be described later, and includes an image processing control unit (not shown) and a plurality of registers. The image processing control unit is controlled by the control unit 10 to read a plurality of edge enhancement filters stored in the edge enhancement filter storage unit 15a in accordance with an image processing procedure of a program stored in the ROM 15, The color image processing unit 12 is controlled to perform image processing on the original image data input to the color image processing unit 12. Also, the original image data or the processed image data during the image processing generated at this time is stored in the register or temporarily saved in the image memory 13. Further, the extracted feature amount and processed image data subjected to various image processing are input to the image memory 13.
[0039]
The image memory 13 is composed of, for example, a dynamic RAM (DRAM), and stores processed image data output from the color image processing unit 12 or a feature amount extraction result.
The printer unit 14 includes an electrophotographic or inkjet printing unit, and forms an image on a sheet supplied to the printing unit by a paper feeding unit (not shown) using the processed image data output from the image memory 13. .
[0040]
FIG. 2 is a block diagram illustrating a configuration of the color image processing unit 12 included in the image forming apparatus 1.
Original image data (RGB analog signals) output from the scanner unit 11 is converted into RGB digital signals by an A / D conversion unit 101, and then the illumination system (light source) of the scanner unit 11 is converted by a shading correction unit 102. ), Various distortions that occur in the imaging system (reflecting mirror and imaging lens) and the imaging system (CCD) are eliminated.
The RGB digital signal (RGB reflectance signal) from which various distortions have been removed by the shading correction unit 102 is adjusted in color balance by the input gradation correction unit 103, and is adopted by the color image processing unit 12. Is converted into a signal (for example, a density signal) which is easy for the image processing system to handle.
[0041]
The original image data of the RGB digital signal output from the input tone correction unit 103 is converted into an LCC (L: luminance, C: chromaticity) color space by a color space conversion unit 104, and a feature amount extraction processing unit 105 Is input to
The original image data of the LCC signal output from the color space conversion unit 104 is converted into processed image data of the LCC signal by the feature amount extraction processing unit 105 and the spatial filter processing unit 106 and input to the scaling processing unit 107. .
[0042]
The feature amount extraction processing unit 105 includes a reducing unit 5a that generates reduced image data obtained by reducing the original image data, a feature amount extracting unit 5b that extracts a feature amount from the generated reduced image data, and the like. 3), the feature quantity extracted by the feature quantity extraction means 5b is temporarily saved in the image memory 13, and the spatial filter processing unit 106, the scaling processing unit 107, and the black generation and under color removal via the image memory 13. The data is input to the section 109 and the tone reproduction processing section 111. Further, the feature amount extraction processing unit 105 inputs the original image data output from the color space conversion unit 104 to the spatial filter processing unit 106 as it is.
Note that the feature amount extracted by the feature amount extraction processing unit 105 is an edge, a noise, a character, a halftone dot, a specific color, or the like. In the following, the edge strength and the isolated point are exemplified.
[0043]
Each unit to which the feature amount is input maps the pixel coordinates of the feature amount and the pixel coordinates of each pixel of the original image data, and assigns the feature amount to the pixel coordinates. (In the case of the spatial filter processing unit 106, the mapping unit 6a; see FIG. 3), and a generation unit that generates processed image data that has been subjected to predetermined processing based on the obtained new feature amount ( The spatial filter processing unit 106 includes a processed image data generating unit 6c (see FIG. 3).
[0044]
That is, the image memory 13 stores a feature amount extraction result having a data amount equal to the data amount of the reduced image data, and a feature amount extraction result having a data amount equal to the data amount of the original image data (the feature amount extraction result is Since the new feature amount extraction result obtained by mapping is generated by each of the above-described units, the storage capacity of the image memory 13 may be smaller than the storage capacity of the image memory included in the conventional image forming apparatus.
[0045]
The spatial filter processing unit 106 uses the original image data of the LCC signal output from the feature value extraction processing unit 105 and the feature value extraction result (for example, the edge strength extraction result), and based on the feature value extraction result, generates a digital filter ( In this case, spatial filter processing (edge enhancement filter processing) using the edge enhancement filter stored in the edge enhancement filter storage unit 15a is performed to correct the spatial frequency characteristics. Therefore, it is possible to prevent the image formed by the printer unit 14 from being blurred or degraded in graininess.
For example, the result of detection of a high edge strength by the feature amount extraction processing unit 105 indicates that the sharpness enhancement processing in the spatial filtering processing by the spatial Is increased. Further, the result detected as an isolated point by the feature extraction processing unit 105 is subjected to a low-pass filter processing for removing the isolated point component in the spatial filter processing unit 106.
[0046]
The processed image data of the LCC signal output from the spatial filter processing unit 106 forms various image forming conditions input by the user using the operation unit 18, the resolution of the printer unit 14, or an image at the scaling unit 107. In accordance with the size of the paper to be processed, a scaling process such as enlargement or reduction is performed, and the image data is converted into processed image data of a predetermined size. The processed image data of the LCC signal subjected to the scaling process is input to the color correction unit 108.
When using the feature extraction processing result at the time of scaling, for example, a plurality of scaling processing units 107, 107,... Are provided, and switching to an appropriate scaling processing unit 107 is performed according to the feature extraction result. The configuration is as follows.
The processed image data of the LCC signal output from the scaling processing unit 107 is converted into a CMY (C: cyan, M: magenta, Y: yellow) color space by the color correction unit 108, and the printer unit 14 has Color correction is performed according to the color characteristics. The processed image data of the CMY signal after the color correction is input to the black generation and under color removal unit 109.
[0047]
The processed image data of the CMY signal output from the color correction unit 108 is subjected to black generation processing for generating a K (black) signal from three CMY color signals in a black generation and under color removal unit 109. Next, a new CMY signal is generated by subtracting the K signal generated in the black generation processing from the CMY signal. That is, the processed image data of the CMY signals (three-color signals) is converted into processed image data of the CMYK signals (four-color signals).
[0048]
As an example of the black generation processing in the black generation and under color removal unit 109, there is a method of performing black generation using skeleton black. In this method, the input / output characteristic of the skeleton curve is set to y = f (x), the data input to the black generation and under color removal unit 109 is set to C, M, and Y, and the data output from the black generation and under color removal unit 109 is output. And C ′, M ′, Y ′, K ′, and a UCR (Under Color Removal) rate α (0 <α <1). Do.
[0049]
(Equation 1)

[0050]
When the result of the feature amount extraction processing is used at the time of black generation, the UCR rate is changed or the characteristic curve of f (x) is changed according to the result of the feature amount extraction.
[0051]
The processed image data of the CMYK signal output from the black generation and under color removal unit 109 is subjected to output gradation correction processing by an output gradation correction unit 110, and finally the image is converted to pixels by a gradation reproduction processing unit 111. The image data is subjected to a tone reproduction process (halftone generation process) for performing processing so as to be able to reproduce each tone separately.
The output gradation correction process is a process of converting the processed image data of the CMYK signal output from the black generation and under color removal unit 109 into a dot area ratio which is a characteristic value of the printer unit 14.
[0052]
In the tone reproduction processing unit 111, for example, a region where the edge strength is high detected by the feature extraction processing unit 105 is particularly suitable for high-frequency reproduction in order to enhance the reproducibility of black characters or color characters. Binarization or multi-value processing on the screen with the resolution is selected. Further, for a region where the edge detection result is small in the feature amount extraction processing unit 105, binarization or multi-value processing is performed on a screen that emphasizes tone reproducibility.
[0053]
The processed image data output from the color image processing unit 12 (gradation reproduction processing unit 111) is stored in the image memory 13, read out from the image memory 13 at a predetermined timing, and input to the printer unit 14.
[0054]
Note that the color image processing unit 12 inputs the original image data to the feature amount extraction processing unit 105 after being input to the input tone correction unit 103, and then to the spatial filter processing unit 106 and the scaling processing unit 107. Although it is configured to be input, it is not limited to such a configuration.
[0055]
For example, the spatial filter processing unit 106, the scaling processing unit 107, and the black generation and under color removal unit 109 do not use the feature amount extraction result, but use the feature amount extraction result only in the tone reproduction processing unit 111 (tone reproduction process In the case where only the unit 111 includes the generation unit, after the original image data is input to the input gradation correction unit 103 and the color space conversion unit 104, the spatial filter processing unit 106, the scaling processing unit 107, and the black generation lower color It may be configured to be input to the removal unit 109 and the output tone correction unit 110, then to the feature extraction processing unit 105, and finally to the tone reproduction processing unit 111. That is, after the original image data of the RGB signal is converted into the original image data of the LCC signal in the input tone correction unit 103, before the converted original image data is subjected to the image processing using the feature amount extraction result. Then, each unit of the color image processing unit 12 may be configured to perform the reduction process and the feature amount extraction process.
[0056]
FIG. 3 is a block diagram illustrating a configuration of the feature amount extraction processing unit 105 and the spatial filter processing unit 106.
The feature amount extraction processing unit 105 includes a reduction unit 5a and a feature amount extraction unit 5b.
Further, the spatial filter processing unit 106 reads the edge enhancement filter from the mapping unit 6a, the edge enhancement filter storage unit 15a of the ROM 15, and selects the edge enhancement filter based on the new feature amount obtained by the mapping unit 6a. 6b and a processed image data generating means 6c.
[0057]
The original image data of the LCC signal output from the color space conversion unit 104 is input to the reduction unit 5a of the feature amount extraction processing unit 105 for each pixel (one line, one plane, or the like). It is input to the processed image data generating means 6c of the spatial filter processing unit 106 via the quantity extraction processing unit 105.
The original image data input to the reduction means 5a is subjected to a reduction processing. In this case, the input of the original image data to the reduction unit 5a is continued until the data (hereinafter, referred to as pixel values) of the predetermined number of pixels is input to the reduction unit 5a. At this time, the data of the N × N square pixels is used as the predetermined number of pixel values, and the reduction process is started when the pixel values corresponding to the pixels are input.
[0058]
The reduction unit 5a performs a reduction process with high reproducibility of details of an image (hereinafter referred to as fine line reproducibility). Hereinafter, a reduction process for calculating an average value of pixel values and a reduction process using a linear interpolation method will be exemplified.
FIG. 4 is an explanatory diagram of the reduction processing used in the reduction means 5a. Each rectangle arranged in a 2 × 2 square indicates a pixel of the original image data, and the numerical value of each rectangle indicates the pixel value of each pixel. Further, a hatched rectangle in the center indicates pixels of reduced image data corresponding to 2 × 2 pixels.
Hereinafter, a case where the original image data is 8-bit data and takes a pixel value of 0 to 255 for each pixel will be exemplified. Further, in the present embodiment, the pixel values of the pixels of the reduced image data are obtained based on the pixel values of the 2 × 2 square pixels in the original image data.
[0059]
For example, when the size of the original image data is 640 × 480, the image is reduced vertically and horizontally by 縦 to obtain reduced image data having a size of 320 × 240. An average value of pixel values is obtained for every 2 × 2 pixels while maintaining the vertical and horizontal size ratios of the image data. In the case of FIG. 4 (pixel values 200, 100, 255, and 200), the average value (pixel value of reduced image data) is 189.
The above-described reduction processing is simple, and the reduced image data is obtained by using all the pixel values of the original image data. Therefore, it is possible to prevent the thin line information included in each pixel from being lost.
[0060]
When the average value of the pixel values is obtained for each N × N pixel, if the value of N is too large (the number of pixels is too large), the thin line information having only a few pixels in the whole N × N pixel is Due to the influence of other pixels, it is difficult to reflect the average value. Therefore, when setting the value of N, care must be taken so that the value does not become excessive.
[0061]
FIG. 5 is an explanatory diagram of the reduction processing by the linear interpolation method used in the reduction means 5a. Each rectangle arranged in a 2 × 2 square indicates a pixel of the original image data. Also, let the pixel coordinates of the upper left pixel be (x, y) and the pixel value be P (x, y). Furthermore, the hatched rectangle at the center indicates a pixel of reduced image data corresponding to 2 × 2 pixels of the original image data, the pixel coordinates of the pixel are (u, v), and the pixel value is P (u). , V). In this case, u and v are x <u <x + 1 and y <v <y + 1.
[0062]
When the pixel value of the pixel of the reduced image data is obtained based on the pixel value of a 2 × 2 square pixel in the original image data, in the reduction processing using the linear interpolation method, linear interpolation is performed using four surrounding points. Use the interpolation method. In this case, the pixel value of the pixel of the reduced image data is the distance between the pixel of the reduced image data and the four pixels in the original image data (the pixel coordinates of the pixel of the reduced image data and the pixel coordinates of the four pixels). The difference is calculated by multiplying the pixel values of the four pixels by a weighting coefficient proportional to the difference between the four pixels.
For this reason, the pixel value P (u, v) that the pixel of the reduced image data should have is obtained by the following equation.
[0063]

[0064]
For example, when the pixel coordinates of (x, y) are (100, 100), the pixel coordinates of (u, v) are set to (100.5, 100.5), and the pixel coordinates of FIG. When P (u, v) is calculated by rounding to the first decimal place,

It becomes.
[0065]
FIG. 6 is a schematic diagram showing reduced image data by the conventional bilinear method and reduced image data by the bilinear method used in the reduction means 5a. In the figure, (a) is original image data, and (b) and (c) are reduced image data obtained by reducing the original image data by a bilinear method. However, each pixel value in FIG. 6B is calculated by substituting u = x and v = y into Expression (2), and each pixel value in FIG. It is calculated by substituting u = x + 0.5 and v = y + 0.5.
[0066]
In FIG. 6B, the pixel coordinates of each pixel of the reduced image data match the pixel coordinates of the upper left pixel among the 2 × 2 pixels of the original image data. For this reason, the reduced image data is equal to the reduced image data obtained by simply subsampling the pixels of the original image data and reducing the pixel values of the 2 × 2 pixels other than the upper left pixel. It is not reflected in the pixel value of the image data.
In FIG. 6C, the pixel coordinates of each pixel of the reduced image data are different from the pixel coordinates of 2 × 2 pixels of the original image data. Therefore, the pixel value of the 2 × 2 pixel is reflected on the pixel value of the reduced image data.
The difference between the pixel coordinates of each pixel of the reduced image data and the pixel coordinates of N × N pixels of the original image data is preferably 0.5 pixels or less. Thus, it is possible to prevent the loss of the thin line information, perform the reduction process, and obtain reduced image data.
[0067]
FIG. 7 is a schematic diagram showing an example of original image data, and FIG. 8 is a schematic diagram showing reduced image data obtained by reducing the original image data shown in FIG. 7 by the reducing means 5a.
In the present embodiment, calculation is performed regardless of whether the original image data is subjected to the reduction processing for obtaining the average value of the pixel values or the reduction processing using the bilinear method. The results are equal. For this reason, reduced image data obtained by reducing the original image data of 8 × 8 pixels in FIG. 7 is reduced image data of 4 × 4 pixels in the central part of FIG.
The 20 pixels on the outer side in FIG. 8 are reduced image data obtained by using original image data for pixels (not shown) outside the original image data of 8 × 8 pixels. That is, the whole of FIG. 8 corresponds to a reduced image obtained by reducing the original image data of 12 × 12 pixels.
[0068]
As described above, after the reduced image data is obtained from the original image data by the reduction unit 5a, the feature amount is extracted from the reduced image data by the feature amount extraction unit 5b.
In this case, the feature amount extraction processing by the feature amount extraction means 5b may be started after obtaining the pixel values of all the pixels of the reduced image data, and the number of pixel values required for the feature amount extraction processing is obtained. It may be started at the time.
Hereinafter, an edge strength extraction process using a Laplacian filter will be exemplified.
[0069]
FIG. 9 is a schematic diagram of a Laplacian filter used in the feature amount extracting means 5b. The Laplacian filter is a filter having a kernel size of 3 × 3, and has a center of 8 and a periphery of −1.
When the Laplacian filter is used to perform a filtering process on the reduced image data by the feature amount extracting unit 5b to extract the edge strengths, the extracted edge strength values (hereinafter, referred to as edge strength values) are −2040 to −2040. 2040 (decimal number). Next, the edge strength value is shifted right by 4 bits. At this time, the edge strength value becomes -128 to 127 (decimal number). Therefore, the edge strength can be represented by an 8-bit value.
Each edge intensity value corresponds to the pixel coordinates of each pixel of the reduced image data.
[0070]
The edge portion is where the sign of the edge strength value is switched, and the difference between the edge strength values whose signs are switched is determined as a new edge strength.
Therefore, the sign of each edge strength value and the edge strength value corresponding to each of the pixel coordinates near eight of the pixel coordinates corresponding to the edge strength value are checked, and a new edge strength is obtained for the place where the sign is switched. The point where the obtained new edge strength value is the maximum is set as the center position of the edge strength extraction result. Where the sign does not change in the vicinity of 8, the extraction result is set to 0, and the sum of the absolute values of the peripheral edge strength values is determined for the portion determined to be an edge. If the sum of the absolute values exceeds 255, it is rounded to 255.
[0071]
FIG. 10 is a schematic diagram showing a feature amount extraction result obtained by extracting a feature amount from the reduced image data shown in FIG. 8 by the feature amount extraction unit 5b.
9A shows a result obtained by performing a filtering process on the 4 × 4 pixels at the center of the reduced pixel data by the feature amount extracting means 5b using the Laplacian filter shown in FIG. 9, and FIG. The result obtained by shifting the result of the filter processing to the right by 4 bits is shown. (C) shows an edge strength extraction result input to the image memory 13 from the feature amount extracting unit 5b after the process by the feature amount extracting unit 5b.
[0072]
In the edge strength extraction results shown in FIG. 10B, each edge strength extraction result is compared with eight neighboring extraction results. For example, assuming that the upper left pixel coordinate is (sx, sy), the sign of (sx + 1, sy) is reversed between (sx + 2, sy) and (sx + 2, sy + 1), and the edge is at that part. Will exist. Also, the center part of the edge strength extraction result includes the absolute value of the difference between (sx + 2, sy) and (sx + 1, sy) and the absolute value of the difference between (sx + 2, sy + 1) and (sx + 1, sy). , That is, the direction of (sx + 2, sy).
Next, for the portion extracted as the edge portion, the sum of the absolute values with the neighboring eight pixels is calculated, and the value of the portion other than the edge portion is set to 0, and the edge strength corresponding to each pixel is obtained.
[0073]
The edge strength extraction result obtained as described above and the data of the center position of the edge strength extraction result are temporarily saved in the image memory 13. Next, mapping processing is performed by the mapping means 6a of the spatial filter processing unit 106.
In the present embodiment, when the edge strength extraction result is input to each unit, the edge strength extraction result stored in the image memory 13 is used (the edge strength extraction result is input to each unit via the image memory 13). Although the processing using the edge strength extraction result is performed at only one place and continues after the feature quantity extraction processing, the edge strength extraction result and the data of the center position are temporarily stored in the image memory 13. It is not necessary to retreat, and the processing may be performed using the edge strength extraction result and the data of the center position stored in the register provided in the color image processing unit 12. In this case, image processing can be speeded up.
[0074]
The mapping process performed on the feature amount extraction result includes the feature amount extraction result temporarily saved in the image memory 13 or the feature amount extraction result stored in the register provided in the color image processing unit 12 and the original image data. And correspond.
The feature amount of one pixel of the feature amount extraction result extracted from the reduced image data obtained by reducing the original image data at a reduction ratio of 1/2 corresponds to the feature amount of 2 × 2 pixels of the original image data. Therefore, pixel coordinates of 2 × 2 pixels corresponding to the pixel coordinates in the original image data are calculated from the pixel coordinates of the feature amount of the feature amount extraction result, and the calculated pixel coordinates are compared with the feature of the one pixel. Sprinkle the amount.
[0075]
Assuming that the pixel coordinates of the feature amount as a result of the feature amount extraction are (sx, sy), the corresponding pixel coordinates (x, y) of 2 × 2 pixels are calculated using the following equations (4) and (5), respectively. Ask.
x = sx × 2 + a (4)
y = sy × 2 + b (5)
Here, a and b are integers satisfying 0 ≦ a, b <(reciprocal of the reduction ratio １ ／), and in this case, a, b = 0,1.
At this time, the four pixel coordinates corresponding to (sx, sy) are (sx × 2, sy × 2), (sx × 2 + 1, sy × 2), (sx × 2, sy × 2 + 1), (sx × 2 + 1, sy × 2 + 1).
[0076]
Next, a feature amount is assigned to the mapped pixel coordinates according to the method of the reduction process used by the reduction unit 5a. In this case, the feature amount assigned corresponding to the center position of the edge strength extraction result is the highest, and the feature amount assigned corresponding to the pixel coordinates around the center position is assigned corresponding to the center position. Each feature amount is assigned so as to be lower than the assigned feature amount. In this case, when the reduction process using the linear interpolation method is performed, the feature amount is respectively increased according to a function distribution having a feature in which the value is high only at the center position in a Gaussian distribution, and the value sharply decreases toward the periphery. Sprinkle. In addition, when the reduction processing for obtaining the average value of the pixel values is performed, the characteristic amount is assigned according to a function distribution having a characteristic in which the value of the center position is high and the value decreases linearly toward the periphery. Hit it.
[0077]
Note that, in this case, a feature amount to be assigned may use a filter having a coefficient such that it is Gaussian-distributed or linearly reduced, or a calculation formula corresponding to the function distribution may be used. The coefficients may be stored in the ROM 15 as a table.
[0078]
FIG. 11 is a schematic diagram showing a feature amount extraction result obtained by mapping the feature amount extraction result shown in FIG. 10 and assigning the feature amount by the mapping means 6a. In the figure, (a) shows the feature amount extraction result, (b) shows the weighting factor at the time of mapping, (c) shows the feature amount weighted to the feature amount and assigned to each pixel, and (d) shows The original image data corresponding to the feature amount is shown.
The mapping means 6a performs mapping between the pixel coordinates corresponding to the extracted edge strength and the pixel coordinates of each pixel of the original image data. In this case, as described above, the pixel coordinates (sx, sy) of the four pixels of the original image data are compared with the pixel coordinates (sx, sy) of the edge strength extraction result. × 2), (sx × 2, sy × 2 + 1), and (sx × 2 + 1, sy × 2 + 1).
[0079]
Also, a weighting factor at the time of mapping is obtained according to the center position of the edge strength extraction result. For example, as shown in FIG. 11B, the weight coefficient at the time of mapping for (sx + 1, sy) is (sx + 1, sy) on the right side since the center position of the edge strength extraction result is in the (sx + 2, sy) direction. And (sx + 1, sy + 1) are pixel coordinates close to the center position, and 0.75 is used as a weight coefficient corresponding to the pixel coordinates, and (sx, sy) and (sx, sy + 1) on the left are pixel coordinates far from the center position. And 0.25 is used as a weight coefficient corresponding to the pixel coordinates.
[0080]
In this case, it is preferable to use a Gaussian distribution weighting coefficient when the reduction processing using the bilinear method is performed, and to use a linear weighting coefficient when performing the reduction processing for calculating the average value of the pixel values. In the embodiment, since the same reduction result is obtained regardless of which reduction processing is performed, a linear weighting coefficient that can easily calculate a weighting coefficient is exemplified. That is, the weight coefficient is set so that the value decreases from the center position of the feature amount extraction result toward the opposite side where the center of 2 × 2 pixels is point-symmetric.
[0081]
As shown in FIG. 11, the feature amount extraction result is multiplied by a weighting coefficient at the time of mapping to obtain a feature amount extraction result for each pixel of the original image data. Next, an adaptive filter process is performed by comparing the value with a preset threshold value.
Note that FIG. 11 shows a mapped feature amount extraction result for easy understanding of the mapping result. For this reason, it seems that the feature extraction result is enlarged, but in actual processing, the result of multiplying each edge intensity value by the weighting factor at the time of mapping is obtained, and a filter coefficient is selected based on the value. However, since the filtering process is performed, a feature amount extraction result having the same size as the original image data is not held.
[0082]
Next, adaptive image processing is performed using the feature amount extraction result assigned to the mapped pixel coordinates. In the following, a case where adaptive filter processing using the edge strength extraction result is performed will be exemplified.
First, the magnitude of the extraction result in the part where the sign of the edge strength extraction result is switched is classified by a plurality of thresholds, and when the magnitude is equal to or larger than each threshold, the edge enhancement processing is performed by the filter coefficient corresponding to the threshold. For example, three edge enhancement filters with different edge enhancement degrees are stored in the edge enhancement filter storage unit 15a, and the edge enhancement filters are read into the filter selection unit 6b.
[0083]
FIG. 12 is a schematic diagram illustrating an example of the edge enhancement filter stored in the edge enhancement filter storage unit 15a.
In the figure, (a) is the strongest edge enhancement filter, (b) is the second strongest edge enhancement filter, and (c) is the third strongest edge enhancement filter.
The filter selection means 6b uses three thresholds 192, 128, and 64 to classify 0 to 255 into four ranges substantially equally.
When the edge strength value is 192 or more and 255 or less, the filter selecting means 6b selects the strongest edge emphasis filter (a). When it is 128 or more and 191 or less, the second strongest edge emphasis filter (b) is selected. When it is 64 or more and 127 or less, the third strongest edge emphasis filter (c) is selected. Further, when the value is 0 or more and 63 or less, the edge enhancement filter is not selected.
[0084]
Next, in the processed image data generating means 6c, when the edge strength value is 64 or more and 255 or less, the original image data is subjected to filter processing using the edge enhancement filter selected by the filter selecting means 6b, and the processed image data is processed. Is generated and output to the scaling unit 107. In the case of 0 or more and 63 or less, the filter processing is not performed, and the original image data is output to the scaling processing unit 107 as processed image data.
As described above, adaptive edge enhancement filter processing is performed according to the extracted edge strength.
[0085]
FIG. 13 is a schematic diagram illustrating processed image data obtained by performing a filtering process on the original image data by using the edge enhancement filter by the processed image data generating unit 6c.
In the above image processing, image processing for one line of reduced image data is performed while shifting the pixel of interest one pixel at a time. When the processing for one line is completed, image processing for the next one line is performed, and finally, image processing is performed for all pixels.
[0086]
FIG. 14 is a flowchart illustrating a procedure of image processing (feature amount extraction processing) of the feature amount extraction processing unit 105.
The input of the original image data output from the color space conversion unit 104 is received. (S11). In this case, the original image data is input to the reduction means 5a of the feature amount extraction processing unit 105, and is delayed through the feature amount extraction processing unit 105, and processed image data of the spatial filter processing unit 106 at a predetermined timing. It is input to the generation means 6c (see S35 described later).
Next, the input original image data is subjected to a reduction process for obtaining an average value of 2 × 2 pixels (or using a bilinear method) at a reduction ratio of に て by a reduction unit 5 a, and the reduced image data is It is generated (S12).
[0087]
The generated reduced image data is subjected to a filtering process using a Laplacian filter (S13), and a 4-bit right shift operation is performed on the calculated edge strength (S14), and each edge strength corresponds to the edge strength. The sign of the edge strength corresponding to each of the pixel coordinates in the vicinity of the pixel coordinates 8 is checked, and the difference in the edge strength value is extracted as a new edge strength at the place where the sign is switched (S15), and the new edge obtained is obtained. The position where the intensity is maximum is set as the center position of the edge intensity extraction result (S16).
[0088]
The obtained edge strength extraction result and the data of the center position of the edge strength extraction result are output to the image memory 13 (S17). Finally, all the pixels of the original image data are input to the feature value extraction processing unit 105. It is determined whether or not the processing of S12 to S17 has been completed (S18). If the processing has not been completed (NO in S18), the processing returns to S11 to accept the input of the original image data. If the above process has been completed (YES in S18), the feature amount extraction process ends.
[0089]
FIG. 15 is a flowchart illustrating a procedure of image processing (processed image data generation processing) of the spatial filter processing unit 106.
First, three edge enhancement filters are read (S31).
The edge intensity input from the image memory 13 is received for each pixel (S32).
The edge intensity of one pixel is mapped and assigned to 2 × 2 pixels (S33), each assigned edge intensity is compared with a threshold, and an edge enhancement filter is selected based on the comparison result (S34).
[0090]
Next, input of the original image data via the feature amount extraction processing unit 105 is received (S35). Filter processing is performed on each of the pixels corresponding to the 2 × 2 pixels of the input original image data using the edge enhancement filter selected in S34 to generate processed image data (S36).
Finally, the generated processed image data is output to the scaling processing unit 107 (S37), and it is determined whether or not the processing of S32 to S37 has been completed for all pixels of the feature amount extraction result (S38). If the process has not been completed (NO in S38), the process returns to S32 to accept the input of the edge strength, and if completed (YES in S38), the processed image data generation process ends.
[0091]
The image forming apparatus 1 as described above extracts the feature amount using the reduced image data obtained by reducing the original image data to 縦 in length and width, so that the data amount to be handled when extracting the feature amount is 1 / It becomes 4. For this reason, the time required for the process of extracting the feature amount is reduced, and the storage capacity required by the image memory 13 is reduced to １ ／.
Further, when extracting the edge strength in all directions, since the edge strength is extracted using the Laplacian filter, it is not necessary to use a plurality of filters as in the case of extracting the edge strength using a Sobel filter, for example. .
[0092]
In the present embodiment, as an image forming apparatus, a copying apparatus that reads an original with a built-in scanner unit and prints the original with a printer unit has been illustrated, but receives original image data transmitted from the outside, and receives the received original image data. A copying apparatus that generates processed image data from image data and prints the processed image data with a printer unit may be used. Further, it may be a facsimile machine or a multifunction machine of a copy machine and a facsimile machine.
[0093]
Further, in the present embodiment, when processing image data is generated, a case has been exemplified in which filter processing (in this case, edge enhancement filter processing based on edge strength) is performed on original image data based on a feature amount. Alternatively, color correction or resolution conversion may be performed based on the feature amount.
[0094]
When performing color correction such that RGB data obtained by reading an image with a scanner unit is converted into CMYK data for printing by a printer unit, data (color information) of original image data is used as a chromaticity value of each pixel. a * or b * (chromaticity defined in the International Commission on Illumination CIE1976L * a * b * space) or the like is extracted as a feature amount, and the color gamut is compressed and color corrected based on the extraction result.
[0095]
Also, human skin, blue sky, or greenery of vegetation are not reproduced with colors faithful to the original image so that humans can preferably feel it, and colors close to the colors memorized by humans ( When performing color correction to reproduce with memory color, it is determined whether or not the color is a memory color based on the extraction result. If the color is a memory color, predetermined memory color processing (saturation is reduced for skin color, For colors other than the flesh color, a process of increasing the saturation and brightness is performed, and when the color is not a memory color, a normal color correction process is performed.
[0096]
When performing resolution conversion, for example, an edge, a color of a pixel, or a run length (the number of consecutive pixels having the same density) is extracted as a feature amount of a character. To convert the resolution. In this case, the edge reproducibility of the character can be improved.
Further, not only a plurality of edge enhancement filters but also other filters or appropriate processing means among processing means such as color correction or resolution conversion are selected based on the extracted feature amount, and the selected processing means is selected from the original image. It may be configured to perform processing on data to generate processed image data.
[0097]
Further, in this embodiment, the edge strength extraction processing using eight neighboring pixels is used as the feature extraction processing, but other peripheral pixel numbers may be used. Further, the edge enhancement filter processing is described as an adaptive image processing using the feature amount extraction result, but it goes without saying that the present invention can be applied to other processing.
Further, in the present embodiment, a printer unit for forming an image on a sheet is exemplified as a unit for forming an image using processed image data, but a display unit for forming an image on a display may be used.
[0098]
Embodiment 2.
FIG. 16 is a block diagram showing a configuration of the image processing device 2 according to Embodiment 2 of the present invention.
In the figure, reference numeral 20 denotes a control unit of the image processing apparatus 2. The control unit 20 includes a display unit 21 using a CRT or a liquid crystal display and an operation unit 22 using a mouse or a keyboard. Each part is connected via a bus.
The control unit 20 controls each unit of the apparatus according to a program stored in the auxiliary storage unit 26 or the ROM 24 using a hard disk, and executes various processes.
[0099]
The external storage unit 27 is, for example, a CD-ROM drive, and is controlled by the control unit 20 to execute a portable recording medium such as a CD-ROM (for example, a program according to the present embodiment and a program that executes the program. Information is read from the recording medium 3) on which the edge enhancement filter and data used in the case are recorded, and stored in the auxiliary storage unit 26.
The LAN interface 23 is connected to the local line L to which the scanner device 41 and the printer device 42 are connected. The control unit 20 controls the scanner device 41 and the printer device 42 via the local line L and the LAN interface 23. Sends and receives image data.
[0100]
When image data is received from the scanner device 41 via the LAN interface 23, the control unit 20 stores the received image data in the RAM 25 or the auxiliary storage unit 26. Further, in accordance with a user's instruction, image processing is performed on the image data stored in the RAM 25 in accordance with the image processing procedure of the program to generate processed image data. Further, in accordance with a user's instruction, the generated processed image data is stored in the auxiliary storage unit 26 or transmitted to the printer device 42.
[0101]
FIG. 17 is a flowchart illustrating an image processing procedure of the image processing apparatus 2.
The control unit 20 causes the RAM 25 to store the original image data received via the LAN interface 23 or the original image data read from the auxiliary storage unit 26 (S51).
The original image data stored in the RAM 25 is subjected to a reduction process for obtaining an average value of 2 × 2 pixels (or using a bilinear method) at a reduction ratio of 、 to generate reduced image data (S52). In this case, reduced image data is generated without processing / erasing the original image data. In addition, before performing the reduction processing, a predetermined image processing (for example, a color space conversion processing) may be performed.
[0102]
The generated reduced image data is subjected to a filtering process using a Laplacian filter (S53), and a 4-bit right shift operation is performed on the calculated edge strength (S54), and each edge strength corresponds to the edge strength. The sign of the edge strength corresponding to each of the pixel coordinates in the vicinity of the pixel coordinates 8 is checked, and the difference in the edge strength value is extracted as a new edge strength at the place where the sign is switched (S55), and the new edge obtained is obtained. The position where the intensity is maximum is set as the center position of the edge intensity extraction result (S56). After the edge strength extraction result is obtained, the reduced image data may be deleted.
Next, three edge enhancement filters are read from the auxiliary storage unit 26 into the RAM 25 (S57).
[0103]
Each edge intensity is mapped and assigned to 2 × 2 pixels (S58), each assigned edge intensity is compared with a threshold, and an edge enhancement filter is selected based on the comparison result (S59).
Filter processing is performed on each pixel corresponding to the 2 × 2 pixels of the original image data stored in the RAM 25 using the edge enhancement filter selected in S59 to generate processed image data (S60). In this case, the original image data may be processed / erased. Further, predetermined image processing (for example, scaling processing or color correction processing) may be performed on the generated processed image data.
Finally, the generated processed image data is output (S61). In this case, the processed image data is transmitted to the printer device 42, stored in the auxiliary storage unit 26, or displayed on the display unit 21.
[0104]
The image processing apparatus 2 as described above can generate processed image data from original image data in the same manner as the image forming apparatus 1 of the first embodiment.
In the present embodiment, the image processing method is recorded on a computer-readable recording medium that records a program to be executed by a computer. As a result, a recording medium on which a program for performing the feature extraction method is recorded can be provided in a portable manner.
[0105]
In the present embodiment, as an image processing apparatus, original image data generated by reading a document with an external scanner device is received, and processed image data is generated from the received original image data and sent to a printer device. Although the transmitting personal computer has been exemplified, the present invention is not limited to such a configuration, and may be configured to receive original image data output from a digital camera, for example. Further, the processing image data may be generated from the original image data generated using the paint software installed in the built-in auxiliary storage unit, and output (displayed) to the display unit.
In addition, the configuration may be such that the program is not only installed in the auxiliary storage unit 26 and executed, but also directly read from the recording medium 3 and executed.
[0106]
As a recording medium, a tape system such as a magnetic tape or a cassette tape, a magnetic disk such as a flexible disk or a hard disk, a disk system of an optical disk such as a CD-ROM / MO / mini disk / DVD, an IC card (including a memory card) ) / Fixed memory such as a card system such as an optical card, or a semiconductor memory such as a mask ROM, an EPROM (Erasable Programmable Read Only Memory), an EEPROM (Electrically Erasable Programmable Read Only Memory), or a semiconductor memory such as a flash memory. It may be.
[0107]
Further, in the present embodiment, since a system configuration capable of connecting to a communication network including the Internet is also possible, a medium that carries the program in a fluid manner such that the program is downloaded from the communication network may be used. When the program is downloaded from the communication network as described above, the download program may be stored in the main device in advance, or may be installed from another recording medium.
[0108]
【The invention's effect】
According to the image processing method, the image processing apparatus, the image forming apparatus including the image processing apparatus, the computer program for using a computer as the image processing apparatus, and the recording medium for recording the computer program, the feature amount can be reduced. Since the time required for the extraction processing can be reduced, the time required for the entire image processing can also be reduced. In addition, since a process of extracting a feature amount can be performed using a memory having a small storage capacity, an apparatus with reduced memory cost can be configured. For example, when using reduced image data with a data amount of 1/4, the required storage capacity is 1/4, so that the memory cost can be reduced to 1/4. Furthermore, overall performance can be improved.
[0109]
Further, the present invention is particularly suitable for edge enhancement.
Further, the computer program can cause a computer to execute an image processing method whose overall performance is improved as compared with the related art, and the image processing method can be general-purpose.
Further, the present invention has an excellent effect, for example, the recording medium can easily supply the computer program to a computer.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a configuration of an image forming apparatus according to Embodiment 1 of the present invention.
FIG. 2 is a block diagram illustrating a configuration of a color image processing unit included in the image forming apparatus according to Embodiment 1 of the present invention.
FIG. 3 is a block diagram illustrating a configuration of a feature amount extraction processing unit and a spatial filter processing unit included in the image processing apparatus according to the first embodiment of the present invention.
FIG. 4 is an explanatory diagram of a reduction process used in a reduction unit included in a feature amount extraction processing unit provided in the image processing apparatus according to the first embodiment of the present invention;
FIG. 5 is an explanatory diagram of a reduction process by a linear interpolation method used in a reduction unit included in a feature amount extraction processing unit provided in the image processing apparatus according to the first embodiment of the present invention.
FIG. 6 is a schematic diagram showing reduced image data by a conventional bilinear method and reduced image data by a bilinear method used in the image processing method according to the first embodiment of the present invention;
FIG. 7 is a schematic diagram illustrating an example of original image data.
FIG. 8 is a schematic diagram illustrating reduced image data obtained by reducing original image data by a reducing unit included in a feature amount extraction processing unit included in the image processing apparatus according to the first embodiment of the present invention;
FIG. 9 is a schematic diagram of a Laplacian filter used in a feature extraction unit included in a feature extraction processing unit included in the image processing apparatus according to the first embodiment of the present invention.
FIG. 10 is a schematic diagram illustrating a feature amount extraction result obtained by extracting a feature amount from reduced image data by a feature amount extraction unit included in a feature amount extraction processing unit included in the image processing apparatus according to the first embodiment of the present invention; is there.
FIG. 11 maps a feature amount extraction result extracted by the feature amount extracting unit and assigns a feature amount by a mapping unit included in a spatial filter processing unit included in the image processing apparatus according to the first embodiment of the present invention; FIG. 7 is a schematic diagram showing a feature amount extraction result.
FIG. 12 is a schematic diagram illustrating an example of an edge enhancement filter stored in an edge enhancement filter storage unit included in the image forming apparatus according to Embodiment 1 of the present invention;
FIG. 13 is a diagram illustrating a process in which a processed image data generating unit included in a spatial filter processing unit included in the image processing apparatus according to the first embodiment of the present invention performs a filtering process on original image data using an edge enhancement filter; It is a schematic diagram which shows image data.
FIG. 14 is a flowchart illustrating an image processing procedure of a feature amount extraction processing unit included in the image processing apparatus according to the first embodiment of the present invention.
FIG. 15 is a flowchart illustrating an image processing procedure of a spatial filter processing unit included in the image processing device according to the first embodiment of the present invention.
FIG. 16 is a block diagram illustrating a configuration of an image processing apparatus according to Embodiment 2 of the present invention.
FIG. 17 is a flowchart illustrating an image processing procedure of the image processing apparatus according to the second embodiment of the present invention.
[Explanation of symbols]
11 Scanner section
12 Color image processing unit
13 Image memory
14 Printer section
15a Edge enhancement filter storage unit
5a Reduction means
5b Feature extraction means
6a Mapping means
6c Processing image data generation means

Claims (12)

An image processing method for extracting a feature amount from image data and generating processed image data processed based on the extracted feature amount,
In order to obtain reduced image data obtained by reducing image data, based on data of a plurality of pixels of the image data, obtain data that pixels of reduced image data should have,
From the obtained reduced image data, a feature amount is extracted corresponding to the pixel coordinates of each pixel of the reduced image data,
The pixel coordinates of the extracted feature amount and the pixel coordinates of each pixel of the image data are mapped, and a new feature amount obtained by assigning the feature amount to the pixel coordinates is obtained.
An image processing method, characterized by generating processed image data processed based on a new feature amount obtained. Determining the pixel coordinates of the N × N pixels (N: a natural number of 2 or more) of the image data so that the pixel coordinates of the pixels of the reduced image data are different from each other; 2. The image processing according to claim 1, wherein data to be possessed by the pixel is obtained based on the data of the pixel and the difference between the pixel coordinates of the pixel and the pixel coordinates of the pixel of the reduced image data. Method. 2. The image processing method according to claim 1, wherein data that pixels of the reduced image data should have is obtained by averaging data of N × N pixels of the image data. When the feature amount is edge strength, a plurality of edge enhancement filters each having a different strength are prepared, and pixel coordinates of edge strength extracted from the reduced image data and pixel coordinates of each pixel of the image data are calculated. By mapping, a new edge strength is obtained by assigning the edge strength to the pixel coordinates, and among the edge enhancement filters, the processed image data is processed using an edge enhancement filter selected based on the new edge strength. The image processing method according to claim 1, wherein the value is obtained. 5. The image processing apparatus according to claim 4, wherein an edge intensity obtained by performing a shift operation on the edge intensity extracted from the reduced image data is obtained, and pixel coordinates of the obtained edge intensity are mapped to pixel coordinates of each pixel of the image data. The image processing method according to 1. When the pixel coordinates of the feature amount corresponding to the pixel of the reduced image data and the pixel coordinates of the N × N pixels are mapped and the feature amount is assigned to the pixel coordinates, the assigned feature amount is A pixel coordinate to be the maximum is obtained based on the extracted feature quantity, and a feature quantity larger than a feature quantity assigned to a pixel coordinate distant from the neighborhood of the pixel coordinate is determined for a pixel coordinate in the vicinity of the obtained pixel coordinate. The image processing method according to claim 2, wherein the image is assigned. The feature amount assigned to each pixel coordinate is assigned such that the feature amount is reduced in a Gaussian distribution from pixel coordinates in the vicinity of the pixel coordinates to pixel coordinates separated from the vicinity of the pixel coordinates. The image processing method according to claim 6. The feature quantity is assigned so that the feature quantity assigned to each pixel coordinate decreases linearly from the pixel coordinates near the pixel coordinates to the pixel coordinates separated from the pixel coordinates. Item 7. The image processing method according to Item 6. An image processing apparatus that extracts a feature amount from image data and generates processed image data that is processed based on the extracted feature amount,
Reducing means for obtaining data that pixels of reduced image data should have, based on data of a plurality of pixels of the image data, to obtain reduced image data obtained by reducing image data;
Feature amount extracting means for extracting a feature amount from the obtained reduced image data in correspondence with the pixel coordinates of each pixel of the reduced image data;
Mapping means for mapping the pixel coordinates of the extracted feature quantity and the pixel coordinates of each pixel of the image data to obtain a new feature quantity by assigning the feature quantity to the pixel coordinates;
A generating unit configured to generate processed image data processed based on the obtained new feature amount. An image forming apparatus comprising: the image processing apparatus according to claim 9; and means for forming an image using the processed image data generated by the image processing apparatus. In a computer program for causing a computer to extract a feature amount from the image data and generate processed image data processed based on the extracted feature amount,
Causing the computer to determine data that a pixel of the reduced image data should have, based on data of a plurality of pixels of the image data, so as to determine reduced image data obtained by reducing the image data;
Extracting a feature amount from the obtained reduced image data in correspondence with the pixel coordinates of each pixel of the reduced image data;
Mapping the pixel coordinates of the extracted feature quantity and the pixel coordinates of each pixel of the image data to obtain a new feature quantity by assigning the feature quantity to the pixel coordinates;
Generating processed image data that has been processed based on the obtained new feature amount. The computer, to extract the feature amount from the image data, to record a computer program for generating a processed image data processed based on the extracted feature amount, in a computer-readable recording medium,
Causing the computer to determine data that a pixel of the reduced image data should have, based on data of a plurality of pixels of the image data, so as to determine reduced image data obtained by reducing the image data;
Extracting a feature amount from the obtained reduced image data in correspondence with the pixel coordinates of each pixel of the reduced image data;
Mapping the pixel coordinates of the extracted feature amount and the pixel coordinates of each pixel of the image data, and calculating a new feature amount by assigning the feature amount to the pixel coordinates;
Generating a processed image data based on the obtained new feature amount. A computer-readable recording medium characterized by recording a computer program.

Images