JP4247889B2 - Image processing device - Google Patents

Description

図５は、地肌レベル予測回路404のより詳細な回路構成を例示する。図５を参照すると、地肌レベルバッファ回路403から入力された地肌レベル信号501は、Ｆ／Ｆ（フリップ・フロップ）502に入力される。また、Ｆ／Ｆ502から出力された地肌レベル信号503はＦ／Ｆ504に入力され、更に、Ｆ／Ｆ504から出力された地肌レベル信号505はＦ／Ｆ506に入力され、Ｆ／Ｆ506からは地肌レベル信号507が出力される。ここで、各Ｆ／Ｆ502,504,506には、上述した区間の切替りに対応した区間クロック508が入力されており、地肌レベル信号503,505,507は、それぞれ上述した右隣接区間、注目領域対応区間、左隣接区間の地肌レベルに対応する。これらの地肌レベル信号は積和回路509に入力され、積和回路509は、上記式(1)に示したような積和演算を各色成分毎に行う。これにより、注目領域の地肌レベルを予測した予測地肌レベル信号510を生成できる。
【００２３】
図４に戻ると、特徴量抽出回路405は、入力された原稿画像信号の特徴量を抽出する回路で、抽出した特徴量は、原稿の地肌部分の画像信号であるか否かを判定し、予測地肌レベルを補正するためのデータとして、地肌レベル決定回路406に出力する。なお、本実施例では、地肌レベルバッファ403の１区間に対応するライン画像信号の部分を注目領域としており、特徴量抽出の範囲（以下では対象範囲と呼ぶ）も注目領域内として、対象範囲における画像信号の最大値、最小値および平均値を各信号成分毎に求め、特徴量として出力する。
図６は、特徴量抽出回路405（１信号成分分）のより詳細な回路構成を例示する。図６を参照すると、入力された画像信号601は、比較器602,603、セレクタ604,605および加算器606に入力される。
比較器602,603及びセレクタ604,605の他方の入力端子には、セレクタ607,608を介してＦ／Ｆ609,610の出力が入力されており、比較器602は、両入力の大きさを比較して、小さい方の信号をセレクタ604が選択するような制御信号611をセレクタ604に出力し、他方、比較器603は、両入力の大きさを比較して、大きい方の信号をセレクタ605が選択するような制御信号612をセレクタ605に出力する。
セレクタ604,605の出力信号は、上記Ｆ／Ｆ609,610に入力されおり、Ｆ／Ｆ609,610は、画像信号の同期信号613に応じて上述の選択結果を保持する。
また、上記セレクタ607,608の他方の入力端子には、それぞれ画像信号がとり得る最大値と最小値が入力されており、セレクタ607,608の動作は、上記区間クロック508により制御されている。即ち、区間の切替りにおいては最大値と最小値をそれぞれ選択する。
また、Ｆ／Ｆ609,610の出力信号は、Ｆ／Ｆ614,615にも入力されており、Ｆ／Ｆ614,615は、区間クロック508によりこれを保持する。これにより対象範囲における画像信号の最大値、最小値がＦ／Ｆ614,615より出力されることになる。
一方、加算器606の出力信号は、Ｆ／Ｆ616に入力されており、Ｆ／Ｆ616は、画像信号の同期信号613に応じてこれを保持する。
また、Ｆ／Ｆ616の出力信号は、セレクタ617を介して、加算器606の他方の入力端子に接続される。ここでセレクタ617は、区間クロック508により制御されており、他方に入力端子には“２０”が入力され、区間の切替りにおいては“０”を選択する。
また、Ｆ／Ｆ616の出力信号はＦ／Ｆ618にも入力されており、Ｆ／Ｆ618は、区間クロック508によりこれを保持する。これにより対象範囲における画像信号の総和がＦ／Ｆ618より出力されることになる。なお、上述したように本実施形態では１区間１６画素としているので、対象範囲における画像信号の総和を示すＦ／Ｆ618の出力を１／１６、即ち、下位４ビットを除いた結果が、平均値となる。
【００２４】
再び図４に戻ると、地肌レベル決定回路４０６は、上記予測地肌レベルおよび抽出した特徴量に基づいて、注目領域の地肌レベルを決定する回路で、決定された地肌レベルは、地肌レベルバッファ回路４０３に出力され記憶されると共に、注目領域の地肌レベルの検出結果として地肌補正回路４０２へ出力される。
ここで、地肌補正に用いる地肌レベルを求める地肌レベル決定回路４０６を以下に示す実施例に基づいて詳細に説明する。
図３を参照して説明したように、地肌レベルの決定手順では、注目領域の画像が地肌部分であるか否かを入力画像の特徴量に基づいて判定するが、原稿の地肌部分は、通常、原稿に使用されている紙そのものであり、一般に、明暗の変化が少なく、無彩色で、一定以上に明るいという特徴を有しており、ここで行う原稿の地肌判定は、この三つの特徴に基づいて行う。
上記の特徴量抽出回路４０５（図６）からは、対象領域の画像信号各成分の最大値、最小値および平均値が入力されるので、本例の地肌レベル決定回路４０６は、上記した三つの特徴を次のような方法により評価する。即ち、明暗の変化が少ないかどうかは、各成分毎に最大値と最小値の差を求めて、それぞれ所定の閾値と比較し、評価し、又無彩色かどうかは、各成分の平均値の最大値と最小値の差を求めて所定の閾値と比較し、評価し、更に一定以上に明るいかどうかは、各成分の平均値をそれぞれ所定の閾値と比較し、評価する。
【００２５】
これらの評価結果が地肌部分の特徴を全て備えていれば、対象領域を地肌部分と判定し、地肌レベル決定回路406は、各成分毎に下記式(2)に例示するような予測地肌レベルと対象領域における画像信号の平均値との加重平均により、地肌レベルを決定する。この例では、予測地肌レベルへの重み付けを大きくし、対象領域の重み付けを比較的小さくすることにより、予測地肌レベルを対象領域の画像により補正するという方法をとっている。

他方、この例では、特徴の評価結果の中に地肌部分の特徴を備えていない特徴があれば、対象領域を地肌部分と判定せず、この場合は、予測地肌レベルを補正する根拠がないので、予測地肌レベルをそのまま地肌レベルとして決定する。
【００２６】
図７及び図８は、地肌レベル決定回路４０６のより詳細な回路構成を例示し、図７は、対象領域の地肌判定部、図８は、地肌判定結果を受けて地肌レベルを決定する決定部を示す。
図７を参照すると、ＭａｘＲ，ＭａｘＧ，ＭａｘＢ、ＭｉｎＲ，ＭｉｎＧ，ＭｉｎＢ、ＡｖｅＲ，ＡｖｅＧ，ＡｖｅＢは、特徴量抽出回路４０５から出力された対象範囲の画像信号の最大値、最小値および平均値の各色成分信号であり、ＭａｘＲとＭｉｎＲ信号、ＭａｘＧとＭｉｎＧ信号、ＭａｘＢとＭｉｎＢ信号は、それぞれ差分回路７０１，７０２，７０３に入力される。差分回路７０１，７０２，７０３は、入力信号の差分を出力する回路であり、その結果は比較器７０４，７０５，７０６に入力される。
比較器７０４，７０５，７０６には、それぞれ所定の閾値信号ＤｉｆＴＨｒ，ＤｉｆＴＨｇ，ＤｉｆＴＨｂが入力されており、比較器７０４，７０５，７０６は、それぞれ差分値が閾値ＤｉｆＴＨｒ，ＤｉｆＴＨｇ，ＤｉｆＴＨｂより小さければ真をＡＮＤゲート７０７に出力する。これにより、ＡＮＤゲート７０７からは明暗の変化が少ない時、真が出力される。
また、ＡｖｅＲ，ＡｖｅＧ，ＡｖｅＢ信号は、所定の閾値信号ＤａｒｋＴＨｒ，ＤａｒｋＴＨｇ，ＤａｒｋＴＨｂ共に、それぞれ比較器７０８，７０９，７１０に入力されており、比較器７０８，７０９，７１０はそれぞれ平均値が閾値より大きければ真をＡＮＤゲート７１１に出力する。これにより、ＡＮＤゲート７１１からは一定以上に明るい時、真が出力される。
また、ＡｖｅＲ，ＡｖｅＧ，ＡｖｅＢ信号は、それぞれ最大値選択回路７１２および最小値選択回路７１３にも入力されており、最大値選択回路７１２は各色成分の平均値の最大値を、最小値選択回路７１３は各色成分の平均値の最小値をそれぞれ選択して、差分回路７１４に出力する。差分回路７１４は入力信号の差分を出力する回路であり、その結果は比較器７１５に入力される。
比較器７１５には、それぞれ閾値信号ＢａｌＴＨも入力されており、比較器７１５は差分値が閾値より小さければ真を出力する。これにより、比較器７１５からは無彩色の時、真が出力される。
また、ＡＮＤゲート７０７，７１１および比較器７１５の出力は、ＡＮＤゲート７１６に入力されている。これにより、ＡＮＤゲート７１６からは対象領域が地肌部分の特徴を全て備えている時、真が出力される。
【００２７】
地肌判定の結果を受けて地肌レベルを決定する回路部を示す図８を参照すると、ExpR,ExpG,ExpBは、地肌レベル予測回路404から出力された予測地肌レベル信号であり、対象領域の画像信号の平均値AveR,AveG,AveBと共に、それぞれ差分回路721,722,723に入力される。差分回路721,722,723は、入力信号の差分を出力する回路であり、その結果は比較器724,725,726に入力される。
比較器724,725,726には、それぞれ所定の閾値信号DownTHr,DownTHg,DownTHbも入力されており、比較器724,725,726はそれぞれ差分値が閾値より小さければ真をＡＮＤゲート727に出力する。なお、ここでは平均値の方が予測地肌レベルより暗い場合に、差分回路が正の値を出力している。これにより、ＡＮＤゲート727からは各色成分の平均値が予測地肌レベルより僅かに暗いか、明るい場合に、真が出力される。
また、上記図７に関して説明したＡＮＤゲート716及びＡＮＤゲート727の出力信号は、ＡＮＤゲート728に入力されている。
他方、729,730,731は、上記式(2)で示した予測地肌レベルの補正演算を行う加重平均回路であり、予測地肌レベル信号ExpR,ExpG,ExpB及び対象領域の画像信号の平均値AveR,AveG,AveBが入力され、演算結果をセレクタ732,733,734に出力する。セレクタ732,733,734の他方の入力端子には、それぞれExpR,ExpG,ExpB信号が入力されており、セレクタ732,733,734の動作は、上述のＡＮＤゲート728出力信号で制御されている。
ここで、ＡＮＤゲート728出力信号は、対象領域が地肌部分の特徴を全て備えており、各色成分の平均値が予測地肌レベルより僅かに暗いか、明るい場合に真となり、この時セレクタ732,733,734は、加重平均回路の演算結果を選択出力する。なお、後者の条件、即ち、ＡＮＤゲート727の条件は、銀塩写真のように連続的に濃淡が変化する画像のハイライト部分を地肌と判定して地肌レベルが追従してしまうのを、防ぐのに有効である。即ち、ハイライト部分と真の地肌部分との濃淡差を、平均値が予測地肌レベルより僅かに暗いかどうかで識別し、これにより地肌レベルが追従しないようにすることができる。
以上のように、本例によれば、セレクタ732,733,734から地肌レベル決定回路406によって決定された地肌レベルが出力される。
【００２８】
再び図４に戻り、地肌レベル決定回路406によって決定された地肌レベルを受けて行う地肌補正について説明する。
地肌補正回路402には、地肌検出回路401が検出した地肌レベル（地肌レベル決定回路406が決定した地肌レベル）が入力されるとともに、遅延バッファ407を介して画像信号も入力されており、地肌補正回路402により入力画像信号に対し地肌レベルに応じた補正を行う。
ここで、本発明に係わる地肌補正方法の概略を説明する。
図９は、地肌補正処理における画像データの入出力特性を示す線図で、本実施形態の地肌補正と補正を行わない場合の特性を対比して示す。同図を参照すると、地肌レベルに応じた補正を行わない場合、傾き１の直線にするものとする。また、地肌部分の補正目標となる所定出力レベルを基準地肌レベルとして示す。
本実施形態の補正方法は、地肌検出回路401が検出した地肌レベルによって、対象領域（図３に示した主走査方向に分割した区間）の画像信号に対して適用する変換特性を変更する。検出地肌レベルによって変更する変換特性の例を検出地肌レベルが目標とする基準地肌レベルより明るい場合と、暗い場合をあげて以下に説明する。
【００２９】
検出地肌レベルが目標の基準地肌レベルより明るい （入力軸上Ａで示す）場合、入力画像データが検出地肌レベルより明るい領域では、検出地肌レベルと同値の入力画像データが基準地肌レベルに変換され、傾きが１である直線（図９中にＡ０−Ａ１特性線として示す）により、入力画像データを変換する特性を用い、入力画像データが検出地肌レベルより暗い領域では、検出地肌レベルと同値の入力画像データが基準地肌レベルに変換され、１より小さい所定の傾きの直線（図９中にＡ０−Ａ２特性線として示す）により、入力画像データを変換する特性を用いる。但し、この変換結果が、補正を行わない場合よりも大きくなる領域では、補正無しの変換とする（図９中にＡ２−原点 特性線として示す）特性を用いる。
他方、検出地肌レベルが基準地肌レベルより暗い （入力軸上Ｂで示す）場合、検出地肌レベルから所定の閾値ＮＲを引いた値（入力軸上Ｂ'で示す）より入力画像データが明るい領域では、検出地肌レベルと同値の入力画像データが基準地肌レベルに変換され、傾きが１である直線（図９中にＢ１−Ｂ０−Ｂ２特性線として示す）により、入力画像データを変換する特性を用い、入力画像データが上述の値Ｂ'より暗い領域では、Ｂ２を通り、１より大きい所定の傾きの直線（図９中にＢ２−Ｂ３特性線として示す）により、入力画像データを変換する特性を用いる。但し、この変換結果が、補正を行わない場合よりも小さい領域では、補正無しの変換とする（図９中にＢ３−原点 特性線として示す）特性を用いる。
なお、上記の変換特性例においてはいずれも、検出地肌レベルの明暗に拘わらず、入力画像データが検出地肌レベルより大きい場合、傾き１の直線で変換を行うようにした。これは、例えば、特性線（Ａ０−Ａ２）を延長した特性線（Ａ０−Ａ１'）或いは特性線（Ｂ２−Ｂ３）と同じ傾きを有する特性線（Ｂ０−Ｂ１'）にしても良いが、検出地肌レベルが、基準地肌レベルを跨いで変化した際に、画像データの変換結果が不連続になってしまい、好ましくないからである。このため、入力画像データが検出地肌レベルより大きい場合、傾き１の直線で変換を行うようにしてこれを防いでいる。
また、以上においては、閾値ＮＲを設定して、検出地肌レベル近傍の入力画像データの変換傾きも１としている。これに代え、Ｂ０を通り特性線（Ｂ２−Ｂ３）のような傾きを有する特性線（Ｂ１'−Ｂ２'）で変換しても良い。但し、一般に、地肌部分の画像データは、必ずしも均一でなくある程度のばらつきを有しており、特性線（Ｂ１'−Ｂ２'）のような傾きを有する特性線で変換すると、このばらつきも拡大され、地汚れしたように見えてしまう場合がある。このため、閾値ＮＲを設定して、検出地肌レベル近傍における入力画像データの変換特性線の傾きを１にすることで、このようなばらつきの拡大を防いでいる。
【００３０】
次に、上記した地肌補正方法（図９に関する説明、参照）を実現する地肌補正回路402を図１０に示す回路構成例により詳細に説明する。
図１０を参照すると、地肌検出回路401から入力された検出地肌レベル信号Detectは、上記基準地肌レベル（図９、参照）に相当する基準地肌レベル信号BaseWと共に、比較器901に入力され、これにより、検出地肌レベルが基準地肌レベルより暗いか否かが判定される。
セレクタ902、減算器903及び比較器904は、図９に示した特性線（Ｂ１−Ｂ０−Ｂ２）或いは特性線（Ａ０−Ａ１）による変換を行うか否かの判定を行う回路を構成する。
このために、セレクタ902には、上記閾値ＮＲ（図９、参照）に相当する閾値信号ＮＲまたは固定値“０”が入力され、比較器901の出力信号による選択が行われる。つまり、セレクタ902は、基準地肌レベルより暗い時に閾値信号ＮＲを、その他の（明るい）時に“０”を選択出力する。また、減算器903は、検出地肌レベル信号Detectからセレクタ902の出力信号を減算する。これにより、図９に示したＢ２或いはＡ０の値が得られるので、比較器904により、遅延バッファ407（図４参照）から入力される画像信号ImgINと比較して、上記判定を実現する。
また、減算器905および加算器906は、上記比較器904の判定出力が真の時の補正処理を行う回路を構成する。即ち、減算器905により基準地肌レベル信号BaseWと検出地肌レベル信号Detectの差を算出して補正量を求めると共に、この減算結果を画像信号ImgINに加算することで、補正後の画像信号ImgOUTAを求める。
【００３１】
次に、図９に示した特性線（Ｂ２−Ｂ３）あるいは特性線（Ａ０−Ａ２）による変換を行う場合を説明する。なお、本実施例では、Ｂ２あるいはＡ０を基点に変換を実現している。
セレクタ907は、特性線（Ｂ２−Ｂ３）あるいは特性線（Ａ０−Ａ２）部分の傾きを選択しており、本実施例ではその傾きをそれぞれ“７／４”、“３／４”として、これを上述の比較器901の出力信号（検出地肌レベルが基準地肌レベルより暗いか否かの信号）で選択する。また、減算器908によりＢ２あるいはＡ０と画像信号ImgINと差を求め、これにセレクタ907が選択した傾きを乗算器909により乗算する。このようにして、Ｂ２あるいはＡ０からの補正後の差を求めることができる。
他方、Ｂ２あるいはＡ０の補正後の値は、基準地肌レベル信号BaseWからセレクタ902の出力信号を減算器910により減算することで求めることができる。従って、減算器910の出力から乗算器909の出力を減算器911により減算し、特性線（Ｂ２−Ｂ３）或いは特性線（Ａ０−Ａ２）による補正を行った画像信号ImgOUTBを得る。
但し、画像信号ImgOUTBは、特性線（Ｂ２−Ｂ３）あるいは特性線（Ａ０−Ａ２）の延長線上の変換も含んでいるため、この結果を比較器912により、補正を行わない場合と比較し、これにより画像信号ImgOUTBまたは画像信号ImgIN(補正を行わない場合)をセレクタ913により選択し、セレクタ913は画像信号ImgOutB'を出力する。なお、セレクタ914,915は、比較器901の出力信号（検出地肌レベルが基準地肌レベルより暗いか否かの信号）に応じて、それぞれ画像信号ImgOUTBまたは画像信号ImgINを選択して比較器912に出力しており、これにより、検出地肌レベルが基準地肌レベルより暗いか否かで、画像信号ImgOUTBと画像信号ImgINの大小からどちらを選ぶかを逆転している。
また、以上にようにして得られた画像信号ImgOUTAあるいはImgOUTB'は、比較器904の出力信号（特性線(Ｂ１−Ｂ０−Ｂ２)或いは特性線(Ａ０−Ａ１)による変換を行うか否かの判定結果を示す信号）で制御されるセレクタ916により選択され、地肌レベル信号Detctに応じて補正された画像信号ImgOUTが得られる。
【００３２】
【発明の効果】
本発明によると、画像データを主走査方向に複数画素からなる区間に分割し区間ごとに地肌レベルを検出するようにし、検出にあたり既に決定された周辺区間（注目区間に対する前の主走査ラインにおける対応区間及び前の主走査ライン上の該注目区間に対応する区間の前後の区間）の地肌レベルからの予測データを注目区間の現在の主走査ラインにおける画像データから抽出した特徴量（注目区間内における画像信号各色成分の最大値と最小値又は画像信号各色成分の平均値）に基づいて補正の要否を判定し、検出地肌レベルを決定することを可能にしたことにより、プレスキャンを必要とせず、原稿、読み取り装置条件等に起因する様々な地肌濃度変動（特に２次元的に連続する変動）に追従して正しい地肌レベルが検出でき、地肌補正の適正化が可能となる。
また、周辺区間の地肌レベルの加重平均を地肌レベルの予測データとして用いることにより、画像データに基づいて決定された地肌レベルが主走査方向に伝達されるので、非地肌部分の区間における地肌レベルの予測精度を向上することが可能になる。
また、特徴量に基づいて補正の要否を判定し、補正が必要であると判定されたときに、予測された地肌レベルと注目区間の現在の平均地肌レベルの平均値との加重平均値を算出し、これを検出地肌レベルとして決定する（即ち、予測地肌レベルを補正する）ようにしたので、検出地肌レベルが急変することを防止でき、地肌補正処理による処理ムラを低減することが可能になる。
【００３３】
さらに、注目区間の画像データから地肌を判定するために抽出した特徴量（注目区間内における画像信号各色成分の最大値と最小値又は画像信号各色成分の平均値）をもとに、明暗の変化、無彩色、一定以上の明るさに対する閾値処理により注目区間の画像を地肌とみなすか否かを判定し、地肌とみなせないと判定されたときに、予測された地肌レベル（周辺区間から予測）を検出地肌レベルとして決定するようにしたので、対象区間の地肌レベルとして決定される検出地肌データの精度を維持することが可能になる。つまり、明暗の変化をチェックすることにより、文字部や網点部を地肌と判定しないので、地肌部分の特徴を端的に表す特徴量を抽出できるので、検出地肌レベルの不要な変動を防止することができ、また、一定以上の明るさであることをチェックすることにより、検出地肌レベルが暗くなり過ぎないので、画像の濃淡が連続的に変化する原稿も処理することができ、また、無彩色であることをチェックすることにより、地肌の判定精度を高めることが可能になる。
【図面の簡単な説明】
【図１】 本発明の実施形態に係わるデジタル複写機の機構部の概略構成を示す。
【図２】 本発明の実施形態に係わるデジタル複写機の電装部の概略構成を示す。
【図３】 本発明に係わる地肌検出方法の原理説明図を示す。
【図４】 本発明の実施形態に係わる地肌除去回路の構成例を概略的に示す。
【図５】 地肌除去回路（図４）における地肌レベル予測回路のより詳細な回路構成を例示する。
【図６】 地肌除去回路（図４）における特徴量抽出回路のより詳細な回路構成を例示する。
【図７】 地肌除去回路（図４）における地肌レベル決定回路のより詳細な回路構成を例示する。
【図８】 地肌除去回路（図４）における地肌レベル決定回路のより詳細な回路構成を例示する。
【図９】 地肌補正処理における画像データの入出力特性を示す線図である。
【図１０】 地肌除去回路（図４）における地肌補正回路のより詳細な回路構成を例示する。
【符号の説明】
101…スキャナユニット、 102…プリンタユニット、
103…原稿、 111…画像処理ユニット、
201…操作表示ユニット、 202…デジタル複写機、
203…システム制御ユニット、 213…地肌除去回路、
401…地肌検出回路、 402…地肌補正回路、
403…地肌レベルバッファ回路、 404…地肌レベル予測回路、
405…特徴量抽出回路、 406…地肌レベル決定回路。[0001]
BACKGROUND OF THE INVENTION
The present invention provides an image processing apparatus (for example, an image scanner apparatus, a facsimile apparatus, a digital copying machine, or a function of these functions) that has a correction means for a document background component included in document image data input by a reading means such as a scanner. More specifically, the background density that varies depending on the document conditions such as paper and image type and the reading conditions, in particular, the background density (level) that continuously varies two-dimensionally. The present invention relates to the image processing apparatus including means for correctly detecting and enabling appropriate correction of image data.
[0002]
[Prior art]
Today, a scanner is generally used for inputting an image written on a manuscript, and examples thereof can be seen in an image scanner, a facsimile, a digital copying machine, or the like, which is an image processing apparatus that handles manuscript images.
In these image processing apparatuses, various image processes such as various corrections / conversions are performed by digital operations in order to make output image data necessary for use on original image data captured by a scanner. Includes a process called “skin removal”.
“Background removal” is generally desired that the background portion of a document is white when outputting an image from a device such as a copying machine based on a document image captured by a scanner. If the background is output at a density that is read, it is considered that the document is soiled, and the evaluation of the image quality is lowered. This is a correction process performed to prevent this.
By the way, there are various types of manuscripts handled by the image processing apparatus using the scanner as described above, and the density of the paper used for the manuscript, that is, the background density, is in a wide range. The method conventionally used to remove the background corresponding to these various originals uniquely detects the background density of the original and outputs “white” for the portion below the detected background density of the original image data. As it depends on how to erase the background.
[0003]
The following “Patent Document 1” to “Patent Document 4” can be cited as conventional examples of an image processing apparatus that performs background removal.
In “Patent Document 1”, a document in which a plurality of independent background densities are mixed, such as a newspaper or magazine clipped on a white paper, is edited, and a density histogram of image data created by prescanning is used. A plurality of background densities corresponding to each area are detected, and at the time of the main scan, it is determined by determining from the image data which of the detected background densities corresponds to the image area being processed. An image processing apparatus that performs background removal processing based on background density is shown.
In “Patent Document 2”, the background density is detected for each line without performing pre-scanning, and the background removal processing is performed. For this purpose, the background density is maintained, and the image is brighter than the retained background density. The difference from the background density of the data is aggregated for each line, and the background density of the next line is updated based on the accumulated difference and the retained background density, and is also updated in the direction of darkening the background density at a certain rate. However, an image processing apparatus that performs background removal processing is shown.
In “Patent Document 3”, a pattern highlight region of an image is detected, a low density pixel that is not included in the region is detected as a background pixel, and the background density is updated while performing background removal processing, thereby performing a pattern removal process. There is shown an image processing apparatus that prevents white spots in a highlight portion of an area and prevents deterioration in image quality due to background removal.
According to “Patent Document 4”, by selecting whether or not the background is a halftone dot, a line drawing, a photograph, or a fluorescent color region by the image region separation unit, an update value of the background removal threshold value is selected for each pixel. An image processing apparatus that performs a background removal process suitable for this area is shown.
[0004]
[Patent Document 1]
Japanese Patent No. 3134292
[Patent Document 2]
JP 2002-94795 A
[Patent Document 3]
Japanese Patent Laid-Open No. 7-74952
[Patent Document 4]
JP 7-264409 A
[0005]
[Problems to be solved by the invention]
By the way, the background density variation includes a two-dimensional continuous variation in addition to the above-described variation caused by the type of document paper. This is because the background density of the original document itself is not necessarily uniform, the floating of the original document caused by saddle stitching of the original document or the saddle stitching of a book original, the influence of flare due to the black solid image existing around, the reading mechanism, etc. Such continuous fluctuations in the background density could not be eliminated, and therefore it was necessary to remove the background more, and the highlight reproducibility was insufficient. .
In recent years, there has been an increasing demand for higher image quality, particularly highlight reproducibility of image portions, and there is a strong demand for performing background removal adaptively when the background density fluctuates two-dimensionally.
However, no solution has been proposed so far, including the above-mentioned “Patent Document 1” to “Patent Document 4”, as a solution that can sufficiently meet this requirement.
Further, the background correction processing proposed by “Patent Document 1” to “Patent Document 4” has the following problems. That is, in “Patent Document 1”, an extra operation of pre-scanning is required, so that a document cannot be processed at high speed. In “Patent Document 2”, since the background density is updated in units of lines, only one-dimensional fluctuation can be dealt with, and when the background density is updated, the background density becomes darker and is updated at a constant rate. bad. In “Patent Document 3”, since a printing halftone dot is targeted, a highlight area of a continuous tone image cannot be detected, and for example, a highlight portion such as a silver halide photograph is recognized as a background and is removed. . In “Patent Document 4”, image area separation is a technique originally developed to apply filter processing such as edge enhancement and smoothing to a necessary part, both of which are a continuous tone region and a background portion where smoothing is desirable. Therefore, as in Patent Document 3, a highlight portion such as a silver salt photograph is recognized as the background and is removed.
The present invention has been made in view of the above-described problems of the prior art in detecting the background density necessary for the background correction processing for a document image captured by an input means such as a scanner. The object of the present invention is to detect the background level. It does not require pre-scanning, and it responds to fluctuations in the background density caused by various factors such as document conditions and image input (reading) device conditions, especially two-dimensional continuous background density fluctuations. An object of the present invention is to detect the correct background level without erroneously recognizing the key highlight area as the background, and to enable proper correction of the input image signal.
[0006]
[Means for Solving the Problems]
According to the first aspect of the present invention, the background level detecting means for detecting the background level of the original document from the raster-format image data read by scanning the main and sub-scans, and the background level detected by the background level detecting means are read. In the image processing apparatus having means for correcting the image data to a target reference background level, the background level detecting means divides the image data into sections made up of a plurality of pixels in the main scanning direction, and the background level for each section. A background level storage means for storing the background level of each section on the previous main scanning line that has already been determined, and the background level of the section of interest on the current main scanning line. A section corresponding to the section of interest on the previous main scanning line stored in the storage means and a section before and after the section corresponding to the section of interest on the previous main scanning line. A background level predicting means that predicts by a weighted average of the skin levels in the surrounding sections, and a feature amount that extracts the maximum value and the minimum value of each color component of the image signal in the attention section from the image data of the attention section as a feature amount A difference value between the maximum value and the minimum value of each color component of the image signal extracted by the extraction means and the feature amount extraction means, and when the obtained difference value of each color component is smaller than a predetermined threshold, When the determination result cannot be regarded as the background, the background level predicted by the background level prediction means is determined as the background level of the attention section, and when the determination result can be regarded as the background, A ground in which a weighted average of the average background level of the target section on the scan line and the background level predicted by the background level prediction unit is determined as the background level of the target section. It is characterized in that it comprises a level decision means.
According to the second aspect of the present invention, the background level detecting means for detecting the background level of the original document from the raster-format image data read by performing main / sub-scanning on the original, and the background level detected by the background level detecting means. In the image processing apparatus having means for correcting the image data to a target reference background level, the background level detecting means divides the image data into sections made up of a plurality of pixels in the main scanning direction, and the background level for each section. A background level storage means for storing the background level of each section on the previous main scanning line that has already been determined, and the background level of the section of interest on the current main scanning line. A section corresponding to the section of interest on the previous main scanning line stored in the storage means and a section before and after the section corresponding to the section of interest on the previous main scanning line. A background level predicting unit that predicts by a weighted average of skin levels in various surrounding sections, and a feature amount extracting unit that extracts, as a feature amount, an average value of each color component of the image signal in the target section from the image data of the target section. , Each color component of the image signal extracted by the feature amount extraction means each Average value while When a difference value between the maximum value and the minimum value is obtained and the obtained difference value is smaller than a predetermined threshold value, a determination is made that the target section is regarded as the background, and when the determination result cannot be regarded as the background, the background level prediction is performed. The background level predicted by the means is determined as the background level of the attention section, and when the determination result can be regarded as the background, the average background level of the attention section on the current main scanning line and the background level prediction means The present invention is characterized by comprising a background level determining means for determining a weighted average with the background level as the background level of the section of interest.
According to the third aspect of the present invention, the background level detecting means for detecting the background level of the original document from the raster format image data read by scanning the main and sub-scans, and the background level detected by the background level detecting means are used for reading. In the image processing apparatus having means for correcting the image data to a target reference background level, the background level detecting means divides the image data into sections made up of a plurality of pixels in the main scanning direction, and the background level for each section. A background level storage means for storing the background level of each section on the previous main scanning line that has already been determined, and the background level of the section of interest on the current main scanning line. A section corresponding to the section of interest on the previous main scanning line stored in the storage means and a section before and after the section corresponding to the section of interest on the previous main scanning line. A background level predicting unit that predicts by a weighted average of skin levels in various surrounding sections, and a feature amount extracting unit that extracts, as a feature amount, an average value of each color component of the image signal in the target section from the image data of the target section. When the average value of each color component of the image signal extracted by the feature amount extraction unit is larger than a predetermined threshold, the background level is determined when the target section is determined as the background, and the determination result cannot be considered as the background. When the background level predicted by the prediction means is determined as the background level of the attention section, and the determination result can be regarded as the background, the average background level of the attention section on the current main scanning line and the background level prediction means are predicted. The present invention is characterized by comprising a background level determining means for determining a weighted average with the background level as the background level of the section of interest.
[0009]
DETAILED DESCRIPTION OF THE INVENTION
An image processing apparatus according to the present invention will be described with reference to the following embodiments shown in the accompanying drawings. In the following embodiment, an example in which the image processing apparatus according to the present invention is implemented in one type of digital copying machine will be described.
FIG. 1 is a diagram showing an outline of a mechanism part of a digital copying machine according to this embodiment.
With reference to FIG. 1, the structure of the mechanism of the digital copying machine will be described below together with related operations.
The mechanism section mainly includes a scanner unit 101 that reads a document and a printer unit 102 that records an image on a recording sheet.
The scanner unit 101 has the following elements necessary for photoelectrically reading an image of the original 103 by performing main / sub scanning and inputting raster format data.
The original 103 is placed at a predetermined position on the platen 104 and illuminated by halogen lamps 105-1 and 105-2. Reflected light from the original 103 passes through the first mirror 106, the second mirror 107, the third mirror 108 and the lens 109 and is imaged on the CCD 110 which is a three-line color line image sensor, and is photoelectrically converted into an image signal by the CCD 110. (The original is scanned in the line direction of the CCD). Here, the halogen lamps 105-1 and 105-2 and the first mirror 106 are mounted on a first carriage (not shown), and the second mirror 107 and the third mirror 108 are mounted on a second carriage (not shown). The first and second carriages are moved from left to right in FIG. 1 at a speed ratio of 2: 1 by a carriage drive motor (not shown) (the document is sub-scanned). As a result, the entire surface of the document placed on the platen 104 is scanned while keeping the optical path length between the document and the lens constant.
The image signal photoelectrically converted by the CCD 110 is subjected to various processes (detailed later) by the image processing unit 111 and the like, and then input as a write signal to an LD (laser diode) (not shown) of the printer unit 102. Here, it is converted into laser light.
[0010]
Further, the printer unit 102 includes the following elements necessary for performing image forming processing by a laser beam writing type electrophotographic process in correspondence with color.
In the printer unit 102, laser light emitted from an LD (not shown) is reflected by a polygon mirror 112, passes through an fθ lens 113 and a fourth mirror 114, and rotates in a counterclockwise direction. The image is irradiated on the surface of the body drum 115. Here, the polygon mirror 112 is fixed to the rotating shaft of the polygon motor 116, and the polygon motor 116 rotates at a constant speed to rotationally drive the polygon mirror 112. By the rotation of the polygon mirror 112, the laser light described above is scanned in a direction perpendicular to the rotational movement direction of the photosensitive drum 115, that is, a direction along the drum axis.
On the other hand, the surface of the photosensitive drum 115 is charged to a uniform positive potential in advance by a charging charger 117 connected to a high voltage generator (not shown). Further, when the photosensitive drum 115 is irradiated with laser light, the surface charge flows to the device ground of the drum body due to a photoconductive phenomenon and disappears. Here, in accordance with the writing signal, the portion with a low original density turns on the LD weakly, and the portion with a high original density turns on the LD strongly. By performing such writing, an electrostatic latent image corresponding to the density of the original is formed on the surface of the photosensitive drum 115 by main scanning by the polygon mirror 112 and sub scanning by rotation of the photosensitive drum 115. The
[0011]
The developing unit 118 has developing units K, C, M, and Y that store black, cyan, magenta, and yellow positively charged toners, respectively, for the electrostatic latent image formed on the photosensitive drum 115. Thus, any one developing unit is selected. The selected developing unit is biased to a predetermined positive potential by a high voltage generator (not shown), develops the electrostatic latent image by adhesion of toner, and generates a toner image corresponding to the density of the original on the photosensitive drum 115. Form on the surface.
The transfer belt 119 is biased to a predetermined negative potential by a high voltage generator (not shown), and is rotated clockwise at the same speed as the photosensitive drum 115, so that the photosensitive drum 115 and the transfer belt 119 approach each other. The toner image is attracted by the action of the bias and transferred from the photosensitive drum 115 to the surface of the transfer belt 119.
The electrostatic latent image formation, toner image formation, and toner image transfer operation are repeated as many times as necessary depending on the developing sections K, C, M, and Y to be used. That is, in each of the full color, document color, and registered color modes, the process is performed four times, the developing unit is selected in the order of K, C, M, and Y, and the formed toner image is aligned. Overlaid on the surface of the transfer belt 119. In each of the black, cyan, magenta, and yellow modes, the process is performed only once, and the developing portions K, C, M, and Y are selected, respectively. In each of the red, green, and blue modes, the process is performed twice. The developing portions are selected in the order of M and Y, C and Y, and C and M, and the formed toner image is aligned. The image is superimposed on the surface of the transfer belt 119.
On the other hand, recording sheets 121-1 and 121-2 are respectively stored in the sheet feeding cassettes 120-1 and 120-2, and one of the sheet feeding cassettes is selected. The recording paper 121-1 of the selected paper feeding cassette, for example, the paper feeding cassette 120-1, is fed out by the paper feeding operation of the paper feeding roller 122-1 and reaches the registration rollers 123-1, 123-2. . The registration rollers 123-1 and 123-2 are initially stopped, start rotating at a predetermined timing according to the position of the toner image on the rotating transfer belt 119, and feed the recording paper.
The transfer charger 124 is connected to a negative voltage high voltage generator (not shown), and the toner image on the transfer belt 119 is retransferred onto the fed recording paper by the action of the transfer charger 124. When the toner image is retransferred to the recording paper, the bias of the transfer belt 119 is released to promote the retransfer.
The recording paper on which the toner image has been retransferred is sent to the thermal fixing units 125-1 and 125-2, where the toner image is fixed to the recording paper and then discharged outside the apparatus.
The toner remaining on the surface of the photosensitive drum 115 after the transfer is removed by the cleaning unit 126, and the photosensitive drum 115 is prepared for the next operation. Further, the toner remaining on the surface of the transfer belt 119 after retransfer is removed by the cleaning unit 127, and the transfer belt 119 is also prepared for the next operation.
[0012]
FIG. 2 is a diagram showing an outline of the electrical component of the digital copying machine according to this embodiment.
With reference to FIG. 2, the configuration of the electrical component of the digital copying machine will be described below together with related operations.
The electrical unit mainly includes the scanner unit 101, the image processing unit 111, and the printer unit 102 described above, and an operation display unit 201 that performs detection and display of inputs such as processing mode selection, and controls the above units. Each unit includes a system control unit 203 that communicates with a circuit and controls the overall operation of the digital copying machine 202.
The digital copying machine 202 is connected to a LAN (local area network) connection device 204, and other devices such as a workstation 206 and a personal computer 207 connected to the LAN 205 are connected to the LAN connection device 204 and the like. Input / output of image signals.
[0013]
The scanner unit 101 includes elements necessary for processing of a read image signal, lamp lighting control in the scanner, and carriage movement control in the following configuration.
The CCD 110 (see FIG. 1) color-separates incident light into red, green, and blue (R, G, B), and then performs photoelectric conversion to output three types of analog image signals to the A / D conversion circuit 208. . The A / D conversion circuit 208 converts the input image signal into a digital signal and outputs it to the shading correction circuit 209 while correcting the dark current of the CCD 110 and the like.
The shading correction circuit 209 is a circuit that corrects the uneven illumination of the above-mentioned halogen lamps 105-1 and 105-2, the sensitivity unevenness of the light receiving element in the CCD 110, and the like when the white reference plate 126 shown in FIG. The output signal of the A / D conversion circuit 208 is stored as shading correction data, and the gain adjustment of the image signal for each color and each pixel is performed based on this. As a result, the unevenness generated in the output image signal is corrected, that is, the output image signal of the shading correction circuit 209 when the white reference plate 126 is read becomes a predetermined value. Note that the shading correction circuit 209 also performs processing such as correction for a reading position shift between red, green, and blue image signals due to the three-line structure of the CCD 110.
The scanner control circuit 210 communicates with the system control unit 203 and controls the entire scanner unit 101. For example, the operation control of the A / D conversion circuit 208 and the shading correction circuit 209, the halogen lamp 105- The lighting control of 1, 105-2 and the rotation control of the carriage drive motor 211 are performed. The document size sensor 212 is a sensor that detects the size of the document placed on the platen 104, and outputs the detection result to the scanner control circuit 210.
An image signal output from the shading correction circuit 209 is input to the image processing unit 111.
[0014]
The image processing unit 111 has the following components necessary for correcting and converting the read image signal received from the scanner unit 101 into a signal suitable for the signal used by the printer unit 102 as a write signal. Prepare.
The image processing unit 111 receives the image signal output from the shading correction circuit 209 as an input signal to the background removal circuit 213.
The background removal circuit 213 detects the background level of the document based on the input image signal, corrects the input image signal according to the background level, and outputs it to the γ conversion circuit 214. The detection of the background level of the document and the correction of the input image signal in accordance with the background level are matters relating to the problem to be solved by the present invention, and will be described in detail later.
The γ conversion circuit 214 is also connected to the LAN connection device 204, and converts the gradation of the image signal input from the background removal circuit 213 or the LAN connection device 204 into an image signal proportional to the density, etc. 215, and output to the area control circuit 216 and the delay circuit 217. Further, depending on the setting (for example, setting of an operation mode using the scanner distribution function), the γ conversion circuit 214 may perform gradation conversion on the image signal input from the background removal circuit 213 and output the image signal to the LAN connection device 204.
The image area separation circuit 215 is a circuit for determining whether or not the image portion being processed is a line drawing such as a character, a monochrome image or a color image based on the input image signal, and a determination signal indicating the result Is output to the filter circuit 218.
The delay circuit 217 delays the determination by the image area separation circuit 215, and delays the image signal in accordance with this delay and outputs the delayed image signal to the filter circuit 218.
On the other hand, the area control circuit 216 is a circuit that stores an input image signal, generates a switching signal for performing different image processing for each partial area of the image, and the like. Output to.
[0015]
The filter circuit 218 is a circuit that outputs the image signal from the delay circuit 217 by performing two-dimensional filter processing such as edge enhancement and smoothing. These processes are controlled by the determination signal of the image area separation circuit 215 and the switching signal of the area control circuit 216. For example, if the determination signal is a line image, edge enhancement is performed. If the determination signal is a non-line image, smoothing filter processing is performed. To do. Also, edge enhancement or smoothing filtering may be performed by the switching signal regardless of the determination signal.
The color correction circuit 219 converts the image signal from the filter circuit 218 into a toner recording amount corresponding to each of the K, C, M, and Y developing units selected in the developing unit 118 as described above. Perform processing. This processing is controlled by the determination signal of the image area separation circuit 215 and the switching signal of the area control circuit 216, for example, converting to a toner recording amount suitable for the full color mode or the black mode according to the switching signal, In the full color mode, conversion may be performed to set the UCR rate to 100% if the determination signal is a line drawing and a monochrome image, and to 70% if the determination signal is a non-line drawing or a color image. In addition, a painting process for outputting a constant image signal value or determination signal value according to the switching signal, a color erasing process for erasing the designated color, and a color conversion process for converting the designated color into another color are performed. There is also. From the color correction circuit 219, one type of image signal converted into the toner recording amount, the determination signal, and the switching signal are output and input to the scaling circuit 220.
The scaling circuit 220 is a circuit that performs scaling processing for enlarging / reducing the three types of signals from the color correction circuit 219 in the main scanning direction. The scaling process for enlarging / reducing the document in the sub-scanning direction is performed by controlling the rotation of the drive motor 211 that moves the carriage.
[0016]
The gradation processing circuit 221 performs γ correction processing for correcting the recording characteristics (image signal vs. toner recording amount characteristics) of the printer unit 102 that fluctuate depending on temperature and humidity, and one or several pixels in the main scanning direction and one in the sub scanning direction. Alternatively, gradation processing or the like for expressing gradation with emphasis on resolution or gradation is performed with several pixels as one unit. Here, the γ correction processing and the gradation processing are controlled by the determination signal and the switching signal. For example, if the determination signal is a line drawing, the gradation processing with an emphasis on resolution is performed. Perform important gradation processing. In addition, since the above-described recording characteristics change depending on the type of gradation processing, different γ correction processing is performed in conjunction with gradation processing. Further, predetermined gradation processing or γ correction processing may be performed by the switching signal regardless of the determination signal. Note that the gradation processing circuit 221 is also connected to the LAN connection device 204, and selectively outputs to the printer unit 102 the image signal subjected to the above processing or the image signal input from the LAN connection device 204. Further, the gradation processing circuit 221 may output the image signal subjected to the above processing to the LAN connection device 204 depending on the setting (for example, setting of an operation mode using a distribution function such as linked copy).
The image processing control circuit 222 communicates with the system control unit 203 and, in response to a request from the system control unit 203, the background removal circuit 213, the γ conversion circuit 214, the image area separation circuit 215, and the area control circuit 216 described above. The delay circuit 217, filter circuit 218, color correction circuit 219, scaling circuit 220 and gradation processing circuit 221 are set to control the entire image processing unit 111.
[0017]
The printer unit 102 includes elements necessary for performing laser beam writing on a photoconductor using an image signal received from the image processing unit 111 and controlling image forming processing by an electrophotographic process in the following configuration.
In the printer unit 102, the image signal output from the gradation processing circuit 221 is input to the LD control circuit 223. The LD control circuit 223 drives the LD 224 by performing pulse width modulation, power modulation, or the like according to the input image signal, thereby controlling the lighting intensity of the LD 224.
The printer control circuit 225 communicates with the system control unit 203 and controls the entire printer unit 102. For example, the LD control circuit 223 is controlled to forcibly turn off the LD 224, the rotation control of the polygon motor 116, the motor 226 that rotates the photosensitive drum 115, the motor 227 that rotates the transfer belt 119, and the like. Selection control of developing units K, C, M, and Y, output control for each load (for example, charging charger, developing unit, transfer belt, transfer charger, etc.) of the high-voltage power supply 228, thermal fixing units 125-1, 125-2 Perform temperature control. The printer unit 102 has a paper size sensor 229 for detecting the size of the stored recording paper for each of the paper feed cassettes 120-1 and 120-2, and the detection result is a printer control circuit. Is output to 225.
The operation display unit 201 displays various mode selections, setting states, and other scanned document images as elements for enabling the user to perform input operations interactively and to know the state of the device. A TPD (touch panel display) 230 in which a display unit and a detection unit for detecting the pressed position of the display unit are integrated, a keyboard 231 for inputting the number of copies, start of copying, and the like, and an operation display control circuit 232 are provided. . The operation display control circuit 232 prompts key input etc. by displaying the options of each mode on the TPD 230, detects the input from the TPD 230 and the keyboard 231 and displays the state set in the TPD 230 as well as system control. It communicates with the unit 203 and transmits the input result. The TPD 230 is also controlled when an image signal of a read document or the like is sent to the TPD 230 via the area control circuit 216 described above.
[0018]
Here, an outline of the background detection method according to the present invention will be described.
FIG. 3 is an explanatory diagram of this background detection method. Referring to the figure, 301 indicates the position of the document, and 302 indicates the range of the document image data. Here, the image data refers to data obtained by disassembling an area including a document portion into pixels. Here, basically, assuming that data read by the raster scan method in the main / sub-scanning direction is assumed, the horizontal direction in FIG. 3 is matched with the main scanning direction of the raster scan, and the vertical direction is matched with the sub-scanning direction.
The background detection method is characterized in that the main scanning direction is divided into a plurality of sections and handled, and detection processing is performed for each section. For this purpose, a background level buffer 303 that holds the background level for each section is prepared.
Here, since it is premised on a method of reading the entire image data by repeating the raster scan, that is, the main scan in the sub-scan direction, the background detection processing operation is performed on the image data input according to this reading method. Is called.
Accordingly, assuming that the image signal at the sub-scanning position 304 is a processing target now, at this time, the background level buffer 303 stores the line before the current main scanning line (that is, at the previous sub-scanning position). It is assumed that the background level determined by (main scanning line) is held.
[0019]
Since detection processing is executed for each section, attention is paid to a section 305 with main scanning at the sub-scanning position 304 as a detection section (hereinafter, this area is referred to as “attention area” or “attention section”). Usually, the background level in the image data hardly varies depending on the position, and even if it changes, it is gentle. For this reason, the background level of the attention area can be predicted by referring to the background level of the adjacent peripheral area. In the background level prediction processing block 306, the background level is stored in the background level buffer 303 as a peripheral section adjacent to the attention area. The background level of the attention area is predicted with reference to the background level of the section corresponding to the attention area and the right and left adjacent sections determined (that is, determined in the previous line).
The predicted value of the background level may be applied as it is as the background level of the region of interest when there is very little variation due to the position. The predicted value is corrected and used as an appropriate background level. In addition, since correction cannot be performed when the attention area is not the background, a method of applying the predicted value as the background level of the attention area is applied.
[0020]
As a processing procedure for executing this method, here, it is determined whether or not the attention area is a background part by referring to the image data of the attention area. Correct the background level.
Referring to FIG. 3, the feature amount extraction processing block 307 refers to the image data of the attention area 305 in order to determine whether or not it is a background portion in the above processing procedure. The feature quantity for determining is extracted. Also, in the background level determination processing block 308, it is determined whether or not it is a background portion based on the extracted feature amount. If it is the background, the predicted background level is corrected by the feature amount and the attention area 305 is corrected. If the background is not the background, the predicted background level is determined as the background level of the attention area 305 as it is. It should be noted that the feature level extraction and the background level determination process using the feature amount will be described in detail in an embodiment described later.
In addition, the determined background level is used as detection data for the attention area 305, and is stored in the background level buffer 303 so that it can be used as data of a peripheral section adjacent to the attention area in the background level determination processing of the next line. The above-mentioned premise is established. Note that since there is no previous line in the main scanning line (first line) at the first sub-scanning position, there is no background level to be held in the background level buffer 303, but for example, a standard background level is held as an initial value. By doing so, the background detection by this method is enabled from the first line in the sub-scanning direction.
As described above, according to this background detection method, pre-scanning is unnecessary, the background level can be detected as the raster scan progresses, and the division is performed following the two-dimensional fluctuation of the background level. It is possible to provide appropriate data for each section.
[0021]
Next, the configuration of a circuit as means for realizing the above-described background detection method (see FIG. 3) will be described.
In the present embodiment, as described above, the background removal circuit 213 (FIG. 2) of the image processing unit 111 detects the background level of the document based on the input image signal. A background level detection circuit is configured.
FIG. 4 schematically shows a configuration example of the background removal circuit in the present embodiment.
Referring to FIG. 4, a background removal circuit 213 mainly detects a background level of a document based on an input image signal as an element necessary for background correction, and an image corresponding to the detected background level. A background correction circuit 402 for correcting the signal is provided.
A background level buffer (line buffer) circuit 403 in the background detection circuit 401 stores a background level for each section when the main scanning direction of the image signal is divided into a plurality of sections (see FIG. 3). This corresponds to the background level buffer 303), and the stored background level is output to the background level prediction circuit 404 in accordance with the position of the image signal input to the background detection circuit 401 in the main scanning direction. As shown in FIG. 3, in this embodiment, the main scanning direction of the image signal is divided with 16 pixels as one section.
[0022]
The background level prediction circuit 404 is a peripheral section corresponding to an area (hereinafter, referred to as “attention area” or “attention area”) that is focused on by a feature amount extraction circuit 405 described later, that is, an area corresponding to the attention area and its section. Based on the background level of the previous line stored in the background level buffer circuit 403 as the background level of the left and right adjacent sections, the circuit predicts the background level of the attention area and outputs the predicted background level to the background level determination circuit 406. In the present embodiment, the background level is predicted by calculating the following weighted average from the background levels of the surrounding sections.

FIG. 5 illustrates a more detailed circuit configuration of the background level prediction circuit 404. Referring to FIG. 5, the background level signal 501 input from the background level buffer circuit 403 is input to an F / F (flip-flop) 502. The background level signal 503 output from the F / F 502 is input to the F / F 504. Further, the background level signal 505 output from the F / F 504 is input to the F / F 506, and the background level signal is output from the F / F 506. 507 is output. Here, each F / F 502, 504, 506 is input with a section clock 508 corresponding to the switching of the section described above, and the background level signals 503, 505, 507 are respectively the right adjacent section, the attention area corresponding section, and the left adjacent section described above. Corresponds to the background level of the section. These background level signals are input to the product-sum circuit 509, and the product-sum circuit 509 performs a product-sum operation as shown in the above equation (1) for each color component. As a result, a predicted background level signal 510 in which the background level of the region of interest is predicted can be generated.
[0023]
Returning to FIG. 4, the feature amount extraction circuit 405 is a circuit that extracts the feature amount of the input document image signal. The feature amount extraction circuit 405 determines whether the extracted feature amount is an image signal of the background portion of the document, The data is output to the background level determination circuit 406 as data for correcting the predicted background level. In this embodiment, the line image signal portion corresponding to one section of the background level buffer 403 is set as the attention area, and the feature amount extraction range (hereinafter referred to as the target area) is also set as the attention area. The maximum value, the minimum value, and the average value of the image signal are obtained for each signal component and output as a feature amount.
FIG. 6 illustrates a more detailed circuit configuration of the feature amount extraction circuit 405 (for one signal component). Referring to FIG. 6, the input image signal 601 is input to comparators 602 and 603, selectors 604 and 605 and adder 606.
The other input terminals of the comparators 602 and 603 and the selectors 604 and 605 receive the output of the F / F 609 and 610 via the selectors 607 and 608, and the comparator 602 compares the magnitudes of both inputs, The control signal 611 is output to the selector 604 so that the signal is selected by the selector 604, while the comparator 603 compares the magnitudes of both inputs and the control signal is selected by the selector 605. 612 is output to the selector 605.
The output signals of the selectors 604 and 605 are input to the F / F 609 and 610, and the F / F 609 and 610 hold the above selection result according to the synchronization signal 613 of the image signal.
The other input terminals of the selectors 607 and 608 are respectively input with a maximum value and a minimum value that can be taken by the image signal, and the operations of the selectors 607 and 608 are controlled by the interval clock 508. That is, the maximum value and the minimum value are selected for switching between sections.
The output signals of the F / Fs 609 and 610 are also input to the F / Fs 614 and 615, and the F / Fs 614 and 615 hold this by the section clock 508. As a result, the maximum and minimum values of the image signal in the target range are output from the F / Fs 614 and 615.
On the other hand, the output signal of the adder 606 is input to the F / F 616, and the F / F 616 holds this according to the synchronization signal 613 of the image signal.
Further, the output signal of the F / F 616 is connected to the other input terminal of the adder 606 via the selector 617. Here, the selector 617 is controlled by the section clock 508, and on the other hand, “20” is input to the input terminal, and “0” is selected in switching the section.
The output signal of the F / F 616 is also input to the F / F 618, and the F / F 618 holds this by the section clock 508. As a result, the sum of the image signals in the target range is output from the F / F 618. Note that, as described above, in this embodiment, since one section has 16 pixels, the output of F / F 618 indicating the sum of the image signals in the target range is 1/16, that is, the result of excluding the lower 4 bits is the average value. It becomes.
[0024]
Returning to FIG. 4 again, the background level determination circuit 406 is a circuit that determines the background level of the region of interest based on the predicted background level and the extracted feature amount. Is The background level is output to and stored in the background level buffer circuit 403 and is output to the background correction circuit 402 as the detection result of the background level of the region of interest.
Here, the background level determination circuit 406 for obtaining the background level used for background correction will be described in detail based on the following embodiments.
As described with reference to FIG. 3, in the background level determination procedure, it is determined based on the feature amount of the input image whether or not the image of the attention area is a background part. , The paper itself used for the manuscript, change Therefore, the background determination of the original document performed here is performed based on these three characteristics.
Since the maximum value, the minimum value, and the average value of each component of the image signal in the target region are input from the feature amount extraction circuit 405 (FIG. 6), the background level determination circuit 406 of the present example has the above three types. Features are evaluated by the following method. That is, light and dark change The difference between the maximum value and the minimum value for each component is obtained and compared with a predetermined threshold value for evaluation. Also, whether the color is achromatic is determined by the maximum and minimum values of the average value of each component. The difference between the values is compared with a predetermined threshold value and evaluated, and whether or not it is brighter than a certain level is evaluated by comparing the average value of each component with the predetermined threshold value.
[0025]
If these evaluation results have all the features of the background portion, the target region is determined to be the background portion, and the background level determination circuit 406 determines the predicted background level as exemplified in the following formula (2) for each component. The background level is determined by a weighted average with the average value of the image signal in the target region. In this example, the predicted background level is corrected by the image of the target area by increasing the weight on the predicted background level and relatively reducing the weight of the target area.

On the other hand, in this example, if there is a feature that does not include the feature of the background part in the evaluation result of the feature, the target region is not determined as the background part, and in this case, there is no basis for correcting the predicted background level. The predicted background level is determined as it is as the background level.
[0026]
7 and 8 exemplify a more detailed circuit configuration of the background level determination circuit 406, FIG. 7 shows a background determination unit for the target region, and FIG. 8 shows a determination unit that determines the background level in response to the background determination result. Indicates.
Referring to FIG. 7, MaxR, MaxG, MaxB, MinR, MinG, MinB, AveR, AveG, and AveB are the respective colors of the maximum value, minimum value, and average value of the image signal in the target range output from the feature amount extraction circuit 405. Component signals, MaxR and MinR signals, MaxG and MinG signals, and MaxB and MinB signals, are input to difference circuits 701, 702, and 703, respectively. The difference circuits 701, 702, and 703 are circuits that output the difference between the input signals, and the result is input to the comparators 704, 705, and 706.
Predetermined threshold signals DifTHr, DifTHg, and DifTHb are input to the comparators 704, 705, and 706, respectively. Output to the AND gate 707. As a result, the AND gate 707 is bright and dark. change When there are few, true is output.
The AveR, AveG, and AveB signals are input to comparators 708, 709, and 710, respectively, along with predetermined threshold signals DarkTHr, DarkTHg, and DarkTHb. The comparators 708, 709, and 710 each have an average value greater than the threshold value. True is output to the AND gate 711. Thus, true is output from the AND gate 711 when it is brighter than a certain level.
The AveR, AveG, and AveB signals are also input to the maximum value selection circuit 712 and the minimum value selection circuit 713, respectively. The maximum value selection circuit 712 determines the maximum value of the average value of each color component, and the minimum value selection circuit 713. Selects the minimum value of the average value of each color component and outputs it to the difference circuit 714. The difference circuit 714 is a circuit that outputs a difference between input signals, and the result is input to the comparator 715.
Each of the comparators 715 also receives a threshold signal BalTH, and the comparator 715 outputs true if the difference value is smaller than the threshold value. As a result, true is output from the comparator 715 when the color is achromatic.
The outputs of the AND gates 707 and 711 and the comparator 715 are input to the AND gate 716. As a result, when the target region has all the features of the background portion, the AND gate 716 outputs true.
[0027]
Referring to FIG. 8 showing a circuit unit that determines the background level in response to the result of the background determination, ExpR, ExpG, and ExpB are predicted background level signals output from the background level prediction circuit 404, and are image signals of the target region. The average values AveR, AveG, and AveB are input to the difference circuits 721, 722, and 723, respectively. The difference circuits 721, 722, and 723 are circuits that output the difference between the input signals, and the results are input to the comparators 724, 725, and 726.
The comparators 724, 725, and 726 also receive predetermined threshold signals DownTHr, DownTHg, and DownTHb, respectively. The comparators 724, 725, and 726 output true to the AND gate 727 if the difference values are smaller than the threshold values. Here, when the average value is darker than the predicted background level, the difference circuit outputs a positive value. As a result, the AND gate 727 outputs true when the average value of each color component is slightly darker or brighter than the predicted background level.
The output signals of the AND gate 716 and the AND gate 727 described with reference to FIG. 7 are input to the AND gate 728.
On the other hand, 729, 730, and 731 are weighted average circuits that perform the correction calculation of the predicted background level shown in the above equation (2), and the average values AveR, AveG, and AveB of the predicted background level signals ExpR, ExpG, and ExpB and the image signal of the target region. Is input, and the calculation result is output to the selectors 732, 733, and 734. ExpR, ExpG, and ExpB signals are input to the other input terminals of the selectors 732, 733, and 734, respectively, and operations of the selectors 732, 733, and 734 are controlled by the above-described AND gate 728 output signal.
Here, the output signal of the AND gate 728 is true when the target area has all the features of the background portion, and the average value of each color component is slightly darker or brighter than the predicted background level. At this time, the selectors 732, 733, 734 Selects and outputs the operation result of the weighted average circuit. Note that the latter condition, that is, the condition of the AND gate 727, prevents the background level from following the background of a highlight portion of an image where the density changes continuously as in a silver halide photograph. It is effective. That is, it is possible to identify the difference in density between the highlight portion and the true background portion based on whether the average value is slightly darker than the predicted background level, thereby preventing the background level from following.
As described above, according to this example, the background level determined by the background level determination circuit 406 is output from the selectors 732, 733, and 734.
[0028]
Returning to FIG. 4 again, the background correction performed in response to the background level determined by the background level determination circuit 406 will be described.
The background correction circuit 402 receives the background level detected by the background detection circuit 401 (the background level determined by the background level determination circuit 406) and also receives an image signal via the delay buffer 407. The circuit 402 corrects the input image signal according to the background level.
Here, an outline of the background correction method according to the present invention will be described.
FIG. 9 is a diagram showing the input / output characteristics of the image data in the background correction process, and compares the background correction of this embodiment with the characteristics when no correction is performed. Referring to the figure, when correction according to the background level is not performed, a straight line with a slope of 1 is assumed. In addition, a predetermined output level that is a correction target for the background portion is shown as a reference background level.
According to the correction method of the present embodiment, the conversion characteristics applied to the image signal in the target region (section divided in the main scanning direction shown in FIG. 3) are changed according to the background level detected by the background detection circuit 401. An example of conversion characteristics to be changed according to the detected background level will be described below with reference to cases where the detected background level is brighter and darker than the target reference background level.
[0029]
When the detected background level is brighter than the target reference background level (indicated by A on the input axis), in the area where the input image data is brighter than the detected background level, the input image data equivalent to the detected background level is converted to the reference background level, In the region where the input image data is darker than the detected background level, an input having the same value as the detected background level is used, using a characteristic for converting the input image data by a straight line having an inclination of 1 (shown as an A0-A1 characteristic line in FIG. 9). The image data is converted to a reference background level, and a characteristic for converting the input image data by a straight line having a predetermined inclination smaller than 1 (shown as an A0-A2 characteristic line in FIG. 9) is used. However, in a region where the conversion result is larger than that in the case where correction is not performed, the conversion without correction (characterized as A2-origin characteristic line in FIG. 9) is used.
On the other hand, if the detected background level is darker than the reference background level (indicated by B on the input axis), the input image data is brighter than the value obtained by subtracting the predetermined threshold NR from the detected background level (indicated by B ′ on the input axis). The input image data having the same value as the detected background level is converted to the reference background level, and a characteristic for converting the input image data by a straight line having a slope of 1 (shown as a B1-B0-B2 characteristic line in FIG. 9) is used. In a region where the input image data is darker than the above-mentioned value B ′, a characteristic of converting the input image data by a straight line (shown as a B2-B3 characteristic line in FIG. 9) passing through B2 and having a predetermined inclination greater than 1. Use. However, in a region where the conversion result is smaller than that in the case where correction is not performed, conversion without correction (characterized as B3-origin characteristic line in FIG. 9) is used.
In any of the above conversion characteristic examples, when the input image data is larger than the detected background level, conversion is performed with a straight line having a slope of 1 regardless of the brightness of the detected background level. This may be, for example, a characteristic line (A0-A1 ′) obtained by extending the characteristic line (A0-A2) or a characteristic line (B0-B1 ′) having the same inclination as the characteristic line (B2-B3). This is because when the detected background level changes across the reference background level, the conversion result of the image data becomes discontinuous, which is not preferable. For this reason, when the input image data is larger than the detected background level, this is prevented by performing conversion with a straight line having an inclination of 1.
In the above description, the threshold NR is set, and the conversion inclination of the input image data near the detected background level is also set to 1. Instead, it may be converted by a characteristic line (B1′-B2 ′) having a slope such as a characteristic line (B2-B3) passing through B0. However, in general, the image data of the background portion is not necessarily uniform and has a certain degree of variation, and if converted by a characteristic line having an inclination such as the characteristic line (B1′−B2 ′), this variation is also enlarged. , May appear dirty. For this reason, by setting the threshold value NR and setting the slope of the conversion characteristic line of the input image data in the vicinity of the detected background level to 1, such an increase in variation is prevented.
[0030]
Next, the background correction circuit 402 for realizing the above-described background correction method (refer to the description regarding FIG. 9) will be described in detail with reference to a circuit configuration example shown in FIG.
Referring to FIG. 10, the detected background level signal Detect input from the background detection circuit 401 is input to the comparator 901 together with the reference background level signal BaseW corresponding to the reference background level (see FIG. 9). It is determined whether or not the detected background level is darker than the reference background level.
The selector 902, the subtractor 903, and the comparator 904 constitute a circuit that determines whether to perform conversion using the characteristic line (B1-B0-B2) or the characteristic line (A0-A1) illustrated in FIG.
For this purpose, the selector 902 receives a threshold signal NR corresponding to the threshold NR (see FIG. 9) or a fixed value “0” and performs selection based on the output signal of the comparator 901. That is, the selector 902 selects and outputs the threshold signal NR when darker than the reference background level, and “0” when other (bright). The subtractor 903 subtracts the output signal of the selector 902 from the detected background level signal Detect. As a result, the value B2 or A0 shown in FIG. 9 is obtained, so that the comparator 904 compares the image signal ImgIN input from the delay buffer 407 (see FIG. 4) with the above determination.
The subtractor 905 and the adder 906 constitute a circuit that performs correction processing when the determination output of the comparator 904 is true. That is, the subtractor 905 calculates the difference between the reference background level signal BaseW and the detected background level signal Detect to obtain the correction amount, and adds the subtraction result to the image signal ImgIN to obtain the corrected image signal ImgOUTA. .
[0031]
Next, a case where conversion is performed using the characteristic line (B2-B3) or the characteristic line (A0-A2) shown in FIG. 9 will be described. In this embodiment, the conversion is realized using B2 or A0 as a base point.
The selector 907 selects the slope of the characteristic line (B2-B3) or the characteristic line (A0-A2). In this embodiment, the slopes are set to “7/4” and “3/4”, respectively. Is selected by the output signal of the above-described comparator 901 (a signal indicating whether or not the detected background level is darker than the reference background level). Further, the difference between B2 or A0 and the image signal ImgIN is obtained by the subtracter 908, and the slope selected by the selector 907 is multiplied by the multiplier 909. In this way, the corrected difference from B2 or A0 can be obtained.
On the other hand, the corrected value of B2 or A0 can be obtained by subtracting the output signal of the selector 902 from the reference background level signal BaseW by the subtractor 910. Accordingly, the output of the multiplier 909 is subtracted from the output of the subtracter 910 by the subtracter 911, and the image signal ImgOUTB corrected by the characteristic line (B2-B3) or the characteristic line (A0-A2) is obtained.
However, since the image signal ImgOUTB includes conversion on the characteristic line (B2-B3) or an extension line of the characteristic line (A0-A2), the result is compared by the comparator 912 with no correction. Accordingly, the image signal ImgOUTB or the image signal ImgIN (when correction is not performed) is selected by the selector 913, and the selector 913 outputs the image signal ImgOutB ′. The selectors 914 and 915 select the image signal ImgOUTB or the image signal ImgIN, respectively, according to the output signal of the comparator 901 (the signal indicating whether the detected background level is darker than the reference background level) and output it to the comparator 912. Thus, depending on whether or not the detected background level is darker than the reference background level, which one of the image signal ImgOUTB and the image signal ImgIN is selected is reversed.
Further, the image signal ImgOUTA or ImgOUTB ′ obtained as described above is used to determine whether or not to perform conversion using the output signal of the comparator 904 (characteristic line (B1-B0-B2) or characteristic line (A0-A1)). The image signal ImgOUT selected by the selector 916 controlled by the signal indicating the determination result and corrected according to the background level signal Sett is obtained.
[0032]
【The invention's effect】
According to the present invention, the image data is divided into a plurality of pixels in the main scanning direction. Shi The background level is detected for each section, and peripheral sections that have already been determined for detection (the section before and after the section corresponding to the target section on the previous main scan line and the section corresponding to the target section on the previous main scan line) ) Extracted from the image data of the current main scanning line in the current section of the prediction data from the background level (Maximum value and minimum value of each color component of the image signal within the interval of interest or an average value of each color component of the image signal) Based on the above, it is possible to determine the necessity of correction and to determine the detected background level, so that no pre-scan is required, and various background density fluctuations (particularly two-dimensional) caused by the document, reading device conditions, etc. Therefore, it is possible to detect the correct background level following the continuous fluctuation) and to optimize the background correction.
In addition, by using the weighted average of the background level of the surrounding sections as the prediction data of the background level, the background level determined based on the image data is transmitted in the main scanning direction. The prediction accuracy can be improved.
Further, the necessity of correction is determined based on the feature amount, and when it is determined that correction is necessary, a weighted average value of the predicted background level and the average value of the current average background level of the target section is calculated. Since it is calculated and determined as the detected background level (that is, the predicted background level is corrected), it is possible to prevent the detected background level from changing suddenly and to reduce processing unevenness due to the background correction processing. Become.
[0033]
In addition, to determine the background from the image data of the section of interest Extracted to Feature value (Maximum value and minimum value of each color component of the image signal within the interval of interest or an average value of each color component of the image signal) , Change of light and darkness, achromatic color, brightness above a certain level Soon It is determined whether or not the image of the target section is regarded as the background by threshold processing, and when it is determined that the image cannot be regarded as the background, the predicted background level (predicted from the surrounding sections) is determined as the detected background level. Therefore, it is possible to maintain the accuracy of the detected background data determined as the background level of the target section. In other words, by checking for changes in light and dark, the character part and halftone dot part are not determined to be the background, so it is possible to extract feature quantities that directly represent the features of the background part, thus preventing unnecessary fluctuations in the detected background level. By checking that the brightness is above a certain level, the detected background level does not become too dark, so it is possible to process a document whose image density changes continuously. By checking that it is, it becomes possible to improve the determination accuracy of the background.
[Brief description of the drawings]
FIG. 1 shows a schematic configuration of a mechanism part of a digital copying machine according to an embodiment of the present invention.
FIG. 2 shows a schematic configuration of an electrical component of a digital copying machine according to an embodiment of the present invention.
FIG. 3 is a diagram for explaining the principle of the background detection method according to the present invention.
FIG. 4 schematically shows a configuration example of a background removal circuit according to an embodiment of the present invention.
FIG. 5 illustrates a more detailed circuit configuration of a background level prediction circuit in the background removal circuit (FIG. 4).
6 illustrates a more detailed circuit configuration of a feature amount extraction circuit in the background removal circuit (FIG. 4).
7 illustrates a more detailed circuit configuration of a background level determination circuit in the background removal circuit (FIG. 4).
8 illustrates a more detailed circuit configuration of a background level determination circuit in the background removal circuit (FIG. 4).
FIG. 9 is a diagram showing input / output characteristics of image data in the background correction processing.
10 illustrates a more detailed circuit configuration of a background correction circuit in the background removal circuit (FIG. 4).
[Explanation of symbols]
101 ... Scanner unit, 102 ... Printer unit,
103 ... Original, 111 ... Image processing unit,
201 ... Operation display unit, 202 ... Digital copier,
203 ... System control unit, 213 ... Skin removal circuit,
401 ... background detection circuit, 402 ... background correction circuit,
403 ... background level buffer circuit, 404 ... background level prediction circuit,
405: Feature amount extraction circuit, 406: Background level determination circuit.

Claims (3)

A background level detecting means for detecting the background level of the original from raster format image data read by scanning the main and sub-scans, and the image data read by the background level detected by the background level detecting means. In the image processing apparatus having means for correcting to the reference background level,
The background level detection means is a means for dividing the image data into sections composed of a plurality of pixels in the main scanning direction, and detecting a background level for each section,
A background level storage means for storing the background level of each section on the previous main scanning line which has already been determined;
The background level of the current section on the main scan line corresponds to the section corresponding to the target section on the previous main scan line and the target section on the previous main scan line stored in the background level storage unit. A skin level predicting means for predicting by a weighted average of the skin levels of the surrounding sections composed of sections before and after the section to be performed;
Feature quantity extraction means for extracting the maximum value and the minimum value of each color component of the image signal in the notice section from the image data of the notice section as a feature quantity;
The difference value between the maximum value and the minimum value of each color component of the image signal extracted by the feature amount extraction unit is obtained, and when the obtained difference value of each color component is smaller than a predetermined threshold value, the determination is made that the attention section is regarded as the background. When the determination result cannot be regarded as the background, the background level predicted by the background level prediction unit is determined as the background level of the attention section, and when the determination result can be regarded as the background, the current main scanning line An image processing apparatus comprising: a background level determination unit that determines a weighted average of an average background level of a target section and a background level predicted by the background level prediction unit as a background level of the target section. A background level detecting means for detecting the background level of the original from raster format image data read by scanning the main and sub-scans, and the image data read by the background level detected by the background level detecting means. In the image processing apparatus having means for correcting to the reference background level,
The background level detection means is a means for dividing the image data into sections composed of a plurality of pixels in the main scanning direction, and detecting a background level for each section,
A background level storage means for storing the background level of each section on the previous main scanning line which has already been determined;
The background level of the current section on the main scan line corresponds to the section corresponding to the target section on the previous main scan line and the target section on the previous main scan line stored in the background level storage unit. A skin level predicting means for predicting by a weighted average of the skin levels of the surrounding sections composed of sections before and after the section to be performed;
Feature quantity extraction means for extracting an average value of each color component of the image signal in the notice section as a feature quantity from the image data of the notice section;
Calculates a difference value between the maximum value and the minimum value between the average value of the image signal each color component extracted by the feature extracting unit, when obtained difference value is smaller than a predetermined threshold value, regarded as the section of interest and background When the determination result cannot be regarded as the background, the background level predicted by the background level prediction unit is determined as the background level of the attention section, and when the determination result can be regarded as the background, the current main scanning line An image processing apparatus comprising: a background level determining unit that determines a weighted average of an average background level of an upper attention section and a background level predicted by the background level prediction means as a background level of the attention section. A background level detecting means for detecting the background level of the original from raster format image data read by scanning the main and sub-scans, and the image data read by the background level detected by the background level detecting means. In the image processing apparatus having means for correcting to the reference background level,
The background level detection means is a means for dividing the image data into sections composed of a plurality of pixels in the main scanning direction, and detecting a background level for each section,
A background level storage means for storing the background level of each section on the previous main scanning line which has already been determined;
The background level of the current section on the main scan line corresponds to the section corresponding to the target section on the previous main scan line and the target section on the previous main scan line stored in the background level storage unit. A skin level predicting means for predicting by a weighted average of the skin levels of the surrounding sections composed of sections before and after the section to be performed;
Feature quantity extraction means for extracting an average value of each color component of the image signal in the notice section as a feature quantity from the image data of the notice section;
When the average value of each color component of the image signal extracted by the feature amount extraction unit is larger than a predetermined threshold, it is determined that the target section is the background, and when the determination result cannot be considered the background, the background level prediction The background level predicted by the means is determined as the background level of the attention section, and when the determination result can be regarded as the background, the average background level of the attention section on the current main scanning line and the background level prediction means An image processing apparatus comprising a background level determining means for determining a weighted average with a background level as a background level of a section of interest.

Images