JP3581596B2 - Stripe detection device, photoconductor drum inspection device using the same, liquid crystal panel inspection device using the same, stripe detection method, and medium recording stripe detection program - Google Patents

Description

【００５０】
図５は、変換係数Ｓ_ｕｖに対応する基底画像を示す図である。基底画像の濃淡変化はＣＯＳ波形となるが、図表現の便宜上ＣＯＳの値が正となる部分を白、負となる部分を黒として２値で表わすものとする。また、図５に示す基底画像のそれぞれは、６４個の直交基底画像行列の（ｕ，ｖ）位置の画像に対応しており、変換係数Ｓ_ｕｖの値が大きいと（ｕ，ｖ）位置の基底画像の成分が強く含まれていることを意味している。
【００５１】
図５から判るように、基底画像は左から右に進むにつれて高周波の水平周波成分（縦縞）を多く含み、上から下に進むにつれて高周波の垂直周波成分（横縞）を多く含むことになり、８×８画素の矩形ブロックＭＰはこの６４個の基底画像の線形結合で表わされることになる。したがって、変換係数マトリックスＭＳにおけるＳ_ｕｖ値の分布状態を読み取ることによって、どの周波数成分が強く含まれているか、およびどの方向の縞がどれほど強く含まれているかという縞の特徴量（向き、周期、振幅）を容易に把握することが可能になる。
【００５２】
しかし、高周波領域には一般に白色ノイズが含まれやすく、画像分解能の観点からも高周波領域を扱うことは無意味であるので、高周波成分を切り捨てて比較的低周波領域に着目する方が有効である。したがって、６４個の変換係数Ｓ_ｕｖのうち、図５の破線で囲んだ基底画像に対応する１２個の変換係数Ｅ＝｛（ｕ，ｖ）＝（０，２），（０，３），（０，４），（１，２），（１，３），（１，１），（２，２），（２，１），（３，１），（２，０），（３，０），（４，０）｝のみを縞の評価対象として選択することにする。なお、これ以外の変換係数の組み合わせであっても良いことは言うまでもない。
【００５３】
図６は、上記選択された１２個の変換係数を表にしたものであり、縞方向および縞周期で分類している。横グループ（Ｙ＿Ｅ）には変換係数Ｓ_０２，Ｓ_０３およびＳ_０４が含まれ、図５に示す基底画像から判るように横方向の縞を含んでいる。また、横斜めグループ（ＹＮ＿Ｅ）には変換係数Ｓ_１２およびＳ_１３が含まれ、図５に示す基底画像から判るように横方向に近い斜めの縞を含んでいる。また、斜めグループ（Ｎ＿Ｅ）には変換係数Ｓ_１１およびＳ_２２が含まれ、図５に示す基底画像から判るように斜めの縞を含んでいる。また、縦斜めグループ（ＴＮ＿Ｅ）には変換係数Ｓ_２１およびＳ_３１が含まれ、図５に示す基底画像から判るように縦方向に近い斜めの縞を含んでいる。さらには、縦グループ（Ｔ＿Ｅ）には変換係数Ｓ_２０，Ｓ_３０およびＳ_４０が含まれ、図５に示す基底画像から判るように縦方向の縞を含んでいる。
【００５４】
検出対象の縞周期が、図６に示す縞周期よりも大きい場合には、矩形ブロックＭＰのサイズを２倍またはそれ以上に設定するか、矩形ブロックＭＰのサイズはそのままで撮像画像ＭＩを１／２またはそれ以上に圧縮すれば良い。
【００５５】
ＤＣＴ処理部２２は、上述した矩形ブロック単位のＤＣＴ処理を６４×６４個の全ての矩形ブロック、すなわち矩形ブロックマトリックスＭＢの全要素に対して行なうことによって、図６に示す１２種類の変換係数のそれぞれについて、６４×６４の２次元マトリックスＭＳ_ｕｖで表現される縞特徴量の撮像画像全域にわたる空間分布を求めることが可能になる。ＭＳ_ｕｖは、次式によって表わすことができる。
【００５６】
【数２】

【００５７】
再び、図２に示すフローチャートの説明に戻る。次に、平均値／標準偏差算出部２３は、２次元マトリックスＭＳ_ｕｖ内の６４×６４個の値の平均値および標準偏差を算出する（Ｓ３）。撮像画像に縞が含まれない場合には、２次元マトリックスＭＳ_ｕｖ内における値のヒストグラムは、０付近に度数が集中した正規分布となる。また、撮像画像内の一部の微小領域に縞が発生する場合には、２次元マトリックスＭＳ_ｕｖ内における値のヒストグラムは、大局的には正規分布となるが、縞を含む矩形ブロックの変換係数Ｓ_ｕｖは０から離れた値となるため、正規分布から離れた位置に縞に対応するピークが現れることになる。
【００５８】
この統計的原理を利用するために、平均値／標準偏差算出部２３は、図６に示す１２種類の変換係数のそれぞれについて正規分布の平均値μ_ｕｖ；（ｕ，ｖ）∈Ｅ、および標準偏差σ_ｕｖ；（ｕ，ｖ）∈Ｅを算出する。
【００５９】
そして、縞候補ブロック抽出部２４は、標準偏差σ_ｕｖを所定の定数Ｃ_ｕｖ；（ｕ，ｖ）∈Ｅ倍した値を閾値とし、変換係数Ｓ_ｕｖと平均値μ_ｕｖとの間の距離が閾値以上である矩形ブロックの集合を縞候補ブロック群として抽出する（Ｓ４）。図７は、縞候補ブロック群の抽出の一例を示しており、変換係数Ｓ_ｕｖと平均値μ_ｕｖとの間の距離が閾値以上である矩形ブロック群３７が縞候補ブロック群として抽出される。この縞候補ブロック群ＧＢ_ｕｖは次式によって表わされる。
【００６０】
【数３】

【００６１】
なお、特殊な場合として、縞が撮像画像の全域または大半の領域で発生する場合もあるため、算出された標準偏差σ_ｕｖの値をそのまま用いるのではなく、その値を直接管理して撮像画像全域に対して評価を行なう方法と併用することが望ましい。
【００６２】
また、縞候補ブロック抽出部２４は、閾値として標準偏差σ_ｕｖをＣ_ｕｖ倍した値を用いることにより、撮像画像全域の中から相対的に変換係数値が０から離れている部分、すなわち相対的に縞コントラスト（縞の振幅）の高い部分を抽出している。この処理によって、撮像の際の照明強度の変化や検査対象物表面の光反射率の変化による縞コントラストの絶対的な変化の影響をなくし、ロバスト性の高い検査が可能となる。
【００６３】
再び、図２に示すフローチャートの説明に戻る。次に、方向別フィルタ処理部２５は、縞候補ブロック抽出部２４によって抽出された１２種類の変換係数Ｓ_ｕｖに対応する縞候補ブロック群ＧＢ_ｕｖに対して、縞の２次元的な形状特性を考慮したフィルタ処理を行ない、所定の特徴量を有する縞の発生領域を検出する（Ｓ５）。
【００６４】
縞の２次元的な形状特性、特に縞の方向や縞の周期の変化は２次元空間的には滑らかであるため、微小領域におけるその変化は無視できる程度である。方向別フィルタ処理部２５は、図６に示す５つの方向別グループのそれぞれに対して、縞の２次元的な形状特性を考慮したオペレータを用いてフィルタ処理を行なう。図８は、そのオペレータの形状の一例であって、（ａ）に示す横グループ用マスク（Ｙ＿ＭＳＫ）、（ｂ）に示す横斜めグループ用マスク（ＹＮ＿ＭＳＫ）、（ｃ）に示す斜めグループ用マスク（Ｎ＿ＭＳＫ）、（ｄ）に示す縦斜めグループ用マスク（ＴＮ＿ＭＳＫ）および（ｅ）に示す縦グループ用マスク（Ｔ＿ＭＳＫ）から構成される。なお、オペレータは図８（ａ）〜（ｅ）に示す以外の形状であっても良い。
【００６５】
たとえば、横グループＹ＿Ｅ＝｛（ｕ，ｖ）＝（０，２），（０，３），（０，４）｝に含まれる変換係数にフィルタ処理を行なう場合、３つの変換係数Ｓ_０２，Ｓ_０３およびＳ_０４に対応する３種類の縞候補ブロック群ＧＢ_０２，ＧＢ_０３およびＧＢ_０４から構成される横縞候補ブロック群Ｙ＿ＧＢは次式によって表わすことができる。
【００６６】
【数４】

【００６７】
この横縞候補ブロック群Ｙ＿ＧＢの抽出は、６４×６４個の矩形ブロックマトリックスＭＢ上を順に走査し、上記３種類の縞候補ブロック群ＧＢ_０２，ＧＢ_０３およびＧＢ_０４のうち、少なくとも１つの縞候補ブロックが存在する矩形ブロックを探索することによって可能である。そして、方向別フィルタ処理部２５は、この横縞候補ブロック群Ｙ＿ＧＢの中から、図８（ａ）に示す横グループ用マスクＹ＿ＭＳＫの形で連続して横縞候補ブロックが存在するブロックのみを横縞ブロック群Ｙ＿ＷＢとして残すことにより、孤立した縞候補ブロックを除去する。この連続する横縞候補ブロックのみを抽出する操作を、演算子ｆ_［Ｙ＿_ＭＳＫ］・と定義すれば、横縞ブロック群Ｙ＿ＷＢは次式によって表わすことができる。
【００６８】
【数５】

【００６９】
図９は、この横縞ブロックの抽出を模式的に示す図である。図９（ａ）〜図９（ｃ）に示す縞候補ブロック群ＧＢ_０２，ＧＢ_０３およびＧＢ_０４のうち、少なくとも１つの縞候補ブロックが存在するブロックを探索することによって、図９（ｄ）に示す横縞候補ブロック群Ｙ＿ＧＢを抽出することができる。そして、方向別フィルタ処理部２５が横グループ用マスクＹ＿ＭＳＫを用いて、図９（ｄ）に示す横縞候補ブロック群Ｙ＿ＧＢに対してフィルタ処理を行なうことによって、図９（ｅ）に示す横縞ブロック群Ｙ＿ＷＢを抽出する。
【００７０】
方向別フィルタ処理部２５は、同様のフィルタ処理を横斜めグループＹＮ＿Ｅ、斜めグループＮ＿Ｅ、縦斜めグループＴＮ＿Ｅ、および縦グループＴ＿Ｅに対して行なう。その結果抽出された横斜め縞ブロック群ＹＮ＿ＷＢ、斜め縞ブロック群Ｎ＿ＷＢ、縦斜め縞ブロック群ＴＮ＿ＷＢ、縦縞ブロック群Ｔ＿ＷＢ、および上述した横縞ブロック群Ｙ＿ＷＢの論理和を求めることにより、全方向の縞ブロック群を重ねて得られた全方位縞ブロック群ＷＢを、最終結果である縞検出領域として出力する。この全方位縞ブロック群ＷＢは、次式によって表わすことができる。
【００７１】
【数６】

【００７２】
以上説明した縞検出装置は、コンピュータに縞検出プログラムを実行させることによって実現可能である。このコンピュータによって縞検出装置を実現する方法について以下に説明するが、本実施の形態における縞検出装置はこれに限られるものではない。たとえば、プログラム処理の中で最も時間を要するＤＣＴ処理をハードウェア化し、処理の高速化を図ることも可能である。
【００７３】
図１０は、本発明の縞検出装置の外観例を示す図である。縞検出装置は、コンピュータ本体１、グラフィックディスプレイ装置２、磁気テープ４が装着される磁気テープ装置３、キーボード５、マウス６、ＣＤ−ＲＯＭ（Ｃｏｍｐａｃｔ Ｄｉｓｃ−Ｒｅａｄ Ｏｎｌｙ Ｍｅｍｏｒｙ ）８が装着されるＣＤ−ＲＯＭ装置７、および通信モデム９を含む。縞検出プログラムは、磁気テープ４またはＣＤ―ＲＯＭ８等の記憶媒体によって供給され、コンピュータ本体１によって実行される。また、縞検出プログラムは他のコンピュータより通信回線を経由し、通信モデム９を介してコンピュータ本体１に供給されてもよい。
【００７４】
図１１は、本実施の形態における縞検出装置の構成例を示すブロック図である。図１０に示すコンピュータ本体１は、ＣＰＵ１０、ＲＯＭ（Ｒｅａｄ Ｏｎｌｙ Ｍｅｍｏｒｙ）１１、ＲＡＭ（Ｒａｎｄｏｍ Ａｃｃｅｓｓ Ｍｅｍｏｒｙ）１２およびハードディスク１３を含む。ＣＰＵ１０は、グラフィックディスプレイ装置２、磁気テープ装置３、キーボード５、マウス６、ＣＤ−ＲＯＭ装置７、通信モデム９、ＲＯＭ１１、ＲＡＭ１２またはハードディスク１３との間でデータを入出力しながら処理を行う。磁気テープ４またはＣＤ−ＲＯＭ８に記録された縞検出プログラムは、ＣＰＵ１０により磁気テープ装置３またはＣＤ−ＲＯＭ装置７を介して一旦ハードディスク１３に格納される。ＣＰＵ１０は、ハードディスク１３から適宜縞検出プログラムをＲＡＭ１２にロードして実行することによって縞の検出を行う。
【００７５】
以上説明したように、本実施の形態における縞検出装置によれば、撮像画像を矩形ブロックに細分化し、矩形ブロックのそれぞれにＤＣＴ処理を行なって縞の空間周波数解析を行なうようにしたので、縞の方向を含めた２次元的な特徴解析が可能となった。
【００７６】
また、矩形ブロック単位で縞の特徴量（縞の向き、周期、振幅）を抽出するようにしたので、矩形ブロックの大きさ程度の微小領域に発生する縞であっても検出することが可能となった。
【００７７】
また、２次元マトリックス内で画像の振幅の値が比較的大きい矩形ブロックを縞候補ブロックとして抽出するようにしたので、撮像の際の照明強度の変化や検査対象物の表面の光反射率の変化による縞コントラストの変化による影響をなくし、ロバスト性の高い検出が可能となった。
【００７８】
さらには、縞の２次元的な形状特性を考慮したフィルタ処理を行なうようにしたので、白色ノイズ、検査対象物の表面の反射ムラまたはテクスチャによる明度変化の影響を除去することが可能となった。
【００７９】
（実施の形態２）
実施の形態２における感光体ドラム検査装置は、実施の形態１における縞検出装置を複写機等に用いられる感光体ドラム表面の薄膜に応用したものである。
【００８０】
図１２は、感光体ドラムの表面付近の断面構造を示している。感光体ドラム４０は、アルミ素管４１と、アルミ素管４１の表面に形成された感光体である多層薄膜４２とを含む。この感光体ドラム４０の表面に、波長λの単一波長の光源４３を照射すると、多層薄膜４２の厚さｄに依存した光干渉が発生する。そのため、厚さｄに変化がある場合には地図の等高線のような干渉縞が観察される。多層薄膜４２の厚さが急激に変化する不良部分があると、複写機においては複写紙上にこの不良部分に対応した濃淡ムラが発生する。この不良部分においては、干渉縞の間隔が狭くなるので、干渉縞の縞周期が短い部分を検出することによって感光体ドラム４０の不良箇所を検出することが可能となる。
【００８１】
図１３は、本実施の形態における感光体ドラム検査装置の概略構成を示すブロック図である。この感光体ドラム検査装置は、単一波長の光源４３と、感光体ドラム４０からの正反射光を撮像するラインセンサ４４と、感光体ドラム４０を回転させる回転駆動部４５と、回転駆動部４５の回転を制御する回転制御部４６と、ラインセンサ４４によって撮像された画像を記憶する画像メモリ４７と、縞検出を行なう縞検出処理部４８とを含む。
【００８２】
光源４３は、光が感光体ドラム４０の長手方向全域に照射されるように配置されている。また、ラインセンサ４４は、感光体ドラム４０の長手方向全域を撮像するように配置されている。回転制御部４６は、ラインセンサ４４による撮像に同期して回転駆動部４５の回転を制御する。たとえば、画像メモリ４７の記憶容量が２０４８画素×２０４８画素分であれば、回転制御部４６は感光体ドラム４０の表面全域の画像（１フレーム）が画像メモリ４７の記憶容量に収まるように回転駆動部４５を制御して感光体ドラム４０を回転させる。
【００８３】
縞検出処理部４８は、画像メモリ４７に格納された１フレーム分の画像に対して縞検出処理を行なうが、実施の形態１における縞検出装置の処理手順と同じであるので、詳細な説明は繰り返さない。
【００８４】
以上説明したように、本実施の形態における感光体ドラム検査装置によれば、感光体ドラムの表面の不良部分を自動的に検出することが可能となり、実施の形態１における縞検出装置において説明した効果を感光体ドラムの検査においても得ることが可能となった。
【００８５】
（実施の形態３）
実施の形態３における液晶パネル検査装置は、実施の形態１における縞検出装置を液晶パネルにおける貼り合わせガラス間のギャップムラ検査に応用したものである。
【００８６】
図１４は、液晶パネル４９の断面構造を示している。液晶パネル４９は、２枚のガラス５０と、この２枚のガラス５０間のギャップに注入される液晶５１とを含む。この液晶パネル４９の表面に波長λの単一波長の光源５３を照射すると、ガラス間ギャップの値ｄに依存した光干渉が発生する。そのため、ガラス間ギャップの値ｄに変化がある場合には地図の等高線のような干渉縞が観察される。
【００８７】
液晶パネルの製造工程においては、ガラス間ギャップに液晶５１が注入されるため、ガラス間ギャップのギャップムラが製品品質を左右することになる。したがって、干渉縞が発生する部分を抽出することによってギャップムラを検出し、液晶パネル４９の不良を発見することが可能となる。
【００８８】
図１５は、本実施の形態における液晶パネル検査装置の概略構成を示すブロック図である。この液晶パネル検査装置は、単一波長の光源５３と、液晶パネル４９に発生した干渉縞を撮像するラインセンサ５２と、ラインセンサ５２によって撮像された画像を記憶する画像メモリ４７と、縞検出を行なう縞検出処理部４８とを含む。
【００８９】
光源５３は、光が液晶パネル４９の表面の全域に拡散照射されるように配置されている。また、ラインセンサ５２は、液晶パネル４９の全域を撮像するように配置されている。たとえば、画像メモリ４７の記憶容量が５１２画素×５１２画素分であれば、液晶パネル４９の表面全域の画像（１フレーム）が画像メモリ４７の記憶容量に収まるように撮像が行なわれる。
【００９０】
縞検出処理部４８は、画像メモリ４７に格納された１フレーム分の画像に対して縞検出処理を行なうが、実施の形態１における縞検出装置の処理手順と同じであるので、詳細な説明は繰り返さない。
【００９１】
以上説明したように、本実施の形態における液晶パネル検査装置によれば、液晶パネルの表面の不良部分を自動的に検出することが可能となり、実施の形態１における縞検出装置において説明した効果を液晶パネルの検査においても得ることが可能となった。
【００９２】
今回開示された実施の形態は、すべての点で例示であって制限的なものではないと考えられるべきである。本発明の範囲は上記した説明ではなくて特許請求の範囲によって示され、特許請求の範囲と均等の意味および範囲内でのすべての変更が含まれることが意図される。
【図面の簡単な説明】
【図１】本発明の実施の形態１における縞検出装置の概略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における縞検出装置の処理手順を説明するためのフローチャートである。
【図３】検査対象物の表面に現れる光学的縞を模式的に示す図である。
【図４】撮像画像を矩形ブロックに細分化した様子を示す図である。
【図５】変換係数Ｓ_ｕｖに対応する基底画像を示す図である。
【図６】選択された１２個の変換係数を表にしたものである。
【図７】縞候補ブロック群の抽出の一例を示す図である。
【図８】フィルタ処理に使用されるオペレータの形状の一例を示す図である。
【図９】横縞ブロック群の抽出を説明するための図である。
【図１０】実施の形態１における縞検出を実現するコンピュータの外観の一例を示す図である。
【図１１】図１０に示すコンピュータの構成例を示す図である。
【図１２】感光体ドラムの表面付近の断面構造を示す図である。
【図１３】感光体ドラム検査装置の概略構成を示す図である。
【図１４】液晶パネルの断面構造を示す図である。
【図１５】液晶パネル検査装置の概略構成を示す図である。
【図１６】従来の縞の周期の計測における問題点を説明するための図である。
【符号の説明】
１ コンピュータ本体
２ グラフィックディスプレイ装置
３ 磁気テープ装置
４ 磁気テープ
５ キーボード
６ マウス
７ ＣＤ−ＲＯＭ装置
８ ＣＤ−ＲＯＭ
９ 通信モデム
１０ ＣＰＵ
１１ ＲＯＭ
１２ ＲＡＭ
１３ ハードディスク
２１ 矩形ブロック分割部
２２ ＤＣＴ処理部
２３ 平均値／標準偏差算出部
２４ 縞候補ブロック抽出部
２５ 方向別フィルタ処理部
２６ 処理結果マージ部
４０ 感光体ドラム
４１ アルミ素管
４２ 多層薄膜
４３，５３ 光源
４４，５２ ラインセンサ
４５ 回転駆動部
４６ 回転制御部
４７ 画像メモリ
４８ 縞検出処理部
４９ 液晶パネル
５０ ガラス
５１ 液晶[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a technique for detecting and evaluating optical fringes such as interference fringes and moiré fringes appearing on the surface of an inspection object, and in particular, calculating a fringe feature amount including the direction of the optical fringe to evaluate the inspection object. The present invention relates to a fringe detection device for evaluating the surface condition of a photoconductor, a photoreceptor drum inspection device using the same, a liquid crystal panel inspection device using the same, a fringe detection method, and a medium on which a fringe detection program is recorded.
[0002]
[Prior art]
In recent years, along with mass production of products, various technologies for automatically inspecting an inspection object have been developed. One of the technologies is to image a stripe generated optically depending on the surface state of the inspection object. There is a technique for detecting a defective portion of an inspection object by analyzing the stripe. Related to this technique are the inventions disclosed in JP-A-7-260457, JP-A-9-61125 and JP-A-10-132535.
[0003]
A surface inspection apparatus disclosed in Japanese Patent Application Laid-Open No. 7-260457 includes an imaging unit that captures an interference fringe pattern image of a surface to be inspected, an A / D conversion unit that obtains image data based on a video signal from the imaging unit, An interference fringe state detection unit that detects an arrangement state of interference fringes forming an interference fringe pattern image represented by each unit section corresponding to a predetermined unit period in the image data obtained by the A / D conversion unit; A defect detection unit for detecting a change in interference fringes due to a defect on the surface to be inspected based on a detection output from the stripe state detection unit.
[0004]
This surface inspection apparatus detects a defect on a surface to be inspected by utilizing the fact that interference fringes at a defective portion become dense. That is, the interference fringe state detection unit calculates a fringe interval (fringe cycle) based on a grayscale profile of each horizontal line for a captured image including interference fringes. Then, the defect detection unit determines whether the number of fringes in each of a predetermined range converted from the fringe period on the inspected surface or the portion where the fringe period on the inspected surface is smaller than the fringe period on the predetermined non-defective product is a predetermined number. A portion having more stripes than a good product is detected as a defective portion.
[0005]
In addition, a laminated plate inspection system disclosed in Japanese Patent Application Laid-Open No. 9-61125 includes a laser light source that irradiates a laser beam to an object to be inspected (a laminated plate obtained by laminating two flat glass plates); It includes a TV camera that captures an interference fringe image of an inspection object that appears due to irradiation, and a CPU (Central Processing Unit) that determines a region where the fringe pitch of the interference fringes captured by the TV camera is higher than a predetermined value.
[0006]
The CPU performs an FFT (fast Fourier transform) process on the interference fringe image captured by the TV camera, performs a high-pass filter process so as to leave a portion having a predetermined frequency or higher, and then performs an inverse FFT process. The CPU detects, from the image obtained by the inverse FFT processing, a portion left as a high frequency region as a defective portion.
[0007]
Further, the surface inspection apparatus disclosed in Japanese Patent Application Laid-Open No. 10-132535 includes a light projecting unit that divides a laser beam into two and irradiates the surface to be inspected, and a light receiving unit that receives light reflected from the surface to be inspected. An image conversion unit for optically Fourier-transforming interference fringes formed on the surface to be inspected, and a display unit for displaying an optical Fourier-transformed image formed by the image conversion unit. The presence or absence of a defect on the surface to be inspected is clarified by displaying the optical Fourier transform image formed by the image conversion unit on the display unit.
[0008]
This surface inspection apparatus performs inspection by spot measurement that irradiates a surface to be inspected with laser light, and enables two-dimensional scanning with laser light to detect a defective portion in a wide area.
[0009]
[Problems to be solved by the invention]
However, in the surface inspection apparatus disclosed in Japanese Patent Application Laid-Open No. Hei 7-260457, the period of the interference fringes can be accurately measured only in the horizontal direction 103 of the captured image 102 as shown in FIG. On the other hand, this is the case where the stripe 101 is vertical. That is, in this case, the actual period D of the interference fringes 101 matches the measured value d. However, when the interference fringes 101 are not perpendicular to the horizontal direction 103 of the captured image 102 as shown in FIG. 16B (the interference fringes 101 are at an angle θ (θ <θ with respect to the horizontal direction 103 of the captured image). 90 °), there is a problem that the period d of the interference fringes 101 calculated from the shading profile is larger (1 / sin θ times) than the period D of the actual interference fringes 101. .
[0010]
When the fringe 101 coincides with the horizontal direction 103 of the captured image 102 as shown in FIG. 16C, the feature of the interference fringe 101 does not appear in the shading profile, and the measurement of the period of the interference fringe 101 is practically impossible. Impossible. Also, it is extremely rare that the interference fringes 101 become straight lines as shown in FIGS. 16A to 16C, and generally, the interference fringes 101 are curved lines. The value of the angle θ of the stripe 101 varies depending on the measurement position. Therefore, it is impossible to calculate the angle θ in line-by-line (fringe cycle) measurement, and it is impossible to accurately measure the cycle of the interference fringe 101 at an arbitrary position.
[0011]
Further, the laminated board inspection system disclosed in Japanese Patent Application Laid-Open No. 9-61125 is effective for stripes generated entirely in the inspection area, but is effective for stripes partially generated in the inspection area. There is a problem peculiar to FFT that it is not effective. That is, in the frequency distribution obtained by the FFT, the characteristics of the stripes partially generated in the inspection area do not appear so strongly, so that there is a problem that the sensitivity is weak and the detection of the stripes becomes difficult.
[0012]
Further, even if a fringe that occurs partially in the inspection area can be detected, it is only possible to confirm that the fringe exists in the inspection area, and it is not possible to detect the position of the fringe. there were. This problem is caused by the fact that the Fourier transform is a signal transform based on a sine wave that is repeated infinitely. Therefore, it can be said that the FFT can be effectively used in the image signal processing when the frequency of the target image is known in advance and the purpose is to remove high-frequency noise that spreads over the entire inspection area.
[0013]
The surface inspection apparatus disclosed in Japanese Patent Application Laid-Open No. 10-132535 solves the problem of the Fourier transform. That is, in this surface inspection apparatus, stripes are detected by performing an optical Fourier transform on an image in a minute range (laser spot diameter). By performing this processing while scanning two-dimensionally, a wide range of measurement is possible. Therefore, it is possible to accurately determine the presence or absence of a stripe that spreads over a region having a size approximately equal to or larger than the laser spot diameter. In addition, since the position of the stripe can be specified from the scanning position at that time, if the laser spot diameter is made sufficiently small, the portion where the stripe occurs can be detected with high sensitivity.
[0014]
However, since the process is not a process taking into account the direction of the stripes or the two-dimensional continuity, it is affected by brightness changes due to images other than stripes appearing due to optical interference, for example, white noise, reflectance unevenness on the surface of the inspection object, or texture. In such a case, there is a problem that these cannot be removed separately from the interference fringes.
[0015]
The present invention has been made to solve the above problems, and a first object is to provide a fringe detecting device capable of calculating a fringe feature amount including a fringe direction.
[0016]
A second object of the present invention is to provide a fringe detecting device capable of detecting fringes with high robustness, which is not affected by a change in fringe contrast due to a change in illumination intensity or a change in light reflectance of the surface of the inspection object.
[0017]
A third object is to provide a fringe detecting apparatus capable of eliminating the influence of brightness change due to white noise, reflectance unevenness on the surface of an inspection object, texture, or the like.
[0018]
A fourth object is to provide a photoconductor drum inspection apparatus capable of calculating a fringe feature amount including a direction of a fringe and accurately detecting a defective portion.
[0019]
A fifth object is to provide a liquid crystal panel inspection apparatus capable of calculating a fringe feature amount including the direction of a fringe and accurately detecting a defective portion.
[0020]
[Means for Solving the Problems]
The fringe detecting device according to claim 1 converts an image of an optical fringe of the inspection object captured into an image. , Divided into multiple directions independent of the direction of the optical fringes, A dividing unit for dividing the block into a plurality of pixel sets; a processing unit for performing a discrete cosine transform on information in the block divided by the dividing unit; Inspection object Evaluation including the direction of optical fringes Evaluation means.
[0021]
The processing means performs discrete cosine transform on the information in the block divided by the dividing means, so that it is possible to calculate the fringe feature amount including the direction of the fringe, and it is possible to accurately detect the fringe. Become.
[0036]
Claim 2 The photosensitive drum inspection device described in the above, light emitting means for irradiating light to the photosensitive drum, imaging means for imaging the optical fringes of the photosensitive drum generated by the reflected light of the light from the light emitting means, A control unit for controlling the rotation of the photosensitive drum, and an image captured by the imaging unit is divided into a plurality of directions independent of the direction of the optical fringe, and divided into blocks each including a plurality of pixel sets. Dividing means, a processing means for performing a discrete cosine transform on the information in the block divided by the dividing means, and the direction of the optical stripe of the photosensitive drum based on the processing result by the processing means. Evaluation means for performing the evaluation.
[0037]
Since the processing unit performs the discrete cosine transform on the information in the block divided by the dividing unit, it is possible to calculate the stripe characteristic amount including the direction of the stripe of the photosensitive drum, and to perform the inspection of the photosensitive drum. It is possible to perform it accurately.
[0038]
Claim 3 A light emitting unit for irradiating light to the liquid crystal panel, an imaging unit for imaging optical fringes of the liquid crystal panel generated by reflected light of the light from the light emitting unit, and an imaging unit Means for dividing the image captured by the dividing means into a plurality of directions independent of the direction of the optical fringe, and dividing the image into blocks composed of a plurality of pixel sets, and information in the blocks divided by the dividing means Processing means for performing a discrete cosine transform on the image data, and evaluation means for performing an evaluation including the direction of the optical fringe of the liquid crystal panel based on the processing result of the processing means.
[0039]
Since the processing means performs discrete cosine transform on the information in the block divided by the dividing means, it is possible to calculate the stripe characteristic amount including the direction of the stripe of the liquid crystal panel, and to accurately inspect the liquid crystal panel. It is possible to do.
[0040]
Claim 4 The method for detecting a fringe according to the step of dividing the captured image of the optical fringe of the inspection object into a plurality of directions independent of the direction of the optical fringe, and dividing the image into blocks formed of a plurality of pixel sets And performing a discrete cosine transform on the information in the divided blocks; and performing an evaluation including the direction of the optical fringe of the inspection object based on the processing result of the discrete cosine transform.
[0041]
Since the discrete cosine transform is performed on the information in the divided blocks, the fringe feature amount including the direction of the fringe can be calculated, and the fringe can be accurately detected.
[0042]
Claim 5 The fringe detection program recorded on the medium described in the above, the captured image of the optical fringe of the inspection object is divided into a plurality of directions that do not depend on the direction of the optical fringe, and is configured by a plurality of pixel sets Dividing into blocks, performing a discrete cosine transform on information in the divided blocks, and performing an evaluation including a direction of an optical fringe of the inspection object based on a processing result of the discrete cosine transform And
[0043]
Since the discrete cosine transform is performed on the information in the divided blocks, the fringe feature amount including the direction of the fringe can be calculated, and the fringe can be accurately detected.
[0044]
BEST MODE FOR CARRYING OUT THE INVENTION
(Embodiment 1)
FIG. 1 is a block diagram showing a schematic configuration of the fringe detecting device according to the first embodiment of the present invention. This fringe detection device includes a rectangular block dividing unit 21 that divides a captured image into rectangular blocks each including a plurality of pixel sets, a DCT processing unit 22 that performs DCT (discrete cosine transform) processing on the rectangular blocks, An average value / standard deviation calculation unit 23 for calculating an average value and an average deviation for each conversion coefficient obtained by the processing, and a fringe candidate based on the average value and the standard deviation calculated by the average value / standard deviation calculation unit 23. The directionality is considered for a stripe candidate block extraction unit 24 that extracts rectangular blocks and a set of rectangular blocks serving as stripe candidates extracted by the stripe candidate block extraction unit 24 (hereinafter, referred to as a stripe candidate block group). Direction-dependent filter processing unit 25 that performs the filtered processing, and a processing result map that superimposes the processing results obtained by the direction-specific filter processing unit 25. And a shaped portion 26.
[0045]
FIG. 2 is a flowchart for explaining a processing procedure of the fringe detecting device in the present embodiment. The processing procedure of the fringe detection device will be described with reference to other drawings as appropriate.
[0046]
FIG. 3 schematically shows an image when an optical fringe such as an interference fringe generated depending on the surface state of the inspection object is imaged. FIG. 3 shows a state in which the optical fringe 31 is captured in the entire captured image 32, and the resolution is 512 × 512 pixels (pixel MI = ｛I _ij 0 ≦ i, j ≦ 511}. The resolution of the captured image 32 may be another number of pixels, and the pixels 34 may not be square.
[0047]
First, the rectangular block dividing unit 21 divides the captured image 32 into rectangular blocks each including a plurality of pixel sets (S1). FIG. 4 shows a state where the captured image 32 is subdivided into rectangular blocks 35. This rectangular block is converted into a square block of 8 × 8 pixels (MP = ｛P _xy ; 0 ≦ x, y ≦ 7}, the captured image 32 is converted into a 64 × 64 block matrix (MB = {MP _mn 0 ≦ x, y ≦ 63}. Note that the rectangular block MP may have any other size, and may not be a square.
[0048]
Next, the DCT processing unit 22 performs a DCT process for each rectangular block obtained by the rectangular block dividing unit 21 to obtain a transform coefficient matrix composed of 8 × 8 (in the case of 8 × 8 pixels) transform coefficients ( S2). The position coordinates of each pixel constituting the rectangular block MP are (x, y), and the gray value of the pixel is P _xy 0 ≦ x, y ≦ 7, the transform coefficient position in the transform coefficient matrix MS is (u, v), and the transform coefficient is S _uv Assuming that 0 ≦ u, v ≦ 7, the orthogonal transformation equation is as follows.
[0049]
(Equation 1)

[0050]
FIG. 5 shows the conversion coefficient S _uv FIG. 6 is a diagram showing a base image corresponding to FIG. The shading change of the base image becomes a COS waveform. For convenience of illustration, a portion where the value of COS is positive is represented by white and a portion where the value of COS is negative is represented by binary as black. Each of the base images shown in FIG. 5 corresponds to the image at the position (u, v) of the 64 orthogonal base image matrices, and the transform coefficient S _uv Is large, it means that the component of the base image at the (u, v) position is strongly included.
[0051]
As can be seen from FIG. 5, the base image contains more high-frequency horizontal frequency components (vertical stripes) as going from left to right, and contains more high-frequency vertical frequency components (horizontal stripes) as going from top to bottom. A rectangular block MP of × 8 pixels is represented by a linear combination of the 64 base images. Therefore, S in the transform coefficient matrix MS _uv By reading the value distribution state, it is possible to easily grasp which frequency components are strongly contained and which feature amount (direction, period, amplitude) of the stripes is in which direction. Will be possible.
[0052]
However, since high-frequency regions generally contain white noise, and it is meaningless to treat high-frequency regions from the viewpoint of image resolution, it is more effective to cut off high-frequency components and focus on relatively low-frequency regions. . Therefore, 64 transform coefficients S _uv Among them, twelve transform coefficients E = ｕ (u, v) = (0,2), (0,3), (0,4), (1,1) corresponding to the base image surrounded by the broken line in FIG. 2), (1,3), (1,1), (2,2), (2,1), (3,1), (2,0), (3,0), (4,0) Only｝ is selected as a stripe evaluation target. It goes without saying that other combinations of conversion coefficients may be used.
[0053]
FIG. 6 is a table showing the selected twelve transform coefficients, which are classified by the stripe direction and the stripe period. The horizontal group (Y_E) has a conversion coefficient S ₀₂ , S ₀₃ And S ₀₄ And a horizontal stripe as seen from the base image shown in FIG. The conversion coefficient S is set for the horizontal and oblique group (YN_E). ₁₂ And S _Thirteen And, as can be seen from the base image shown in FIG. The conversion coefficient S is added to the diagonal group (N_E). ₁₁ And S ₂₂ And oblique stripes as can be seen from the base image shown in FIG. The conversion coefficient S is assigned to the vertical / diagonal group (TN_E). ₂₁ And S ₃₁ And, as can be seen from the base image shown in FIG. Further, the vertical group (T_E) has a conversion coefficient S ₂₀ , S ₃₀ And S ₄₀ And vertical stripes as can be seen from the base image shown in FIG.
[0054]
When the fringe period to be detected is larger than the fringe period shown in FIG. 6, the size of the rectangular block MP is set to be twice or more, or the size of the captured image MI is reduced to 1 / with the size of the rectangular block MP unchanged. What is necessary is just to compress to 2 or more.
[0055]
The DCT processing unit 22 performs the above-described DCT processing in units of rectangular blocks on all 64 × 64 rectangular blocks, that is, all elements of the rectangular block matrix MB, thereby obtaining the 12 types of transform coefficients shown in FIG. For each, a 64 × 64 two-dimensional matrix MS _uv It is possible to obtain the spatial distribution of the fringe feature amount expressed by the following expression over the entire captured image. MS _uv Can be expressed by the following equation.
[0056]
(Equation 2)

[0057]
Returning to the description of the flowchart shown in FIG. Next, the average / standard deviation calculation unit 23 calculates the two-dimensional matrix MS _uv The average value and the standard deviation of the 64 × 64 values are calculated (S3). When no fringes are included in the captured image, the two-dimensional matrix MS _uv Is a normal distribution in which the frequencies are concentrated around 0. In addition, when stripes occur in a part of a minute area in a captured image, the two-dimensional matrix MS _uv Although the histogram of the values within {circumflex over ()} is globally distributed, the transform coefficient S _uv Is a value apart from 0, and a peak corresponding to the fringe appears at a position away from the normal distribution.
[0058]
In order to use this statistical principle, the average value / standard deviation calculation unit 23 calculates the average value μ of the normal distribution for each of the 12 types of conversion coefficients shown in FIG. _uv ; (U, v) ∈E and standard deviation σ _uv Calculating (u, v) ∈E;
[0059]
Then, the stripe candidate block extraction unit 24 calculates the standard deviation σ _uv Is a predetermined constant C _uv ; (U, v) ∈E is set as the threshold value, and the conversion coefficient S _uv And mean μ _uv Then, a set of rectangular blocks whose distance between them is equal to or larger than a threshold is extracted as a stripe candidate block group (S4). FIG. 7 shows an example of extraction of a fringe candidate block group. _uv And mean μ _uv Are extracted as a stripe candidate block group. This stripe candidate block group GB _uv Is represented by the following equation.
[0060]
(Equation 3)

[0061]
Note that, as a special case, the fringes may occur in the entire or most of the captured image, and thus the calculated standard deviation σ _uv It is desirable to use this value in combination with a method of directly managing the value and evaluating the entire captured image instead of using the value as it is.
[0062]
Further, the fringe candidate block extraction unit 24 sets the standard deviation σ as a threshold. _uv To C _uv By using the multiplied value, a portion where the conversion coefficient value is relatively apart from 0, that is, a portion where the stripe contrast (the amplitude of the stripe) is relatively high is extracted from the entire captured image. This processing eliminates the influence of an absolute change in the fringe contrast due to a change in the illumination intensity at the time of imaging and a change in the light reflectance of the surface of the inspection object, and enables a highly robust inspection.
[0063]
Returning to the description of the flowchart shown in FIG. Next, the direction-specific filter processing unit 25 calculates the twelve types of transform coefficients S extracted by the stripe candidate block extraction unit 24. _uv Candidate block group GB corresponding to _uv , A filter process is performed in consideration of the two-dimensional shape characteristics of the stripes, and a region where a stripe having a predetermined feature amount is generated is detected (S5).
[0064]
Since the two-dimensional shape characteristics of the stripe, particularly the change in the direction of the stripe and the period of the stripe, are two-dimensionally smooth, the change in the minute area is negligible. The direction-specific filter processing unit 25 performs filter processing on each of the five direction-specific groups illustrated in FIG. 6 using an operator that takes into account the two-dimensional shape characteristics of the stripes. FIG. 8 shows an example of the shape of the operator, which is a horizontal group mask (Y_MSK) shown in (a), a horizontal diagonal group mask (YN_MSK) shown in (b), and a diagonal group mask shown in (c). (N_MSK), a vertical / diagonal group mask (TN_MSK) shown in (d), and a vertical group mask (T_MSK) shown in (e). Note that the operator may have a shape other than those shown in FIGS.
[0065]
For example, when filter processing is performed on transform coefficients included in the horizontal group Y_E = {(u, v) = (0, 2), (0, 3), (0, 4)}, three transform coefficients S ₀₂ , S ₀₃ And S ₀₄ Three types of stripe candidate block groups GB corresponding to ₀₂ , GB ₀₃ And GB ₀₄ Can be represented by the following equation.
[0066]
(Equation 4)

[0067]
The extraction of the horizontal stripe candidate block group Y_GB is performed by sequentially scanning the 64 × 64 rectangular block matrix MB, and the above three types of the stripe candidate block group GB are extracted. ₀₂ , GB ₀₃ And GB ₀₄ Among them, it is possible to search for a rectangular block in which at least one fringe candidate block exists. Then, the direction-specific filter processing unit 25 removes, from the horizontal stripe candidate block group Y_GB, only blocks in which horizontal stripe candidate blocks are continuously present in the form of a horizontal group mask Y_MSK shown in FIG. By leaving it as Y_WB, an isolated fringe candidate block is removed. The operation of extracting only the continuous horizontal stripe candidate blocks is performed by an operator f _[Y _ _MSK] , The horizontal stripe block group Y_WB can be represented by the following equation.
[0068]
(Equation 5)

[0069]
FIG. 9 is a diagram schematically showing the extraction of the horizontal stripe block. The stripe candidate block group GB shown in FIGS. 9A to 9C. ₀₂ , GB ₀₃ And GB ₀₄ By searching for a block in which at least one stripe candidate block exists, a horizontal stripe candidate block group Y_GB shown in FIG. 9D can be extracted. Then, the direction-specific filter processing unit 25 performs a filtering process on the horizontal stripe candidate block group Y_GB shown in FIG. 9D using the horizontal group mask Y_MSK, and thereby the horizontal stripe block group shown in FIG. Extract Y_WB.
[0070]
The direction-specific filter processing unit 25 performs the same filter processing on the horizontal and oblique group YN_E, the oblique group N_E, the vertical and oblique group TN_E, and the vertical group T_E. By obtaining the logical sum of the extracted horizontal and oblique stripe block group YN_WB, the oblique stripe block group N_WB, the vertical and oblique stripe block group TN_WB, the vertical stripe block group T_WB, and the above-described horizontal stripe block group Y_WB, the stripe blocks in all directions are obtained. The omnidirectional fringe block group WB obtained by superimposing the groups is output as a fringe detection area as a final result. This omnidirectional fringe block group WB can be represented by the following equation.
[0071]
(Equation 6)

[0072]
The fringe detecting device described above can be realized by causing a computer to execute a fringe detecting program. A method for realizing a fringe detecting device by this computer will be described below, but the fringe detecting device in the present embodiment is not limited to this. For example, the DCT processing that requires the longest time in the program processing can be implemented by hardware to speed up the processing.
[0073]
FIG. 10 is a diagram showing an example of the appearance of the fringe detection device of the present invention. The fringe detecting device includes a computer main body 1, a graphic display device 2, a magnetic tape device 3 on which a magnetic tape 4 is mounted, a keyboard 5, a mouse 6, and a CD-ROM on which a CD-ROM (Compact Disc-Read Only Memory) 8 is mounted. It includes a ROM device 7 and a communication modem 9. The fringe detection program is supplied by a storage medium such as the magnetic tape 4 or the CD-ROM 8 and executed by the computer main body 1. Further, the fringe detection program may be supplied to the computer main body 1 from another computer via a communication line and a communication modem 9.
[0074]
FIG. 11 is a block diagram illustrating a configuration example of a fringe detection device according to the present embodiment. The computer main body 1 shown in FIG. 10 includes a CPU 10, a ROM (Read Only Memory) 11, a RAM (Random Access Memory) 12, and a hard disk 13. The CPU 10 performs processing while inputting and outputting data to and from the graphic display device 2, the magnetic tape device 3, the keyboard 5, the mouse 6, the CD-ROM device 7, the communication modem 9, the ROM 11, the RAM 12, and the hard disk 13. The stripe detection program recorded on the magnetic tape 4 or the CD-ROM 8 is temporarily stored on the hard disk 13 by the CPU 10 via the magnetic tape device 3 or the CD-ROM device 7. The CPU 10 detects a fringe by appropriately loading a fringe detection program from the hard disk 13 into the RAM 12 and executing the program.
[0075]
As described above, according to the fringe detecting device of the present embodiment, the captured image is subdivided into rectangular blocks, and the rectangular blocks are subjected to DCT processing to perform spatial frequency analysis of the fringes. The two-dimensional feature analysis including the direction of is possible.
[0076]
In addition, since the feature amount (direction, period, and amplitude of the stripe) of the stripe is extracted for each rectangular block, it is possible to detect even a stripe generated in a minute area of about the size of the rectangular block. became.
[0077]
In addition, since a rectangular block having a relatively large image amplitude value in a two-dimensional matrix is extracted as a fringe candidate block, a change in illumination intensity at the time of imaging and a change in light reflectance on the surface of an inspection object are performed. This eliminates the influence of the change in the fringe contrast, and enables detection with high robustness.
[0078]
Further, since the filter processing is performed in consideration of the two-dimensional shape characteristics of the stripes, it is possible to remove the influence of brightness change due to white noise, uneven reflection on the surface of the inspection object, or texture. .
[0079]
(Embodiment 2)
The photoconductor drum inspection device according to the second embodiment is obtained by applying the stripe detection device according to the first embodiment to a thin film on the surface of the photoconductor drum used in a copying machine or the like.
[0080]
FIG. 12 shows a cross-sectional structure near the surface of the photosensitive drum. The photoconductor drum 40 includes an aluminum tube 41 and a multilayer thin film 42 that is a photoconductor formed on the surface of the aluminum tube 41. When the surface of the photosensitive drum 40 is irradiated with a light source 43 having a single wavelength of λ, light interference depending on the thickness d of the multilayer thin film 42 occurs. Therefore, when there is a change in the thickness d, interference fringes such as contour lines of a map are observed. If there is a defective portion where the thickness of the multilayer thin film 42 changes abruptly, in a copying machine, shading unevenness corresponding to the defective portion occurs on the copy paper. In this defective portion, the interval between the interference fringes is narrowed, so that it is possible to detect a defective portion of the photosensitive drum 40 by detecting a portion where the fringe cycle of the interference fringes is short.
[0081]
FIG. 13 is a block diagram illustrating a schematic configuration of the photoconductor drum inspection apparatus according to the present embodiment. The photoconductor drum inspection apparatus includes a light source 43 having a single wavelength, a line sensor 44 for imaging regular reflection light from the photoconductor drum 40, a rotation drive unit 45 for rotating the photoconductor drum 40, and a rotation drive unit 45. A rotation control unit 46 for controlling the rotation of the image sensor, an image memory 47 for storing an image captured by the line sensor 44, and a fringe detection processing unit 48 for performing fringe detection.
[0082]
The light source 43 is arranged so that light is irradiated on the entire area of the photosensitive drum 40 in the longitudinal direction. Further, the line sensor 44 is arranged so as to capture an image of the entire photosensitive drum 40 in the longitudinal direction. The rotation control unit 46 controls the rotation of the rotation drive unit 45 in synchronization with the imaging by the line sensor 44. For example, if the storage capacity of the image memory 47 is 2048 pixels × 2048 pixels, the rotation control unit 46 rotates the image (one frame) of the entire surface of the photosensitive drum 40 so that the image (one frame) fits in the storage capacity of the image memory 47. The controller 45 is controlled to rotate the photosensitive drum 40.
[0083]
The fringe detection processing unit 48 performs fringe detection processing on one frame of the image stored in the image memory 47. Since the processing procedure is the same as that of the fringe detection device in the first embodiment, a detailed description thereof will be omitted. Do not repeat.
[0084]
As described above, according to the photoconductor drum inspection apparatus of the present embodiment, it is possible to automatically detect a defective portion on the surface of the photoconductor drum, and this has been described in the fringe detection apparatus of the first embodiment. The effect can be obtained even in the inspection of the photosensitive drum.
[0085]
(Embodiment 3)
The liquid crystal panel inspection device according to the third embodiment is obtained by applying the fringe detection device according to the first embodiment to inspection for gap unevenness between bonded glasses in a liquid crystal panel.
[0086]
FIG. 14 shows a sectional structure of the liquid crystal panel 49. The liquid crystal panel 49 includes two glasses 50 and a liquid crystal 51 injected into a gap between the two glasses 50. When the surface of the liquid crystal panel 49 is irradiated with the light source 53 having a single wavelength of λ, optical interference occurs depending on the value d of the gap between the glasses. Therefore, when there is a change in the value d of the gap between glasses, interference fringes such as contour lines of a map are observed.
[0087]
In the manufacturing process of the liquid crystal panel, since the liquid crystal 51 is injected into the gap between the glasses, the unevenness of the gap between the glasses affects the product quality. Therefore, it is possible to detect gap unevenness by extracting a portion where interference fringes occur, and to find out a defect of the liquid crystal panel 49.
[0088]
FIG. 15 is a block diagram illustrating a schematic configuration of a liquid crystal panel inspection device according to the present embodiment. This liquid crystal panel inspection apparatus includes a light source 53 having a single wavelength, a line sensor 52 for capturing an interference fringe generated on the liquid crystal panel 49, an image memory 47 for storing an image captured by the line sensor 52, and a fringe detection. And a fringe detection processing unit 48.
[0089]
The light source 53 is arranged so that light is diffused and radiated to the entire surface of the liquid crystal panel 49. Further, the line sensor 52 is arranged so as to capture an image of the entire area of the liquid crystal panel 49. For example, if the storage capacity of the image memory 47 is 512 pixels × 512 pixels, an image is taken so that an image (one frame) of the entire surface of the liquid crystal panel 49 fits in the storage capacity of the image memory 47.
[0090]
The fringe detection processing unit 48 performs fringe detection processing on one frame of the image stored in the image memory 47. Since the processing procedure is the same as that of the fringe detection device in the first embodiment, a detailed description thereof will be omitted. Do not repeat.
[0091]
As described above, according to the liquid crystal panel inspection device of the present embodiment, it is possible to automatically detect a defective portion on the surface of a liquid crystal panel, and the effect described in the fringe detection device of the first embodiment is obtained. It has become possible to obtain even in the inspection of the liquid crystal panel.
[0092]
The embodiments disclosed this time are to be considered in all respects as illustrative and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.
[Brief description of the drawings]
FIG. 1 is a block diagram illustrating a schematic configuration of a fringe detection device according to a first embodiment of the present invention.
FIG. 2 is a flowchart for explaining a processing procedure of the fringe detecting device according to the first embodiment of the present invention.
FIG. 3 is a diagram schematically showing optical fringes appearing on the surface of an inspection object.
FIG. 4 is a diagram illustrating a state in which a captured image is subdivided into rectangular blocks.
FIG. 5 shows a conversion coefficient S _uv FIG. 6 is a diagram showing a base image corresponding to FIG.
FIG. 6 is a table showing twelve selected transform coefficients.
FIG. 7 is a diagram illustrating an example of extraction of a stripe candidate block group.
FIG. 8 is a diagram illustrating an example of the shape of an operator used for the filtering process.
FIG. 9 is a diagram for explaining extraction of a horizontal stripe block group.
FIG. 10 is a diagram illustrating an example of an external appearance of a computer that realizes fringe detection according to the first embodiment.
11 is a diagram illustrating a configuration example of a computer illustrated in FIG.
FIG. 12 is a diagram illustrating a cross-sectional structure near a surface of a photosensitive drum.
FIG. 13 is a diagram illustrating a schematic configuration of a photoconductor drum inspection apparatus.
FIG. 14 is a diagram showing a cross-sectional structure of a liquid crystal panel.
FIG. 15 is a diagram showing a schematic configuration of a liquid crystal panel inspection device.
FIG. 16 is a diagram for explaining a problem in the conventional measurement of a fringe period.
[Explanation of symbols]
1 Computer body
2 Graphic display device
3 Magnetic tape device
4 Magnetic tape
5 Keyboard
6 mice
7 CD-ROM device
8 CD-ROM
9 Communication modem
10 CPU
11 ROM
12 RAM
13 Hard Disk
21 Rectangular block division unit
22 DCT processing unit
23 Average / standard deviation calculator
24 Stripe candidate block extraction unit
25 Filtering unit for each direction
26 Processing result merger
40 Photoconductor drum
41 Aluminum tube
42 Multilayer Thin Film
43,53 light source
44, 52 Line sensor
45 rotation drive
46 Rotation control unit
47 Image memory
48 fringe detection processing unit
49 LCD panel
50 glass
51 LCD

Claims (5)

Dividing means for dividing the captured image of the optical fringe of the inspection object into a plurality of directions that do not depend on the direction of the optical fringe, and dividing the image into blocks composed of a plurality of pixel sets,
Processing means for performing a discrete cosine transform on the information in the block divided by the dividing means,
A fringe detecting device comprising: an evaluating unit for performing an evaluation including a direction of an optical fringe of the inspection object based on a processing result by the processing unit. A light emitting unit for irradiating the photosensitive drum with light,
Imaging means for imaging optical fringes of the photosensitive drum generated by reflected light of the light from the light emitting means,
Control means for controlling the rotation of the photosensitive drum,
Division means for dividing the image taken by the imaging means into a plurality of directions independent of the direction of the optical fringe, and dividing the image into blocks formed of a plurality of pixel sets,
Processing means for performing a discrete cosine transform on the information in the block divided by the dividing means,
An evaluation unit for performing an evaluation including a direction of an optical stripe on the photosensitive drum based on a processing result by the processing unit. A light emitting means for irradiating the liquid crystal panel with light,
Imaging means for imaging optical fringes of the liquid crystal panel generated by reflected light of the light from the light emitting means,
Division means for dividing the image taken by the imaging means into a plurality of directions independent of the direction of the optical fringe, and dividing the image into blocks formed of a plurality of pixel sets,
Processing means for performing a discrete cosine transform on the information in the block divided by the dividing means,
A liquid crystal panel inspection apparatus comprising: an evaluation unit for performing an evaluation including a direction of an optical fringe of the liquid crystal panel based on a processing result by the processing unit. A step of dividing the captured image of the optical fringe of the inspection object into a plurality of directions independent of the direction of the optical fringe, and dividing the image into blocks formed of a plurality of pixel sets,
Performing a discrete cosine transform on the information in the divided blocks;
Performing an evaluation including a direction of an optical fringe of the inspection object based on a processing result of the discrete cosine transform. A step of dividing the captured image of the optical fringe of the inspection object into a plurality of directions independent of the direction of the optical fringe, and dividing the image into blocks formed of a plurality of pixel sets,
Performing a discrete cosine transform on the information in the divided blocks;
Performing an evaluation including a direction of an optical fringe of the inspection object based on a processing result of the discrete cosine transform.

Images