JP3605579B2 - Dust detector, dust detection system, particle detector and dust detection method for clean room - Google Patents

Description

(ただし、ｐｉ，ｐｏ，αおよびＩ(ｘ，ｙ)は以下の通り定義される。
ｐｉ： 画像データのうち、(ｘ，ｙ)で示される位置を基準として線対称をなす第１の画素群を構成する画素の輝度値
ｐｏ： 像データのうち、(ｘ，ｙ)で示される位置を基準として線対称をなすとともに前記第１の画素群よりも外側に位置する第２の画素群を構成する画素の輝度値
Ｉ(ｘ，ｙ)： (ｘ，ｙ)で示される位置の画素の輝度値
ａ，ｂ，α： ０および正の数)
演算手段によって得られる輝度情報に基づいて画像のコントラストを検出するコントラスト検出手段と、コントラストの大小関係に基づいて測定面に存在する微粒子を識別する識別手段と、識別手段での識別結果を出力する結果出力手段と、を有することを特徴とする。
【００１１】
ここでは、Ｐ(ｘ，ｙ)を算出するが、ｐｉ： 前記画像データのうち、(ｘ，ｙ)で示される位置を基準として線対称をなす第１の画素群を構成する画素の輝度値、ｐｏ： 前記画像データのうち、(ｘ，ｙ)で示される位置を基準として線対称をなすとともに前記第１の画素群よりも外側に位置する第２の画素群を構成する画素の輝度値である。したがって、Ｐ(ｘ，ｙ)は前述した２重周囲差分値であり、輝度情報を得ることができる。
また、第１の画素群を(ｘ，ｙ)で示される位置に存在する画素から画素ｎ個分(ｎは１以上の整数)だけ離れた位置に存在する画素によって構成し、かつ第２の画素群を(ｘ，ｙ)で示される位置に存在する画素から画素ｍ個分(ｍ＞ｎの整数)だけ離れた位置に存在する画素によって構成することができる。そして、ｎおよびｍを適宜設定することにより、種々の形態の輝度ムラを抽出することを可能にする。
さらに、係数ａ，ｂはそれぞれ、内側差分値および外側差分値を加算するときの重み付けを定義する。したがって、ａおよびｂはａ＋ｂ＝１.０を満足し、かつａおよびｂは評価する画像の特徴に応じて決定されることになる。ここで、ａまたはｂが０(ゼロ)となることを許容している。
加えて、前記αは、前記ｍの値に応じて決定することができる。例えば、ｍの値が大きくなれば、ｐｏの総和に対してＩ(ｘ，ｙ)は無視することもできる。その場合には、αを０(ゼロ)として演算処理することもできる。
【００１２】
または、微粒子を測定する測定器として、微粒子を載せることができる平坦な測定面と当該測定面を撮像する撮像部とを一つの筐体内に収容した測定面撮像手段と、撮像された画像に関する画像データを分割する画像処理手段と、分割された画像データについて、下記式（２）で示す演算処理を実行する演算手段と、
【数８】

(ただし、ｐｉ： 画像データのうち、(ｘ，ｙ)で示される位置に存在する画素を取り囲む画素群を構成する画素の輝度値、Ｉ(ｘ，ｙ)： (ｘ，ｙ)で示される位置の画素の輝度値)
演算手段によって得られる輝度情報に基づいて撮像した画像のコントラストを検出するコントラスト検出手段と、撮像された画像のコントラストの大小関係に基づき、測定面に存在する微粒子を識別する識別手段と、識別手段での識別結果を出力する結果出力手段と、を有するものであってもよい。
【００１３】
更に本発明は塵埃検出方法として捉えることもできる。
塵埃検出方法は、塵埃が載った平坦部の画像を撮像するステップと、撮像された画像の画像データを複数の画素に分割する分割ステップと、複数の前記画素の輝度値の分布に基づいて注目画素を設定する注目画素設定ステップと、注目画素を取り囲む画素群を構成する各画素の輝度値の総和と注目画素の輝度値の差分を算出する差分算出ステップと、差分に基づき、注目画素が塵埃を現すものであるか否かを判定する判定ステップと、判定ステップにおいて塵埃と判定されたものを集計する集計ステップと、を有することを特徴とする。
この塵埃検出方法が光を透過させたときの注目画素と画素群の輝度差を検出する輝度差検出ステップを有する場合、判定ステップにおいて差分が所定の閾値以上であるときに注目画素が塵埃を現すものであると判定することができる。
【００１４】
また本発明は、クリーンルーム内の塵埃検出方法として、クリーンルーム内において落下した塵埃を平坦部で所定時間受け止めるステップと、塵埃を受け止めた平坦部の透過像を撮像するステップと、撮像した透過像の画像データを記憶するステップと、画像データを解析して塵埃を検出するステップと、を有することを特徴とするものである。この場合、塵埃を平坦部にて受け止める前において、平坦部の初期透過像を撮像するステップをさらに有し、塵埃を検出するステップでは、初期透過像と透過像とを比較すると、所定時間内に受け止めた塵埃をより正確に検出することができる。
なお、画像データをシェーディング処理すると画像中の塵埃がより明確に浮き出るため、検出が容易になる。また塵埃を検出するステップでは、塵埃が平坦部に形成する影と、平坦部における輝度ムラとを識別し、塵埃の影のデータを抽出することにより塵埃を検出することができる。
【００１５】
【発明の実施の形態】
以下、添付図面に示す実施の形態に基づいてこの発明を詳細に説明する。
図１は、第１の実施の形態における塵埃検出システムの構成を説明するための図である。図１に示すように、塵埃を測定するための塵埃検出システム（微粒子を測定する測定器）１は、塵埃検出器（塵埃検出装置）１０と、ケーブル１８を介して塵埃検出器１０に接続された解析コンピュータ３０とから構成されている。なお、図示していないが、解析コンピュータ３０には、複数の塵埃検出器１０が接続されている。
【００１６】
塵埃検出器１０は、自由落下する塵埃が載る集塵板（光透過部、測定部、塵埃受け手段、平坦部）１１と、集塵板１１の下に重ねられた拡散板（光拡散部、光拡散手段）１２と、集塵板１１と拡散板１２を透過した光を反射するミラー（偏向部）１３と、レンズ１４と、画像を撮像するＣＣＤカメラ（撮像部、撮像手段）１５とを備えている。
【００１７】
集塵板１１は光を透過するもので構成される。本実施の形態では、この集塵板１１の表面１１ｓ上に載った塵埃を測定する。集塵板１１では塵埃の画像を撮像しやすいように表面１１ｓは平坦となっている。また数回にわたって塵埃の測定を行う場合において、測定前に表面１１ｓ上に載っていた塵埃を除去すること等を考慮し、集塵板１１は強度の高いものを使用することが好ましい。例えばサファイアガラスやクリスタルガラス等を挙げることができる。
【００１８】
拡散板１２は、集塵板１１を透過した光を拡散するものであり、光を拡散できるものであれば特に限定されない。具体例として厚みが２〜３ｍｍ程度のアクリル板等を挙げることができる。
ミラー１３は、拡散板１２を透過した光をＣＣＤカメラ１５の方向へ反射させる。なお、ミラー１３の代わりにレンズや偏向板など、入射光を所定方向へ偏向することができるものを適用することができる。
【００１９】
レンズ１４は、ミラー１３によって反射された光をＣＣＤカメラ１５に結像する。レンズ１４としては特に限定されないが、被写界深度を集塵板１１から拡散板１２に定めることができるものであることが好ましく、更に好ましくは焦点を集塵板１１の表面１１ｓに合わせることができるものである。具体例として光の色による収差を補正することができるガウス型レンズやテッサ型レンズ等の複数のレンズから構成される対象型レンズを挙げることができる。
【００２０】
ＣＣＤカメラ１５は、白黒タイプ、カラータイプのいずれであっても良いが、本実施の形態においては、現状の撮像能力からいって有利な白黒タイプのものを採用する。なお、本実施の形態において、ＣＣＤカメラ１５では、集塵板１１の全域を撮像する。ＣＣＤカメラ１５で撮像された画像は画像信号に変換され、図示しないインターフェース（送出手段）からケーブル１８を介して解析コンピュータ３０へ送出される。
【００２１】
解析コンピュータ３０は、ＣＣＤカメラ１５で撮像した画像に対し、後に詳述する所定の処理を施す画像処理部(コントラスト把握手段、画像処理手段、演算手段)３１、処理を施した画像に基づき、撮像した画像内にどのくらい塵埃が存在しているかを判定部(塵埃検出手段、ムラ検出手段)３２、塵埃と判定されたものを集計する集計部（集計手段、分別手段）３３、ＣＣＤカメラ１５における撮像を制御する撮像制御部３４、判定部３２での判定結果や画像処理部３１で取り込んだ画像、処理後の画像等を適宜表示するモニタ(集計結果出力手段、表示手段)３５、を備えている。
【００２２】
上記のような構成を有する塵埃検出システム１を用いる場合、塵埃を検出する場所（以下、検出場所）に設置して集塵板１１の表面１１ｓの画像をＣＣＤカメラ１５で撮像し、撮像した画像データを解析コンピュータ３０で処理することにより、表面１１ｓ上に載った塵埃の測定を行う。
具体的には、まず塵埃検出器１０を塵埃を検出する場所（以下、検出場所）に、集塵板１１の表面１１ｓが上向きになるように、また表面１１ｓが水平となるようにして設置する。そして、表面１１ｓ上に載った塵埃を画像としてＣＤＤカメラ１５で撮像し、撮像した画像データを画像処理部３１で処理することにより、載った塵埃の測定を行う。ここで、塵埃は表面１１ｓ上に自然に落下したものが測定される。なお、発生した塵埃が落下するのはその塵埃の発生場所のみに限られず、例えばクリーンルーム内の気流等の影響を受けて塵埃の落下箇所は変動する。検出場所は適宜設定できるが、例えば作業者の傍や作業台の上等、作業の対象となる製品の付近に設置される。また、解析コンピュータ３０の設定場所は特に限定されず、塵埃検出器１０の近傍であっても、離れた場所であってもよい。なお、複数の塵埃検出器１０は互いに異なる検出場所に設置される。
【００２３】
画像の取り込みにおいては、まず光源からの光が集塵板１１を透過する。このとき表面１１ｓ上に載った塵埃が存在する箇所において光は透過しないので、得られる表面１１ｓの画像においては塵埃が影となって現れる。光源は外光や検出場所であるクリーンルームに設置された蛍光灯等の通常の電燈を使用するが、塵埃検出器１０の近傍に撮像用光源を設けてもよい。ただしこの場合、クリーンルームの空気の流れを変えないような位置に設置することが好ましい。
【００２４】
集塵板１１を透過した光は、拡散板１２によって拡散される。光源の種類や状態によっては、集塵板１１に入射光の輝度が場所によって異なる「輝度ムラ」を生じる。輝度ムラは塵埃の検出の妨げとなるので、入射光を拡散板１２によって拡散することで、その輝度ムラを減少させる。このとき、塵埃が存在する箇所と塵埃が存在しない箇所における輝度の差は、輝度ムラと比較してコントラストが大きいため、拡散板によって輝度の変化をさせても、塵埃の検出にはさほど影響はない。
【００２５】
拡散板１２を透過した光は、ミラー１３によって横方向へ偏向される。更に、ミラー１３によって偏向された光は、レンズ１４によって集光されてＣＣＤカメラ１５のイメージセンサ部分に捉えられる。このとき、レンズ１４の焦点を調整することにより、結像画像の大きさを調節できるので、ＣＣＤカメラ１５のイメージセンサ部分は測定面積である表面１１ｓの面積と同じ面積を有しなくてもよい。したがって、本実施の形態では、表面１１ｓにおける測定面積をＣＣＤカメラ１５のイメージセンサ部分より大きなものとすることができる。
【００２６】
さらに塵埃検出器１０にはミラー１３が設けられており、ミラー１３により入射光を偏向させることによりＣＣＤカメラ１５を集塵板１１の直下に設けなくてもよい構成となっている。したがって、塵埃検出器１０の高さ寸法が比較的小さくでき、測定面である表面１１ｓを作業対象となる製品のより近辺に位置させることができる。
【００２７】
このようにしてＣＣＤカメラ１５により撮像した画像のデータは、ケーブル１８を介して解析コンピュータ３０へ送出され、その画像のデータは画像処理部３１内の所定のメモリに格納する。塵埃の検出においては、得られた表面１１ｓの画像を解析し、画像中の輝度ムラを排除して表面１１ｓに載った塵埃のみを検出する。この原理について以下に説明する。
【００２８】
図２は、塵埃の検出を行う際に実行する画像処理の流れを示す図である。
まず、画像処理部３１においては、図２に示すような手順にて、画像を処理し、検出を実行する。なお、以下の処理において、周知の画像処理手法については詳細な説明を省略する。本実施の形態による検出方法は、被検査画像の取り込み(ステップＳ１０１)、取り込み画像データの分割処理(ステップＳ１０３)、平滑化処理(ステップＳ１０５)、２重周囲差分法の適用(ステップＳ１０７)、特異値データ抽出処理(ステップＳ１０９)、画像膨張・収縮処理(ステップＳ１１１)という一連のステップによって行なわれる。以下、各ステップの内容をステップ順に説明する。
【００２９】
ステップＳ１０１：画像取り込み
まず、表面１１ｓの画像をＣＣＤカメラ１５で撮像する。撮像された画像の画像データは、ケーブル１８を介して解析コンピュータ３０へ送出される。図３に撮像された表面１１ｓの画像を示す。図３に示すように、この画像には、所々小さな白っぽい点のようなものが微かに観察される。これらは主として表面１１ｓ上に載った塵埃である。
そして、送出された画像データは画像処理部３１においてシェーディング補正（シェーディング処理）が施される。このシェーディング補正により、主にＣＣＤカメラ１５における周辺減光、撮像素子の不均一性によって生じる、測定画像の本来の輝度と画像データとの間の変換特性の不整合を除去する。このとき、輝度を変化させることにより、輝度の高いものは赤色に、逆に輝度の低いものは緑色となるように補正される。このように処理して得られる画像を図４に示す。図４では、図３に示すものより塵埃が明瞭に現れている。その一方で、画像全体に輝度ムラが図３に示すものより顕著に現れている。
【００３０】
ステップＳ１０３：取り込み画像データの分割
ステップＳ１０１により得られた画像データをマトリックス状に分割する。この画像データの分割は、後述のステップＳ１０７における２重周囲差分法を実行するための予備的な処理である。ここでは、ｎｘｍの画素に分割されるものとする。分割された画素は、２次元座標(ｘ，ｙ)でその位置を特定することができる。なお、本実施の形態では、撮像手段としてＣＣＤカメラ１５を用いているから、ｎｘｍはＣＣＤカメラ１５の画素数(解像度)に対応する。もっとも、分割の数は最大でＣＣＤカメラ１５の画素数(解像度)であるが、これ以下の分割数とすることを妨げるものではない。
【００３１】
ステップＳ１０５：平滑化処理
ステップＳ１０３により分割された画像データについて、ノイズを除去するための平滑化処理を行う。例えば、２ｘ２のフィルタリングを行う。
【００３２】
ステップＳ１０７：２重周囲差分法の適用
ステップＳ１０５においてノイズが除去された画像データについて、各画素の輝度値を求めることを前提に、以下を内容とする２重周囲差分法を実行する。
図５に、２重周囲差分法の処理手順を示すフローチャートを示す。ステップＳ１０７では、図５に示すように、まず下記式(３)で示される内側差分値を算出する(ステップＳ２０１)。次いで、下記式(４)で示される外側差分値を算出する(ステップＳ２０３)。次いで、ステップＳ２０１で得られた内側差分値およびステップＳ２０３で得られた外側差分値とを加えることにより下記式(５)で示される２重周囲差分値を算出する(ステップＳ２０５)。以下、ステップＳ２０１〜ステップＳ２０３の具体的内容を説明する。なお、内側差分値を算出するステップＳ２０１と外側差分値を算出するステップＳ２０３の順番は入れ替えても良い。この２重周囲差分法は、本発明者が新たに開発したアルゴリズムである。また、式(３)〜式(５)は、本発明者等が種々検討を重ねた結果としての経験式であり、式(３)の２ｎ＋１(第３項)、式(４)のα(第２項)および２ｍ＋１(第３項)、式(５)の係数ａ，ｂは経験的に導き出されるものである。
【００３３】
【数９】

(ただし、ｐｉ： 前記画像データのうち、(ｘ，ｙ)で示される位置に存在する画素を取り囲む第１の画素群を構成する画素の輝度値、Ｉ(ｘ，ｙ)： (ｘ，ｙ)で示される位置の画素の輝度値、ｎ： 注目画像から第１の画素群までの画素の数)
【００３４】
【数１０】

(ただし、ｐｏ： 前記画像データのうち、(ｘ，ｙ)で示される位置に存在する画素を取り囲む第２の画素群を構成する画素の輝度値、Ｉ(ｘ，ｙ)： (ｘ，ｙ)で示される位置の画素の輝度値、ｍ： 注目画像から第２の画素群までの画素の数，α： ０以上の正の数)
【００３５】
【数１１】

(ただし、ａ，ｂ： ａ＋ｂ＝１.０を満足する正の数)
【００３６】
ステップＳ２０１：内側差分値の算出
内側差分値は前述のように式(３)に基づく演算を行う。図６を参照しつつ式(３)の内容を説明する。なお、図６は、ステップＳ１０３により分割された画像データを示している。
式(３)のうち、第１項は、注目画素５０からｎ個だけ離れた位置に存在する画素の輝度の総和を意味している。図６において、黒塗りされた画素が注目画素５０であり、またｎ個だけ離れた位置に存在する複数の画素からなる第１の画素群(画素群)５１ａは濃い灰色に塗りつぶされている。なお、図６はｎ＝２の例を示している。次に、式(３)の第２項は、注目画素５０の輝度値を示している。さらに、式(３)の第３項は、注目画素５０からｎ個だけ離れた位置に存在する画素列に含まれる画素の数である。
なお、注目画素５０の位置によっては、ｎ個だけ離れた位置に画素が存在しない場合もあり得る。その場合には、ミラーリングによってｎ個だけ離れた位置に存在する画素の輝度値を用いて演算を行なえばよい。ステップＳ２０３における外側差分値の算出についても同様である。
【００３７】
ステップＳ２０３：外側差分値の算出
外側差分値は前述のように式(４)に基づく演算により算出される。
式(４)において、第１項は、注目画素５０からｍ個だけ離れた位置に存在する画素の輝度の総和を意味している。なお、ｎ＜ｍであることから、式(３)についての演算処理を内側差分値の算出と、また式(４)についての演算処理を外側差分値の算出と称している。図６において、注目画素５０からｍ個だけ離れた位置に存在する複数の画素からなる第２の画素群(画素群)５２ａは薄いグレーに塗りつぶされている。なお、図６はｍ＝４の例を示している。次に、式(４)の第２項は、注目画素５０の輝度値を示している。この第２項は係数αがあることを除いて式(３)と共通する。さらに、式(４)の第３項は、注目画素５０からｍ個だけ離れた位置に存在する画素群が構成する列に含まれる画素の数である。
【００３８】
本実施の形態では、内側差分値算出のための第１の画素群５１ａ、外側差分値算出のための第２の画素群５２ａを、図６に示す形態としたが、例えば、図７〜図９に示すような他の形態を採用することもできる。
図７に示す例は、第１の画素群(画素群)５１ｂおよび第２の画素群(画素群)５２ｂが、連続する画素のうち１個おきに選択された画素から構成される。つまり、図６に示す例は、間欠的な位置に配置された複数の画素から第１の画素群５１ｂおよび第２の画素群５２ｂを構成している。図６に示した本実施の形態によれば、１つの注目画素５０についての演算は、ｎまたはｍ個だけ離れた位置の画素の輝度値を積算するにとどまるから、その負荷がそもそも小さい。さらに図６のように、間欠的な位置にある画素のみを対象とすれば、演算に要する負荷は一段と小さくなる。
図８に示す例は、八角形状あるいは円状とでもいうべき位置に存在する複数の画素から第１の画素群(画素群)５１ｃおよび第２の画素群(画素群)５２ｃを構成している。
図９に示す例は、図８の変形例であって、注目画素５０を取り囲む全ての画素から第１の画素群(画素群)５１ｄを構成し、第１の画素群５１ｄを取り囲む全ての画素から第２の画素群(画素群)５２ｄを構成している。図６〜図９に示した第１の画素群５１ａ〜５１ｄおよび第２の画素群５２ａ〜５２ｄは以下の点で共通している。すなわち、第１の画素群５１ａ〜５１ｄおよび第２の画素群５２ａ〜５２ｄは、各々注目画素５０に対して方向性を持たない等方的な形態をなしている。また、第１の画素群５１ａ〜５１ｄおよび第２の画素群５２ａ〜５２ｄは、注目画素５０を基準として線対称(あるいは点対称)の形態をなしている。適切に輝度ムラや塵埃による輝度の変化を抽出するためには、第１の画素群５１ａ〜５１ｄおよび第２の画素群５２ａ〜５２ｄを以上のような形態とすることが重要である。
【００３９】
ステップＳ２０５：２重周囲差分値の算出
２重周囲差分値は式(５)に基づく演算処理を実行することにより算出できる。つまり、ステップＳ２０１およびステップＳ２０３によりそれぞれ算出した内側差分値と外側差分値とを加えることにより２重周囲差分値を算出することができる。ここで、式(５)中のａおよびｂは係数であり、０以上の正の数を採る。ステップＳ２０１，ステップＳ２０３，ステップＳ２０５からなる一連の処理は、分割された画像データのすべての画素に対して実行される。２重周囲差分値は、結局、各画素の相対輝度を示している。
【００４０】
ここで、ステップＳ２０１により求めた内側差分値と、ステップＳ２０３により求めた外側差分値とを加算することによって、ステップＳ２０５で求めた２重周囲差分値を再変換することにより得られた画像を図１０に示す。
この画像は、モニタ３５に表示することができる。なお、この画像は、前述のｎ，ｍをｎ＝１，ｍ＝１６とした場合のものである。また、式(４)中の係数α＝１、式(５)中の係数ａ＝０.１５、ｂ＝０.８５として演算処理した結果に基づいている。なお、内側差分値に基づくことによって、局所的な輝度ムラや塵埃による輝度の違い等が抽出される。一方、外側差分値に基づくことによって、画面全体に生ずる輝度の勾配が抽出される。そして、これら内側差分値および外側差分値を加算した２重周囲差分値に基づく図１０では、輝点および輝度の勾配の両者を抽出できることがわかる。なお、図１１に示すグラフは、図１０中に描かれている直線(Ｚ)における切断面の輝度値を示している。この図１１においては、塵埃が存在していると思われる部分は輝度値が周囲より高くなっていることがわかる。
【００４１】
本実施の形態においては、内側差分値および外側差分値を加えた２重周囲差分値を求めているが、内側差分値のみ、または外側差分値のみを用いて表示画像の画質評価を行うこともできる。また、本実施の形態では、式(４)中の係数α＝１、式(５)中の係数ａ＝０.１５、ｂ＝０.８５としたが、他の値を採用することができる。係数ａおよびｂは、内側差分値と外側差分値との重み付けを決定する。本発明者等の検討によると、抽出したい輝度ムラや塵埃の分類に応じて、内側差分値と外側差分値との重み付けを決定することにより、当該輝度ムラや異物を適切に抽出することができる。
したがって、係数ａ，ｂは、抽出したい画像の特徴に応じて、実験的に定めることが必要である。また、係数αについては、外側差分値を算出するための第２の画素群５２ａ(〜５２ｄ)によって、変動し得る。つまり、ｍが大きい値を採用するときには式(３)の第２項は第１項に比べて極めて小さな値をとることになる。例えば、ｍ＝２０だとすると、第１項は１６０個の画素における輝度値の総和であり、１個の画素(注目画素５０)の輝度値はこの総和に比べて極めて小さい。したがって、このような場合には、αを０(ゼロ)として第２項を無視することもできる。
【００４２】
ステップＳ１０９：特異値データの抽出
ステップＳ１０７で求めた２重周囲差分値、つまり各画素の相対輝度値を所定の閾値（図１１に示す閾値Ｌ１）を使用して２値化する。ここでは、閾値Ｌ１より輝度の低いものと、閾値Ｌ１と同じ若しくは輝度の高いものの二つに分けられる。２値化により抽出された特異値データは、ステップＳ１１１の画像膨張・収縮処理に供される。
ステップＳ１１１：画像膨張・収縮
ステップＳ１０９で求めた特異値データについて、所定の回数だけ画像膨張、収縮を施す。画像膨張は、図形成分の境界にあるがその値を全て背景成分の画素の値に変換して１画素分膨らます処理をいう。また、逆に画像収縮は、図形成分を１画素分だけ縮める処理をいう。画像膨張と収縮を交互に繰り返すことにより、雑音部分を除去することができる。このようにして得られた画像データの結果を出力する（ステップＳ１１３）。得られた画像を図１２に示す。
【００４３】
以上のように、本実施の形態では画像処理部３１における画像処理において、図１１に示すような切断面の輝度値の分布が得られるが、このとき、輝度ムラによる輝度値のピークより、塵埃の存在による輝度値のピーク値の方が大幅に低く、また輝度値の勾配も小さい(緩やか)。つまり、これらの点に基づき、画像処理部３１で検出される輝度値のピークが輝度ムラによるものなのか、塵埃によるのもなのかが識別できるのである。輝度ムラと判別されたものを画像データから除去し、図１２に示すような塵埃のみが白い点となって明瞭に現れた画像をモニタ３５に表示させることができる。
【００４４】
次に、塵埃検出システム１を用いた、クリーンルーム内の具体的な塵埃検出方法について説明する。図１３は塵埃検出方法のフロー図である。
初めに、検出場所における塵埃を正確に測定するため、検出場所に塵埃検出器１０を設置した状態で、測定開始前の表面１１ｓの画像を初期画像として撮像する（ステップＳ３０１）。初期画像は、検出場所ではない場所において付着した塵埃や、表面１１ｓについた傷等を含んでおり、後のステップＳ３０４の塵埃の検出において所定時間経過後の測定画像と比較するときに使用される。なお、初期画像の取り込み前に、実際の検出場所において測定時間内に載った塵埃のみを測定するため、測定前に表面１１ｓ上に既に存在している塵埃を取り除くことが好ましい。具体的方法として、例えば布等で表面１１ｓの塵埃を払拭したり、粘着ローラ等で塵埃を粘着除去することができる。
【００４５】
上記ステップＳ３０１における初期画像の取り込み終了後、画像の取り込みを止め、塵埃検出器１０をそのまま所定時間放置して表面１１ｓ上に落下した塵埃を自由に積もらせる（ステップＳ３０２）。所定時間経過したと判断した場合、次のステップＳ３０３へと進み、所定時間経過していないと判断した場合には、所定時間が経過するまで待機する。ここで、塵埃を受け止める所定時間、即ち測定時間は適宜設定することができるが、例えば１時間程度または数時間〜数日間程度である。
所定時間経過後、測定画面として表面１１ｓの画像をステップＳ３０１と同様にして撮像し（ステップＳ３０３）、撮像した画像の画像データも解析コンピュータ３０内の所定のメモリに格納する。
【００４６】
次に、画像データを解析することにより、所定時間内に表面１１ｓ上に載った塵埃の検出を行う（ステップＳ３０４）。
ステップＳ３０４では、まず初期画像と測定画像とを比較し、所定時間中に変化していない部分を示すデータを排除する。これにより、測定画像から所定時間中以外に落下した塵埃や集塵板１１や拡散板１２やミラー１３に付いた傷等を排除することができる。なお、この比較は図２および図５に示した二重差分法による処理が終了した後に行うものであってもよい。また、表面１１ｓが明らかに汚れていない場合等、測定画像から初期画像を差し引く必要が無い場合には、初期画像の測定および比較は行わなくても良い。
【００４７】
続けて、ステップＳ３０４では、上記の比較処理によって得られた画像データに対して、画像処理部３１において図２および図５に示した画像処理を行う。処理の結果、処理された画像内に塵埃あるいは輝度ムラであると判断された箇所がある場合には、画像処理部３１にて、その箇所の輝度値分布を算出し、その輝度値のピーク値が、輝度分布に基づいて予め設定された閾値Ｌ１を上回るか否かを判定する。ここで、ピーク値が所定の閾値Ｌ１を上回る程低ければ、この箇所は塵埃によるもので、ピーク値が閾値Ｌ１を上回るほど低くなければ、その箇所は輝度ムラによるものであるとして判定し、輝度ムラによるものを除去する。そして、塵埃と判定されたものを所定時間内に落下した塵埃として、測定結果データを出力する。
【００４８】
以上のようにして得られた塵埃の測定結果データをモニタ３５に映す（ステップＳ３０５）。ここで、結果データに基づき、それぞれの塵埃を分類分けし、所定分類ごとの塵埃の数を集計することが好ましい。分類分けすることにより、それぞれの分類における塵埃の発生源を特定しやすいものとなる。塵埃の分類分けは、例えば取り込まれた塵埃の形状と面積によって分類する方法など、特に限定されない。集計結果は、分類ごとの塵埃数を示すヒストグラム等によって画面に表示させることができる。
【００４９】
上述したように、塵埃検出システム１では、塵埃検出器１０における集塵板１１の表面１１ｓの輝度値の分布を定量的に評価し、塵埃や輝度ムラが生じている箇所を検出する。そして、輝度値のピーク値や傾きによって、塵埃と輝度ムラとを識別することができる。加えて、輝度ムラ、塵埃による輝度の局所的な変化を定量的に評価することができる。さらに、２重周囲差分法を用いることにより、輝度勾配およびその中で生ずる局所的な輝度ムラを抽出することができ、抽出された輝度ムラを排除することにより、より正確に塵埃の測定ができる。そして、この２重周囲差分法によれば、演算処理の負荷が小さくて済むため、演算処理の時間をセーブすることができる。したがって、この塵埃検出システム１を用いることにより、簡単、正確且つ効率的に発生した塵埃の検出を行うことができる。
また、複数の塵埃検出器１０が解析コンピュータ３０に接続されているので、複数の検出場所における塵埃の同時測定が可能である。また、解析コンピュータ３０において塵埃の測定を集中的に管理することができ、クリーンルーム内の総合的な塵埃発生を把握することができる。また解析コンピュータ３０を複数箇所設置する必要が無く、従来の塵埃検出装置と比べて場所をとらない。
【００５０】
上記したような塵埃検出システム１の構成や塵埃の検出手順は、必ずしも図１３に示したものに限らず、適宜変更することが可能である。例えば、塵埃検出器１０においてミラー１３を設けず、集塵板１１と拡散板１２にＣＣＤカメラ１５のイメージセンサが対面するように構成することもできる。また、ＣＣＤカメラ１５を拡散板１２の下でスキャンさせることにより、集塵板１１の画像を順次撮像するようにしても良い。その他、複数のミラー等の偏光板を設けることにより、ＣＣＤカメラ１５の設置箇所を適宜調整することができる。
【００５１】
また、本実施の形態においては、塵埃検出器１０が拡散板１２を備える場合、図２や図５に示す画像処理は行わなくても良い。拡散板１２により、入射した光が散乱され、シェーディング処理することにより輝度ムラが塵埃の検出の妨げにならない程度に変換される。一方、塵埃検出器１０は拡散板１２を備えなくてもよい。ただしこのとき、輝度ムラによって塵埃の検出が正確に行うことができない場合があるので、図２や図５に示す画像処理を行うことが好ましい。
【００５２】
図１４は、第２の実施の形態における塵埃検出システムの構成を説明するための図である。
図１４に示す塵埃検出システム２では、塵埃検出器２０と、ケーブル１８を介して接続された解析コンピュータ３０とから構成されている点で第１の実施の形態の塵埃検出システム１と同じであるが、塵埃検出器２０内の構成が、第１の実施の形態の塵埃検出器１０のものと異なる。以下、塵埃検出器２０について詳細を説明する。
【００５３】
塵埃検出器２０は、保護板（光透過部、測定部、塵埃受け手段、平坦部）２１と、保護板２１の下に設けられて保護板２１の表面２１ｓを撮像するＣＣＤカメラ２２と、を外枠２３内に備えて構成されている。保護板２１は、光を透過し、且つその下に設けられたＣＣＤカメラ２２を保護するものであり、例えば第１の実施の形態の集塵板１１と同じ構成のものを使用できる。ＣＣＤカメラ２２は、第１の実施の形態のＣＣＤカメラ１５と同じ構成のものを使用できる。塵埃検出器２０では、ＣＣＤカメラ２２におけるイメージセンサの大きさが、そのまま塵埃測定可能な面積となる。
【００５４】
塵埃検出システム２では、図２に示す第１の実施の形態の塵埃の検出を行う際に実行する画像処理の流れと同じようにして処理が行われる。まず、保護板２１の表面２１ｓ上に載った塵埃をＣＣＤカメラ２２により直接撮像する。図１５に直接撮像した画像を示す。撮像した画像のデータは、次にシェーディング処理を施されて、図１６に示すような画像となる。さらに、ステップＳ１０７における２重周囲差分値を算出して画像処理を行うと、図１７に示すような画像となる。図１８に示すグラフは、図１７中に描かれている直線(Ｚ)における切断面の輝度値を示している。
最終的な処理として、ステップＳ１０９において特異値データを算出した後、ステップＳ１１１の画像膨張、収縮処理を施すと、図１９に示すような画像となる。このような画像処理を、図１３に示すような測定方法に基いて行い、第２の実施の形態の塵埃検出システム２において、所定時間に落下した塵埃が測定される。
【００５５】
第２の実施の形態の塵埃検出器２０は、第１の実施の形態の塵埃検出器１０と比べると、拡散板１２やミラー１３等の構成部材を必要としないので、形状が簡単でまた寸法も小さなものとなる。よって塵埃検出器２０の設置が簡単であり、また作業台上に置く場合であっても場所を取らないので使用しやすいものとなる。
【００５６】
また塵埃検出器２０においては、表面２１ｓの画像を取り込むにあたり、撮像する面積がＣＣＤカメラ２２のイメージセンサ部分と同じ面積となっており、その面積は比較的狭い。また第１の実施の形態の塵埃検出器１０と比較して、塵埃が載る表面２１ｓから撮像部までの距離が短い。これらの結果、表面２１ｓの画像に輝度ムラが生じにくい。したがって、取り込んだ画像について図２や図５に示すような画像処理による輝度ムラを除去する処理を行わなくても、塵埃の検出を比較的精度よく行うことができる。具体的には、表面２１ｓの画像から直接塵埃を検出する方法、表面２１ｓの画像をシェーディング処理した後に塵埃を検出する方法等を採用できる。
【００５７】
上記の第１および第２の実施の形態の塵埃検出システム１または２では、複数の塵埃検出器１０または２０が解析コンピュータ３０に接続されて構成されているが、１つの塵埃検出器１０または２０が１つの解析コンピュータ３０に接続されて構成されていてもよい。さらに、塵埃検出器１０と塵埃検出器２０とが同じ解析コンピュータ３０に接続され、検出場所の状況（測定場所における測定面積、塵埃検出器の設置可能面積、外光の状態）に応じて塵埃検出器１０または塵埃検出器２０を選択して使用することも可能である。
また、塵埃検出システム１、２における塵埃の検出を自動化させることもできる。例えば、所定時間毎に自動的に画像が撮像されるように設定し、測定前に自動的に塵埃検出器の測定部である表面を払拭する機能を備えさせることもできる。
【００５８】
その他、上記の第１および第２の実施の形態では、検出場所に自然に落下した塵埃を検出しているが、塵埃検出器１０または２０の周囲のエアーを吸引することにより、塵埃を検出場所へ故意に落下させることも可能である。
これ以外にも、本発明の主旨を逸脱しない限り、上記実施の形態で挙げた構成を取捨選択したり、他の構成に適宜変更することが可能である。
【００５９】
【発明の効果】
以上説明したように、本発明によれば、クリーンルーム等の塵埃管理がなされている場所において塵埃の検出を効率良くかつ確実に行うことができる。
【図面の簡単な説明】
【図１】第１の実施の形態における塵埃検出システムの構成を説明するための図である。
【図２】塵埃の検出を行う際に実行する画像処理の流れを示す図である。
【図３】塵埃検出システムにより撮像された未処理の画像を示す図である。
【図４】シェーディング処理した画像を示す図である。
【図５】２重周囲差分法の処理手順を示すフローチャートを示す。
【図６】分割された画像データを示している。
【図７】第１の画素群および第２の画素群の他の例を示す図である。
【図８】第１の画素群および第２の画素群のさらに他の例を示す図である。
【図９】第１の画素群および第２の画素群のさらに他の例を示す図である。
【図１０】２値化処理した画像を示す図である。
【図１１】図１０の直線(Ｚ)における切断面の輝度値を示している。
【図１２】画像膨張、収縮処理した画像を示す図である。
【図１３】塵埃検出方法のフロー図である。
【図１４】第２の実施の形態における塵埃検出システムの構成を説明するための図である。
【図１５】塵埃検出システムにより撮像された未処理の画像を示す図である。
【図１６】シェーディング処理した画像を示す図である。
【図１７】２値化処理した画像を示す図である。
【図１８】図１７の直線(Ｚ)における切断面の輝度値を示している。
【図１９】画像膨張、収縮処理した画像を示す図である。
【符号の説明】
１、２・・・塵埃検出システム（測定器）、１０、２０・・・塵埃検出器（塵埃検出装置）、１１・・・集塵板（光透過部、測定部、塵埃受け手段、平坦部）、１２・・・拡散板（光拡散部、光拡散手段）、１３・・・ミラー（偏向部）、１４・・・レンズ、１５、２２・・・ＣＣＤカメラ（撮像部、撮像手段）、１８・・・ケーブル、２１・・・保護板（光透過部、測定部、塵埃受け手段、平坦部）、３０・・・解析コンピュータ、３１・・・画像処理部(コントラスト把握手段、画像処理手段、演算手段)、３２・・・判定部(塵埃検出手段、ムラ検出手段)、３３・・・集計部（集計手段、分別手段）、３４・・・撮像制御部、３５・・・モニタ(集計結果出力手段、表示手段)[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a dust detection device and the like used for managing dust generation in a clean room or the like.
[0002]
[Prior art]
In recent years, various precision devices such as a liquid crystal display (LCD) and a semiconductor device have been manufactured. These dusty instruments are manufactured in a place where dust is controlled, such as a clean room, because dust may cause malfunction of the instruments. In a clean room, there are many types of dust, such as garbage adhering to the materials of manufactured products, threads coming out of workers 'clothes, waste products falling from workers' bodies, and rubber on conveyors for running products. Therefore, in order to prevent the generation of dust, it is necessary to detect the dust, specify the type of the dust, and take measures to prevent the generation of the dust. As a specific example of a device for detecting dust, there is a cyclone dust collecting device that sucks external air and counts fine particles adsorbed on a filter.
[0003]
[Problems to be solved by the invention]
However, in the case of measuring by suctioning air, such as a cyclone dust collector, turbulence of air flow occurs in the clean room during measurement, it is difficult to accurately grasp the source of dust, and it has adhered to the filter. It is not efficient because it is necessary to count fine particles one by one. Therefore, in order to prevent the generation of dust, it is desired to develop a method capable of detecting dust accurately and efficiently.
The present invention has been made based on such a technical problem, and an object of the present invention is to provide a dust detection device, a dust detection system, a particle measurement device, and a dust detection method that can accurately detect dust. I do.
[0004]
[Means for Solving the Problems]
With this object in mind, a dust detection device for a clean room according to the present invention includes a light transmitting portion having an incident surface for receiving fallen dust and an output surface for emitting light incident from the surface, and light emitted from the light transmitting portion. And an image pickup unit for picking up an image. In this dust detection device, an image of a state where dust is placed on the incident surface is captured, and dust can be easily detected from this image. In addition, in this dust detection device, an image is taken from the emission surface side, so that the fall of dust on the incidence surface side is not hindered and the flow of air at the measurement location is not disturbed. Can be measured accurately.
[0005]
The dust detection device can be configured by stacking an imaging unit and a light transmitting unit. In this case, since the light transmitting portion and the imaging portion are continuous with each other, dust is relatively clearly reflected in the captured image, and thus, dust detection is easy. Here, the term “lamination” may mean that another layer is interposed between the imaging unit and the light transmitting unit, or that there may be a slight gap between the imaging unit and the light transmitting unit. .
[0006]
Further, the dust detection device may have a configuration including a deflecting unit that deflects light emitted from the light transmitting unit, and a lens that forms an image of the deflected light on the imaging unit. By providing the deflecting unit, the imaging unit does not need to be provided immediately below the light transmitting unit, and thus the height of the dust detection device itself can be suppressed. Further, since the image formed on the image pickup unit can be made an appropriate size by the lens, the measurement area for detecting dust can be adjusted. In the case of such a configuration, when the light diffusing unit is disposed between the light transmitting unit and the imaging unit, it is possible to reduce luminance unevenness of light generated when the light source depends on external light, that is, Since the background surface can be made closer to uniform illuminance, dust can be easily detected.
[0007]
The present invention can be regarded as a dust detection system. The dust detection system is provided in the measurement unit based on the magnitude relationship between the image contrast, the imaging unit that captures the transmitted light of the measurement unit that places dust on the surface, the contrast determination unit that captures the contrast of the captured image, and the image contrast. And a totaling result output unit that analyzes the dust identified by the detecting unit and outputs dust status data. By counting the detected dust, the degree of dust generation at a predetermined location (location where the measuring unit is installed) can be grasped. Here, the dust state data means data indicating the state of the generated dust, such as the shape and size of the dust, and the distribution according to the number and type of the dust.
[0008]
The dust detection system of the present invention can be considered from other viewpoints. Specifically, the dust detection system is a dust detection system that includes a plurality of dust detection devices and an analysis computer that collects data output from the dust detection devices. A dust receiving unit that can be placed on the dust collecting unit; an imaging unit that captures an image of the dust receiving unit; and a sending unit that sends image data of an image captured by the imaging unit to an analysis computer. It is characterized by comprising: identification means for analyzing image data sent from the detection device to identify dust; counting means for counting the identified dust; and display means for displaying data based on the counting result. . As described above, by using a plurality of dust detection devices, simultaneous measurement of dust at a plurality of locations becomes possible, and the measurement results can be integrated and totalized by an analysis computer. Note that one or more analysis computers can be configured. The dust detection device and the analysis computer can be connected by wire or wirelessly.
It is preferable that the dust detection system further includes a classification unit that classifies the dust identified by the identification unit by type, and the display unit displays a total result of each type by, for example, a histogram or the like. By referring to such a result, it is easy to identify the cause of the generation of dust. However, the data based on the aggregation result displayed by the display means is not particularly limited. For example, the aggregation result may be directly shown as a numerical value, and whether or not the aggregation result satisfies a reference value may be indicated. It may indicate the determination made.
[0009]
If the present invention is taken from another viewpoint, it can be taken as a measuring instrument for measuring fine particles.
In the measurement of the fine particles, the fine particles contained in the image are grasped by detecting the contrast of the captured image. The contrast may be detected by detecting the luminance value of the image. The outline of this procedure is as follows.
(a) The image to be evaluated is divided into a plurality of pixels.
(b) The luminance values of a plurality of pixels that are n pixels away from a specific pixel (target pixel) and surround the target pixel are integrated, and the luminance value of the target pixel is subtracted from the integrated value. This process is performed on all the divided pixels to obtain luminance information.
(c) The luminance values of a plurality of pixels that are located at positions m (n <m) pixels away from the target pixel and surround the target pixel are integrated, and the luminance value of the target pixel is subtracted from the integrated value. This process is performed on all the divided pixels to obtain luminance information.
(d) By using the luminance information based on the two luminance information obtained in (b) and (c) above, a quantitative evaluation close to a sensory test can be obtained.
Here, the luminance information obtained in (b) and (c) is a difference value between the target pixel and the pixels existing around the target pixel. That is, the luminance information obtained by (b) and (c) is a value that should also be called a peripheral difference value, and the luminance information obtained by (d) is based on two peripheral difference values. Called. Hereinafter, a method of obtaining the surrounding difference value and the double surrounding difference value is referred to as a surrounding difference method and a double surrounding difference method.
[0010]
Specifically, the measuring device includes a flat measurement surface on which fine particles can be placed and an imaging unit that images the measurement surface in a single housing. And an arithmetic unit that performs an arithmetic process represented by the following equation (1) on all the divided images to obtain P (x, y) as luminance information:
(Equation 7)

(However, pi, po, α and I (x, y) are defined as follows.
pi: luminance value of a pixel forming a first pixel group which is line-symmetric with respect to the position indicated by (x, y) in the image data
po: A luminance value of a pixel that forms a second pixel group that is line-symmetric with respect to the position indicated by (x, y) and that is located outside the first pixel group in the image data.
I (x, y): luminance value of the pixel at the position indicated by (x, y)
a, b, α: 0 and a positive number)
Contrast detection means for detecting the contrast of the image based on the luminance information obtained by the calculation means, identification means for identifying fine particles present on the measurement surface based on the magnitude relationship of the contrast, and an identification result by the identification means are output. And a result output unit.
[0011]
Here, P (x, y) is calculated, and pi is a luminance value of a pixel constituting the first pixel group that is line-symmetric with respect to the position indicated by (x, y) in the image data. , Po: luminance values of pixels constituting a second pixel group which is line-symmetric with respect to the position indicated by (x, y) and is located outside the first pixel group in the image data It is. Therefore, P (x, y) is the above-described double surrounding difference value, and luminance information can be obtained.
Further, the first pixel group is constituted by pixels existing at positions n pixels away (n is an integer of 1 or more) from pixels existing at the position indicated by (x, y), and The pixel group can be constituted by pixels located at positions separated by m pixels (m> n is an integer) from pixels located at the position indicated by (x, y). By appropriately setting n and m, it is possible to extract various forms of luminance unevenness.
Further, the coefficients a and b respectively define weights when adding the inner difference value and the outer difference value. Therefore, a and b satisfy a + b = 1.0, and a and b are determined according to the features of the image to be evaluated. Here, a or b is allowed to be 0 (zero).
In addition, α can be determined according to the value of m. For example, if the value of m increases, I (x, y) can be ignored for the sum of po. In that case, the arithmetic processing can be performed with α set to 0 (zero).
[0012]
Alternatively, as a measuring device for measuring fine particles, a measuring surface imaging means in which a flat measuring surface on which fine particles can be placed and an imaging unit for imaging the measuring surface are housed in a single housing, and an image related to the captured image. Image processing means for dividing the data, arithmetic means for performing an arithmetic process represented by the following equation (2) on the divided image data,
(Equation 8)

(However, pi: a luminance value of a pixel forming a pixel group surrounding a pixel existing at a position indicated by (x, y) in image data; I (x, y): indicated by (x, y) (The luminance value of the pixel at the position)
Contrast detecting means for detecting the contrast of a captured image based on luminance information obtained by the calculating means, identifying means for identifying fine particles present on a measurement surface based on the magnitude relationship of the contrast of the captured image, and identifying means And a result output means for outputting a result of the identification in step (a).
[0013]
Further, the present invention can be considered as a dust detection method.
The dust detection method focuses on a step of capturing an image of a flat portion on which dust is placed, a step of dividing image data of the captured image into a plurality of pixels, and a distribution of luminance values of the plurality of pixels. A pixel-of-interest setting step of setting a pixel; a difference calculation step of calculating a difference between the sum of the luminance values of the pixels surrounding the pixel of interest and a luminance value of the pixel of interest; And a counting step of counting the items determined to be dust in the determining step.
When the dust detection method includes a luminance difference detecting step of detecting a luminance difference between a pixel of interest and a pixel group when light is transmitted, the pixel of interest represents dust when the difference is equal to or greater than a predetermined threshold in the determination step. Can be determined.
[0014]
The present invention also provides a method for detecting dust in a clean room, wherein the step of receiving the dust dropped in the clean room at the flat portion for a predetermined time, the step of capturing a transmission image of the flat portion receiving the dust, and the step of capturing an image of the captured transmission image The method includes the steps of storing data and detecting dust by analyzing image data. In this case, before receiving the dust on the flat portion, the method further includes capturing an initial transmission image of the flat portion. In the step of detecting dust, the initial transmission image and the transmission image are compared within a predetermined time. The received dust can be detected more accurately.
Note that when shading processing is performed on the image data, dust in the image appears more clearly, which facilitates detection. In the step of detecting dust, the dust can be detected by identifying the shadow formed by the dust on the flat portion and the uneven brightness on the flat portion, and extracting data of the shadow of the dust.
[0015]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, the present invention will be described in detail based on embodiments shown in the accompanying drawings.
FIG. 1 is a diagram for explaining the configuration of the dust detection system according to the first embodiment. As shown in FIG. 1, a dust detection system (measuring device for measuring fine particles) 1 for measuring dust is connected to a dust detector (dust detection device) 10 and a dust detector 10 via a cable 18. And an analysis computer 30. Although not shown, a plurality of dust detectors 10 are connected to the analysis computer 30.
[0016]
The dust detector 10 includes a dust collecting plate (light transmitting unit, measuring unit, dust receiving unit, flat portion) 11 on which dust that falls freely is placed, and a diffuser plate (light diffusing unit, A light diffusing unit) 12, a mirror (deflecting unit) 13 for reflecting light transmitted through the dust collecting plate 11 and the diffusing plate 12, a lens 14, and a CCD camera (imaging unit, imaging unit) 15 for imaging an image. Have.
[0017]
The dust collecting plate 11 is configured to transmit light. In the present embodiment, the dust placed on the surface 11s of the dust collecting plate 11 is measured. The surface 11s of the dust collecting plate 11 is flat so that a dust image can be easily captured. In the case where the dust is measured several times, it is preferable to use a dust collecting plate 11 having high strength in consideration of removing dust placed on the surface 11s before the measurement. For example, sapphire glass and crystal glass can be used.
[0018]
The diffusion plate 12 diffuses the light transmitted through the dust collection plate 11, and is not particularly limited as long as it can diffuse the light. Specific examples include an acrylic plate having a thickness of about 2 to 3 mm.
The mirror 13 reflects the light transmitted through the diffusion plate 12 toward the CCD camera 15. Instead of the mirror 13, a lens or a deflecting plate, which can deflect incident light in a predetermined direction, can be applied.
[0019]
The lens 14 forms an image of the light reflected by the mirror 13 on the CCD camera 15. The lens 14 is not particularly limited. However, it is preferable that the depth of field can be determined from the dust collecting plate 11 to the diffusing plate 12, and it is more preferable that the focal point be adjusted to the surface 11 s of the dust collecting plate 11. You can do it. As a specific example, a symmetric lens composed of a plurality of lenses such as a Gauss-type lens and a Tessa-type lens that can correct aberration due to the color of light can be given.
[0020]
The CCD camera 15 may be either a black-and-white type or a color type, but in the present embodiment, a black-and-white type that is advantageous in terms of the current imaging capability is employed. In the present embodiment, the CCD camera 15 captures an image of the entire area of the dust collecting plate 11. The image picked up by the CCD camera 15 is converted into an image signal, and sent to an analysis computer 30 via a cable 18 from an interface (sending means) (not shown).
[0021]
The analysis computer 30 performs an image processing unit (contrast grasping means, image processing means, calculating means) 31 for performing predetermined processing described later in detail on the image captured by the CCD camera 15, and performs image capturing based on the processed image. Determination unit (dust detection unit, unevenness detection unit) 32 for determining how much dust is present in the extracted image, aggregation unit (aggregation unit, classification unit) 33 for counting the items determined to be dust, and imaging by CCD camera 15 (Monitoring result output means, display means) for appropriately displaying the result of determination by the determination section 32, the image captured by the image processing section 31, the processed image, and the like. .
[0022]
When the dust detection system 1 having the above configuration is used, the dust detection system 1 is installed at a location where dust is detected (hereinafter, a detection location), and an image of the surface 11 s of the dust collecting plate 11 is captured by the CCD camera 15 and the captured image is taken. The data on the analysis computer 30 is processed to measure the dust placed on the surface 11s.
Specifically, first, the dust detector 10 is installed at a location where dust is detected (hereinafter, a detection location) such that the surface 11s of the dust collecting plate 11 faces upward and the surface 11s is horizontal. . Then, the dust placed on the front surface 11s is captured as an image by the CDD camera 15, and the captured image data is processed by the image processing unit 31 to measure the placed dust. Here, the dust that is naturally dropped on the surface 11s is measured. Note that the generated dust falls not only at the place where the dust is generated but also at the place where the dust falls due to the influence of, for example, an air flow in a clean room. Although the detection location can be set as appropriate, it is installed near a product to be worked, for example, near a worker or on a work bench. Further, the setting place of the analysis computer 30 is not particularly limited, and may be near the dust detector 10 or at a remote place. The plurality of dust detectors 10 are installed at different detection locations.
[0023]
In capturing an image, first, light from a light source passes through the dust collecting plate 11. At this time, since light does not pass through a portion where the dust placed on the surface 11s exists, the dust appears as a shadow in the obtained image of the surface 11s. As a light source, an ordinary light such as a fluorescent lamp or the like installed in a clean room as a detection place is used as a light source, but an imaging light source may be provided near the dust detector 10. However, in this case, it is preferable to install in a position where the flow of air in the clean room is not changed.
[0024]
The light transmitted through the dust collection plate 11 is diffused by the diffusion plate 12. Depending on the type and state of the light source, “luminance unevenness” occurs on the dust collecting plate 11 where the luminance of the incident light varies depending on the location. Since the uneven brightness hinders the detection of dust, the uneven brightness is reduced by diffusing the incident light by the diffusion plate 12. At this time, the difference in luminance between the portion where dust is present and the portion where no dust is present is greater in contrast than luminance unevenness. Therefore, even if the luminance is changed by the diffusion plate, the detection of dust is not significantly affected. Absent.
[0025]
The light transmitted through the diffusion plate 12 is deflected by the mirror 13 in the horizontal direction. Further, the light deflected by the mirror 13 is condensed by the lens 14 and captured by the image sensor portion of the CCD camera 15. At this time, since the size of the formed image can be adjusted by adjusting the focal point of the lens 14, the image sensor portion of the CCD camera 15 does not need to have the same area as the area of the surface 11s which is the measurement area. . Therefore, in the present embodiment, the measurement area on the surface 11s can be made larger than the image sensor portion of the CCD camera 15.
[0026]
Further, the dust detector 10 is provided with a mirror 13, and the CCD 13 is not required to be provided directly below the dust collecting plate 11 by deflecting the incident light by the mirror 13. Therefore, the height of the dust detector 10 can be made relatively small, and the surface 11s, which is the measurement surface, can be located closer to the product to be worked.
[0027]
The data of the image captured by the CCD camera 15 in this way is sent to the analysis computer 30 via the cable 18, and the image data is stored in a predetermined memory in the image processing unit 31. In the detection of dust, the obtained image of the surface 11s is analyzed, and luminance unevenness in the image is eliminated to detect only the dust placed on the surface 11s. This principle will be described below.
[0028]
FIG. 2 is a diagram illustrating a flow of image processing executed when dust is detected.
First, the image processing unit 31 processes an image and performs detection in a procedure as shown in FIG. In the following processing, detailed descriptions of well-known image processing methods are omitted. The detection method according to the present embodiment includes capturing of an image to be inspected (step S101), division processing of captured image data (step S103), smoothing processing (step S105), application of a double surrounding difference method (step S107), The process is performed by a series of steps of a singular value data extraction process (Step S109) and an image expansion / contraction process (Step S111). Hereinafter, the contents of each step will be described in the order of steps.
[0029]
Step S101: Image capture
First, an image of the surface 11s is captured by the CCD camera 15. Image data of the captured image is transmitted to the analysis computer 30 via the cable 18. FIG. 3 shows an image of the surface 11s taken. As shown in FIG. 3, small whitish spots are slightly observed in this image in some places. These are mainly dusts placed on the surface 11s.
Then, the transmitted image data is subjected to shading correction (shading processing) in the image processing unit 31. This shading correction removes the mismatch of the conversion characteristic between the original luminance of the measured image and the image data, which is mainly caused by the vignetting in the CCD camera 15 and the non-uniformity of the image sensor. At this time, by changing the luminance, the luminance is corrected so that the higher luminance becomes red and the lower luminance becomes green. FIG. 4 shows an image obtained by such processing. In FIG. 4, dust is more clearly seen than that shown in FIG. On the other hand, luminance unevenness appears more prominently in the entire image than that shown in FIG.
[0030]
Step S103: Division of captured image data
The image data obtained in step S101 is divided into a matrix. This division of the image data is a preliminary process for executing the double surrounding difference method in step S107 described later. Here, it is assumed that the image is divided into nxm pixels. The position of the divided pixel can be specified by the two-dimensional coordinates (x, y). In the present embodiment, since the CCD camera 15 is used as the imaging means, nxm corresponds to the number of pixels (resolution) of the CCD camera 15. Although the maximum number of divisions is the number of pixels (resolution) of the CCD camera 15, this does not prevent the number of divisions from being smaller than this.
[0031]
Step S105: smoothing process
The image data divided in step S103 is subjected to a smoothing process for removing noise. For example, 2 × 2 filtering is performed.
[0032]
Step S107: Applying the double surrounding difference method
Assuming that the luminance value of each pixel is obtained for the image data from which noise has been removed in step S105, the double surrounding difference method having the following contents is executed.
FIG. 5 is a flowchart showing a processing procedure of the double surrounding difference method. In step S107, as shown in FIG. 5, first, an inside difference value represented by the following equation (3) is calculated (step S201). Next, an outer difference value represented by the following equation (4) is calculated (step S203). Next, a double surrounding difference value represented by the following equation (5) is calculated by adding the inside difference value obtained in step S201 and the outside difference value obtained in step S203 (step S205). Hereinafter, the specific contents of steps S201 to S203 will be described. The order of step S201 for calculating the inside difference value and step S203 for calculating the outside difference value may be interchanged. The double surrounding difference method is an algorithm newly developed by the present inventors. Equations (3) to (5) are empirical equations as a result of various studies by the present inventors, and 2n + 1 (third term) of equation (3) and α ( The second term), 2m + 1 (third term), and the coefficients a and b in equation (5) are derived empirically.
[0033]
(Equation 9)

(However, pi: a luminance value of a pixel forming a first pixel group surrounding a pixel existing at a position indicated by (x, y) in the image data, I (x, y): (x, y ), The luminance value of the pixel at the position indicated by n), n: the number of pixels from the target image to the first pixel group)
[0034]
(Equation 10)

(However, po: a luminance value of a pixel constituting a second pixel group surrounding a pixel existing at a position indicated by (x, y) in the image data, I (x, y): (x, y ), The luminance value of the pixel at the position indicated by m), m: the number of pixels from the target image to the second pixel group, α: a positive number equal to or greater than 0)
[0035]
(Equation 11)

(However, a, b: positive numbers satisfying a + b = 1.0)
[0036]
Step S201: Calculation of inside difference value
The inner difference value is calculated based on the equation (3) as described above. The contents of equation (3) will be described with reference to FIG. FIG. 6 shows the image data divided in step S103.
In the equation (3), the first term means the sum of the luminances of the pixels located at positions n away from the target pixel 50. In FIG. 6, a pixel painted black is the target pixel 50, and a first pixel group (pixel group) 51a composed of a plurality of pixels located n positions apart from each other is painted dark gray. FIG. 6 shows an example where n = 2. Next, the second term of the equation (3) indicates the luminance value of the target pixel 50. Further, the third term of the equation (3) is the number of pixels included in a pixel column located at a position n away from the target pixel 50.
It should be noted that, depending on the position of the target pixel 50, there may be a case where no pixel exists at a position separated by n pixels. In that case, the calculation may be performed by using the brightness values of the pixels existing at positions n positions apart by mirroring. The same applies to the calculation of the outer difference value in step S203.
[0037]
Step S203: Calculation of the outside difference value
The outer difference value is calculated by the calculation based on Expression (4) as described above.
In Equation (4), the first term means the sum of the luminances of the pixels located at positions m away from the target pixel 50. Since n <m, the calculation processing of Expression (3) is referred to as calculation of an inner difference value, and the calculation processing of Expression (4) is referred to as calculation of an outer difference value. In FIG. 6, a second pixel group (pixel group) 52a including a plurality of pixels located at positions m away from the target pixel 50 is painted in light gray. FIG. 6 shows an example where m = 4. Next, the second term of Expression (4) indicates the luminance value of the target pixel 50. This second term is common to equation (3) except that there is a coefficient α. Further, the third term of the equation (4) is the number of pixels included in a column formed by a pixel group located m positions away from the target pixel 50.
[0038]
In the present embodiment, the first pixel group 51a for calculating the inside difference value and the second pixel group 52a for calculating the outside difference value are configured as shown in FIG. 6. Other forms as shown in FIG. 9 can also be adopted.
In the example shown in FIG. 7, the first pixel group (pixel group) 51b and the second pixel group (pixel group) 52b are configured by pixels selected every other one of continuous pixels. That is, in the example shown in FIG. 6, the first pixel group 51b and the second pixel group 52b are configured from a plurality of pixels arranged at intermittent positions. According to the present embodiment shown in FIG. 6, the calculation for one target pixel 50 is only a summation of the luminance values of the pixels at positions n or m away from each other, so that the load is small in the first place. Further, as shown in FIG. 6, if only pixels at intermittent positions are targeted, the load required for the calculation is further reduced.
In the example shown in FIG. 8, a first pixel group (pixel group) 51c and a second pixel group (pixel group) 52c are configured from a plurality of pixels existing at positions that can be called an octagon or a circle. .
The example shown in FIG. 9 is a modification of FIG. 8, in which a first pixel group (pixel group) 51d is formed from all the pixels surrounding the target pixel 50, and all the pixels surrounding the first pixel group 51d. To form a second pixel group (pixel group) 52d. The first pixel groups 51a to 51d and the second pixel groups 52a to 52d shown in FIGS. 6 to 9 are common in the following points. That is, the first pixel groups 51a to 51d and the second pixel groups 52a to 52d are each in an isotropic form having no direction with respect to the target pixel 50. The first pixel groups 51a to 51d and the second pixel groups 52a to 52d are line-symmetric (or point-symmetric) with respect to the target pixel 50. It is important that the first pixel groups 51a to 51d and the second pixel groups 52a to 52d be configured as described above in order to appropriately extract luminance variations due to luminance unevenness and dust.
[0039]
Step S205: Calculation of the double surrounding difference value
The double surrounding difference value can be calculated by executing a calculation process based on Expression (5). That is, the double surrounding difference value can be calculated by adding the inner difference value and the outer difference value calculated in steps S201 and S203, respectively. Here, a and b in the equation (5) are coefficients, and take a positive number equal to or greater than 0. A series of processing including Step S201, Step S203, and Step S205 is executed for all pixels of the divided image data. The double surrounding difference value eventually indicates the relative luminance of each pixel.
[0040]
Here, an image obtained by re-converting the double surrounding difference value obtained in step S205 by adding the inner difference value obtained in step S201 and the outer difference value obtained in step S203 is shown in FIG. It is shown in FIG.
This image can be displayed on the monitor 35. This image is obtained when n and m are n = 1 and m = 16. In addition, the calculation is based on the result of calculation processing with the coefficient α = 1 in the equation (4) and the coefficient a = 0.15 and b = 0.85 in the equation (5). Note that, based on the inside difference value, local luminance unevenness, luminance difference due to dust, and the like are extracted. On the other hand, based on the outer difference value, the gradient of the luminance generated in the entire screen is extracted. In FIG. 10 based on the double surrounding difference value obtained by adding the inner difference value and the outer difference value, it can be seen that both the bright spot and the gradient of the luminance can be extracted. Note that the graph shown in FIG. 11 shows the luminance value of the cut surface along the straight line (Z) drawn in FIG. In FIG. 11, it can be seen that the luminance value is higher in the portion where dust is considered to be present than in the surroundings.
[0041]
In the present embodiment, the double surrounding difference value obtained by adding the inside difference value and the outside difference value is obtained. However, the image quality of the display image may be evaluated using only the inside difference value or only the outside difference value. it can. In the present embodiment, the coefficient α = 1 in the equation (4) and the coefficients a = 0.15 and b = 0.85 in the equation (5), but other values can be adopted. . The coefficients a and b determine the weighting of the inner difference value and the outer difference value. According to the study by the present inventors, by determining the weighting of the inside difference value and the outside difference value according to the classification of the luminance unevenness or dust to be extracted, the luminance unevenness or foreign matter can be appropriately extracted. .
Therefore, the coefficients a and b need to be experimentally determined according to the characteristics of the image to be extracted. Further, the coefficient α may vary depending on the second pixel group 52a (to 52d) for calculating the outer difference value. That is, when the value of m is large, the second term of the expression (3) takes an extremely small value as compared with the first term. For example, if m = 20, the first term is the sum of the brightness values of 160 pixels, and the brightness value of one pixel (the target pixel 50) is extremely smaller than this sum. Therefore, in such a case, the second term can be ignored by setting α to 0 (zero).
[0042]
Step S109: Extract singular value data
The double surrounding difference value obtained in step S107, that is, the relative luminance value of each pixel is binarized using a predetermined threshold (threshold L1 shown in FIG. 11). Here, it is divided into two types: one having a luminance lower than the threshold L1 and one having the same or higher luminance than the threshold L1. The singular value data extracted by the binarization is subjected to the image expansion / contraction processing in step S111.
Step S111: Image expansion / contraction
Image expansion and contraction are performed on the singular value data obtained in step S109 a predetermined number of times. The image expansion is a process of expanding the image component by one pixel by converting all the values at the boundary of the graphic component into the values of the pixels of the background component. Conversely, image shrinkage refers to a process of shrinking a graphic component by one pixel. By repeating image expansion and contraction alternately, a noise portion can be removed. The result of the image data thus obtained is output (step S113). FIG. 12 shows the obtained image.
[0043]
As described above, in the present embodiment, in the image processing performed by the image processing unit 31, the distribution of the luminance values of the cut surface as shown in FIG. 11 is obtained. , The peak value of the luminance value is much lower, and the gradient of the luminance value is smaller (slower). That is, based on these points, it is possible to identify whether the peak of the luminance value detected by the image processing unit 31 is caused by luminance unevenness or dust. An image determined to be uneven in luminance is removed from the image data, and an image in which only dust as white dots clearly appear as shown in FIG. 12 can be displayed on the monitor 35.
[0044]
Next, a specific dust detection method in a clean room using the dust detection system 1 will be described. FIG. 13 is a flowchart of the dust detection method.
First, in order to accurately measure dust at the detection location, an image of the surface 11s before the start of measurement is taken as an initial image with the dust detector 10 installed at the detection location (step S301). The initial image includes dust attached to a location other than the detection location, a scratch on the surface 11s, and the like, and is used when comparing with a measurement image after a predetermined time has elapsed in the detection of dust in step S304. . In addition, before taking in the initial image, in order to measure only the dust placed within the measurement time at the actual detection location, it is preferable to remove the dust already existing on the surface 11s before the measurement. As a specific method, for example, dust on the surface 11s can be wiped off with a cloth or the like, or the dust can be removed with an adhesive roller or the like.
[0045]
After completion of the capture of the initial image in step S301, the capture of the image is stopped, and the dust detector 10 is left as it is for a predetermined period of time to allow the dust dropped on the surface 11s to freely accumulate (step S302). If it is determined that the predetermined time has elapsed, the process proceeds to the next step S303. If it is determined that the predetermined time has not elapsed, the process waits until the predetermined time has elapsed. Here, the predetermined time for receiving dust, that is, the measurement time can be set as appropriate, but is, for example, about one hour or several hours to several days.
After a lapse of a predetermined time, an image of the surface 11s is captured as a measurement screen in the same manner as in step S301 (step S303), and image data of the captured image is also stored in a predetermined memory in the analysis computer 30.
[0046]
Next, dust on the surface 11s is detected within a predetermined time by analyzing the image data (step S304).
In step S304, first, the initial image and the measured image are compared, and data indicating a portion that has not changed during a predetermined time is excluded. Thereby, it is possible to eliminate dust, a dust collecting plate 11, a diffusion plate 12, and a scratch on the mirror 13 which fall from the measurement image other than during the predetermined time. This comparison may be performed after the processing by the double difference method shown in FIGS. 2 and 5 is completed. Further, when it is not necessary to subtract the initial image from the measurement image, for example, when the surface 11s is not clearly dirty, the measurement and comparison of the initial image need not be performed.
[0047]
Subsequently, in step S304, the image processing unit 31 performs the image processing shown in FIGS. 2 and 5 on the image data obtained by the above-described comparison processing. As a result of the processing, if there is a portion determined to be dust or uneven brightness in the processed image, the image processing section 31 calculates a brightness value distribution of the portion, and calculates a peak value of the brightness value. Is greater than or equal to a threshold value L1 set in advance based on the luminance distribution. Here, if the peak value is lower as the value exceeds a predetermined threshold L1, this portion is determined to be due to dust, and if the peak value is not lower as to exceed the threshold value L1, the portion is determined to be due to luminance unevenness. Eliminate unevenness. Then, the measurement result data is output as the dust determined as dust falling within a predetermined time.
[0048]
The dust measurement result data obtained as described above is displayed on the monitor 35 (step S305). Here, it is preferable that each dust is classified and classified based on the result data, and the number of dusts for each predetermined classification is counted. The classification makes it easier to identify the source of dust in each classification. The classification of the dust is not particularly limited, for example, a method of classifying the dust according to the shape and area of the taken dust. The tally result can be displayed on the screen by a histogram or the like indicating the number of dusts for each classification.
[0049]
As described above, the dust detection system 1 quantitatively evaluates the distribution of the luminance value of the surface 11 s of the dust collecting plate 11 in the dust detector 10 and detects a portion where dust or uneven luminance occurs. Then, dust and uneven brightness can be distinguished by the peak value or the slope of the brightness value. In addition, it is possible to quantitatively evaluate luminance unevenness and local changes in luminance due to dust. Furthermore, by using the double surrounding difference method, a luminance gradient and local luminance unevenness generated therein can be extracted, and dust can be measured more accurately by eliminating the extracted luminance unevenness. . According to the double peripheral difference method, the load of the arithmetic processing can be reduced, and the time for the arithmetic processing can be saved. Therefore, by using the dust detection system 1, it is possible to simply, accurately, and efficiently detect the generated dust.
Further, since the plurality of dust detectors 10 are connected to the analysis computer 30, it is possible to simultaneously measure dust at a plurality of detection locations. In addition, the analysis computer 30 can centrally manage the measurement of dust, and can grasp the total dust generation in the clean room. Further, there is no need to install the analysis computer 30 at a plurality of locations, and the analysis computer 30 requires less space than a conventional dust detection device.
[0050]
The configuration of the dust detection system 1 and the dust detection procedure as described above are not necessarily limited to those illustrated in FIG. 13 and can be appropriately changed. For example, the dust detector 10 may be configured so that the image sensor of the CCD camera 15 faces the dust collecting plate 11 and the diffusion plate 12 without providing the mirror 13. Alternatively, the CCD camera 15 may be scanned below the diffusion plate 12 to sequentially capture images of the dust collection plate 11. In addition, by providing a plurality of polarizing plates such as mirrors, the installation position of the CCD camera 15 can be appropriately adjusted.
[0051]
In the present embodiment, when the dust detector 10 includes the diffusion plate 12, the image processing illustrated in FIGS. 2 and 5 may not be performed. The diffused plate 12 scatters the incident light, and by shading processing, converts the luminance unevenness to a level that does not hinder dust detection. On the other hand, the dust detector 10 may not include the diffusion plate 12. However, at this time, it may not be possible to accurately detect dust due to uneven brightness. Therefore, it is preferable to perform the image processing shown in FIGS. 2 and 5.
[0052]
FIG. 14 is a diagram for explaining the configuration of the dust detection system according to the second embodiment.
The dust detection system 2 shown in FIG. 14 is the same as the dust detection system 1 of the first embodiment in that the dust detection system 2 includes a dust detector 20 and an analysis computer 30 connected via a cable 18. However, the configuration inside the dust detector 20 is different from that of the dust detector 10 of the first embodiment. Hereinafter, the dust detector 20 will be described in detail.
[0053]
The dust detector 20 includes a protection plate (light transmitting unit, measuring unit, dust receiving means, flat portion) 21 and a CCD camera 22 provided below the protection plate 21 and imaging the surface 21 s of the protection plate 21. It is provided in the outer frame 23. The protection plate 21 transmits light and protects the CCD camera 22 provided thereunder. For example, the protection plate 21 may have the same configuration as the dust collection plate 11 of the first embodiment. As the CCD camera 22, a camera having the same configuration as the CCD camera 15 of the first embodiment can be used. In the dust detector 20, the size of the image sensor in the CCD camera 22 is the area where the dust can be measured as it is.
[0054]
In the dust detection system 2, processing is performed in the same manner as in the flow of image processing executed when dust is detected according to the first embodiment shown in FIG. First, dust captured on the surface 21s of the protection plate 21 is directly imaged by the CCD camera 22. FIG. 15 shows an image directly taken. The data of the captured image is then subjected to shading processing to obtain an image as shown in FIG. Further, when image processing is performed by calculating the double surrounding difference value in step S107, an image as shown in FIG. 17 is obtained. The graph shown in FIG. 18 shows the luminance value of the cut surface along the straight line (Z) drawn in FIG.
As a final process, after calculating the singular value data in step S109, the image is expanded and contracted in step S111 to obtain an image as shown in FIG. Such image processing is performed based on a measuring method as shown in FIG. 13, and the dust detection system 2 according to the second embodiment measures dust that has fallen for a predetermined time.
[0055]
Compared with the dust detector 10 of the first embodiment, the dust detector 20 of the second embodiment does not require components such as the diffusion plate 12 and the mirror 13, so that the shape is simple and the size is small. Will also be small. Therefore, the dust detector 20 can be easily installed, and even if it is placed on a workbench, it does not take up much space and is easy to use.
[0056]
Further, in capturing the image of the surface 21s, the dust detector 20 has the same area as the image sensor portion of the CCD camera 22 for capturing an image of the surface 21s, and the area is relatively small. Further, as compared with the dust detector 10 of the first embodiment, the distance from the surface 21s on which dust is placed to the imaging unit is shorter. As a result, luminance unevenness hardly occurs in the image of the surface 21s. Accordingly, dust detection can be performed with relatively high accuracy without performing a process of removing luminance unevenness by image processing as shown in FIGS. 2 and 5 for a captured image. Specifically, a method of directly detecting dust from the image of the surface 21s, a method of detecting dust after shading the image of the surface 21s, and the like can be adopted.
[0057]
In the dust detection system 1 or 2 of the first and second embodiments, the plurality of dust detectors 10 or 20 are connected to the analysis computer 30, but one dust detector 10 or 20 is used. May be connected to one analysis computer 30. Further, the dust detector 10 and the dust detector 20 are connected to the same analysis computer 30, and the dust detection is performed in accordance with the situation of the detection place (measurement area at the measurement place, area where the dust detector can be installed, state of external light). It is also possible to select and use the detector 10 or the dust detector 20.
In addition, the dust detection in the dust detection systems 1 and 2 can be automated. For example, it is possible to provide a function of automatically setting an image to be taken every predetermined time and automatically wiping the surface of the dust detector before measurement.
[0058]
In addition, in the above-described first and second embodiments, dust naturally dropped to the detection location is detected. However, the air around the dust detector 10 or 20 is sucked to detect the dust. It is also possible to drop intentionally.
In addition, the configuration described in the above embodiment can be selected or changed to another configuration as appropriate without departing from the gist of the present invention.
[0059]
【The invention's effect】
As described above, according to the present invention, it is possible to efficiently and reliably detect dust in a place where dust is managed, such as a clean room.
[Brief description of the drawings]
FIG. 1 is a diagram for describing a configuration of a dust detection system according to a first embodiment.
FIG. 2 is a diagram showing a flow of image processing executed when dust is detected.
FIG. 3 is a diagram showing an unprocessed image captured by a dust detection system.
FIG. 4 is a diagram showing an image subjected to shading processing.
FIG. 5 is a flowchart showing a processing procedure of the double surrounding difference method.
FIG. 6 shows divided image data.
FIG. 7 is a diagram illustrating another example of a first pixel group and a second pixel group.
FIG. 8 is a diagram illustrating still another example of the first pixel group and the second pixel group.
FIG. 9 is a diagram illustrating still another example of the first pixel group and the second pixel group.
FIG. 10 is a diagram illustrating an image subjected to a binarization process.
11 shows a luminance value of a cut surface along a straight line (Z) in FIG.
FIG. 12 is a diagram showing an image subjected to image expansion and contraction processing.
FIG. 13 is a flowchart of a dust detection method.
FIG. 14 is a diagram illustrating a configuration of a dust detection system according to a second embodiment.
FIG. 15 is a diagram showing an unprocessed image captured by the dust detection system.
FIG. 16 is a diagram showing an image subjected to shading processing.
FIG. 17 is a diagram showing an image subjected to a binarization process.
18 shows a luminance value of a cut surface along a straight line (Z) in FIG.
FIG. 19 is a diagram showing an image subjected to image expansion and contraction processing.
[Explanation of symbols]
1, 2 ... dust detection system (measuring device), 10, 20 ... dust detector (dust detection device), 11 ... dust collecting plate (light transmitting part, measuring part, dust receiving means, flat part) ), 12: diffusion plate (light diffusion unit, light diffusion unit), 13: mirror (deflection unit), 14: lens, 15, 22: CCD camera (imaging unit, imaging unit), 18 ... Cable, 21 ... Protective plate (light transmitting part, measuring part, dust receiving means, flat part), 30 ... Analysis computer, 31 ... Image processing part (contrast grasping means, image processing means) , Calculation means), 32 ... determination section (dust detection means, unevenness detection means), 33 ... counting section (counting means, classification means), 34 ... imaging control section, 35 ... monitor (counting) (Result output means, display means)

Claims (19)

A light transmitting portion having an incident surface for receiving the dropped dust, and an exit surface from which light incident from the surface exits,
An imaging unit for imaging light emitted from the light transmitting unit;
A dust detection device for a clean room, comprising: The dust detection device for a clean room according to claim 1, wherein the imaging unit and the light transmission unit are stacked. A deflecting unit that deflects light emitted from the light transmitting unit,
A lens that forms an image of the light deflected by the deflection unit on the imaging unit;
The dust detection device for a clean room according to claim 1, further comprising: The dust detection device for a clean room according to claim 3, wherein a light diffusion unit is disposed between the light transmission unit and the imaging unit. Imaging means for imaging the transmitted light of the measurement unit for placing dust on the surface,
A contrast grasping means for grasping the contrast of the image taken by the imaging means,
Based on the magnitude relationship of the contrast of the captured image, a detection unit that detects dust present in the measurement unit,
Tally result analysis means for analyzing the dust identified by the detection means and outputting dust state data,
Image processing means for dividing image data relating to the image taken by the imaging means,
An arithmetic unit for performing an arithmetic process represented by the following equation (1) on all the images divided by the image processing unit to obtain P ( x, y ) as luminance information ;

(However, pi, po, α and I ( x, y ) are defined as follows.
pi: A luminance value of a pixel constituting a first pixel group that is line-symmetric with respect to a position indicated by ( x, y ) in the image data
po: Among the image data , luminance values of pixels forming a second pixel group that is line-symmetric with respect to the position indicated by ( x, y ) and that is located outside the first pixel group
I ( x, y ) : the luminance value of the pixel at the position indicated by ( x, y )
a, b, α: 0 and positive numbers)
A dust detection system , wherein the contrast grasping means detects the contrast of the image based on luminance information obtained by the calculating means . The image processing apparatus further includes a non-uniformity detecting unit that detects non-uniformity in luminance of the image based on the luminance information obtained by the arithmetic unit. The dust detection system according to claim 5, wherein the dust detection is performed. A light diffusing unit, wherein the light diffused by the light diffusing unit is imaged by the imaging unit. The dust detection system according to claim 5, wherein: Imaging means for imaging the transmitted light of the measurement unit for placing dust on the surface,
  A contrast grasping means for grasping the contrast of the image taken by the imaging means,
  Based on the magnitude relationship of the contrast of the captured image, a detection unit that detects dust present in the measurement unit,
  Tally result analysis means for analyzing the dust identified by the detection means and outputting dust state data,
  Image processing means for dividing image data relating to the image taken by the imaging means,
  The arithmetic processing represented by the following equation (2) is performed on all the images divided by the image processing means, and the luminance information is obtained. A ( x, y ) Computing means for obtaining

(However, pi: Among the image data, ( x, y ) The luminance value of the pixels constituting the pixel group surrounding the pixels existing at the position indicated by,
      I ( x, y ) :  ( x, y ) The luminance value of the pixel at the position indicated by
A dust detection system, wherein the contrast grasping means detects the contrast of the image based on luminance information obtained by the calculating means. 9. The dust detection system according to claim 8 , further comprising a light diffusion unit, wherein the image pickup unit captures light diffused by the light diffusion unit. A dust detection system including a plurality of dust detection devices, and an analysis computer that collects data output from the dust detection devices,
The dust detecting device,
Dust receiving means capable of placing dust on the surface;
Imaging means for capturing an image of the dust receiving means;
Sending means for sending image data of an image taken by the imaging means to the analysis computer;
Wherein the analyzing computer,
The contrast of the image is detected based on P ( x, y ) , which is luminance information obtained by performing an arithmetic process represented by the following equation (1) on the image data sent from the dust detection device. identifying means for identifying dust by,

(However, pi, po, α and I ( x, y ) are defined as follows.
pi: A luminance value of a pixel constituting a first pixel group that is line-symmetric with respect to a position indicated by ( x, y ) in the image data
po: Among the image data , luminance values of pixels forming a second pixel group that is line-symmetric with respect to the position indicated by ( x, y ) and that is located outside the first pixel group
I ( x, y ) : the luminance value of the pixel at the position indicated by ( x, y )
a, b, α: 0 and positive numbers)
Tabulation means for tabulating the dust identified by the identification means,
Separation means for separating the dust identified by the identification means by type,
A display means for displaying data based on the aggregation result by the aggregation means, and a display means for also displaying the aggregation result based on the classification by type ,
A dust detection system comprising: A dust detection system including a plurality of dust detection devices, and an analysis computer that collects data output from the dust detection devices,
  The dust detection device,
    Dust receiving means capable of placing dust on the surface;
    Imaging means for capturing an image of the dust receiving means;
    Sending means for sending image data of an image taken by the imaging means to the analysis computer;
  And the analysis computer comprises:
    A which is luminance information obtained by performing an arithmetic process represented by the following equation (2) on the image data sent from the dust detection device. ( x, y ) Based on, identification means for identifying dust by detecting the contrast of the image,

(However, pi: Among the image data, ( x, y ) The luminance value of the pixels constituting the pixel group surrounding the pixels existing at the position indicated by,
      I ( x, y ) :  ( x, y ) The luminance value of the pixel at the position indicated by
    Tabulation means for tabulating the dust identified by the identification means,
    Separation means for separating the dust identified by the identification means by type,
    A display unit that displays data based on the aggregation result by the aggregation unit, and a display unit that also displays the aggregation result based on the classification by type,
  A dust detection system comprising: A measuring device for measuring fine particles,
A flat measurement surface on which fine particles can be placed, and a measurement surface imaging unit in which an imaging unit for imaging the measurement surface is housed in one housing,
Image processing means for dividing image data relating to an image captured by the measurement surface imaging means,
An arithmetic unit for performing an arithmetic process represented by the following equation (1) on all the images divided by the image processing unit to obtain P (x, y) as luminance information;

(However, pi, po, α and I (x, y) are defined as follows.
pi: a luminance value of a pixel constituting a first pixel group which is line-symmetric with respect to a position indicated by (x, y) in the image data po: (x, y) in the image data A luminance value I (x, y) of a pixel forming a second pixel group that is symmetrical with respect to the indicated position and that is located outside the first pixel group is represented by (x, y). Luminance value a, b, α of pixel at position: 0 and positive number)
A contrast detection unit that detects a contrast of the image captured by the measurement surface imaging unit based on luminance information obtained by the calculation unit;
Identification means for identifying fine particles present on the measurement surface based on the magnitude relationship of the contrast of the captured image,
A result output unit that outputs a result of identification by the identification unit,
A fine particle measuring device comprising: A measuring device for measuring fine particles,
A flat measurement surface on which fine particles can be placed, and a measurement surface imaging unit in which an imaging unit for imaging the measurement surface is housed in one housing,
Image processing means for dividing image data relating to an image captured by the measurement surface imaging means,
An arithmetic unit that performs an arithmetic process represented by the following equation (2) on the image data divided by the image processing unit to obtain A ( x, y ) that is luminance information ;

(However, pi: a luminance value of a pixel forming a pixel group surrounding a pixel existing at a position indicated by (x, y) in the image data,
I (x, y): luminance value of the pixel at the position indicated by (x, y)
A contrast detection unit that detects a contrast of the image captured by the measurement surface imaging unit based on luminance information obtained by the calculation unit;
Identification means for identifying fine particles present on the measurement surface based on the magnitude relationship of the contrast of the captured image,
A result output unit that outputs a result of identification by the identification unit,
A fine particle measuring device comprising: Capturing an image of a flat portion on which dust is placed;
A dividing step of dividing the image data of the captured image into a plurality of pixels,
A target pixel setting step of setting a target pixel based on a distribution of luminance values of the plurality of pixels;
A difference calculating step of calculating a difference between the sum of the brightness values of the pixels constituting the pixel group surrounding the pixel of interest and the brightness value of the pixel of interest,
A determination step of determining whether or not the pixel of interest represents dust based on the difference;
A counting step of counting what is determined to be dust in the determining step;
A dust detection method comprising: Further comprising a luminance difference detecting step of detecting a luminance difference between the pixel of interest and the pixel group when light is transmitted;
15. The dust detection method according to claim 14, wherein in the determining step, when the difference is equal to or greater than a predetermined threshold, the target pixel is determined to represent dust. Receiving the fallen dust on the flat portion for a predetermined time;
Imaging a transmission image of the flat portion that has received the dust,
Storing image data of the captured transmission image;
Analyzing the image data to detect dust,
A method for detecting dust in a clean room, comprising: Before receiving the dust, further comprising capturing an initial transmission image of the flat portion,
17. The dust detection method in a clean room according to claim 16, wherein in the step of detecting the dust, the initial transmitted image is compared with the transmitted image to detect dust received within the predetermined time. . 17. The method for detecting dust in a clean room according to claim 16, further comprising a step of shading the image data. In the step of detecting the dust, identifying the dust formed by the dust on the flat portion and the uneven brightness in the flat portion, and extracting the data of the shadow of the dust to detect the dust. The method for detecting dust in a clean room according to claim 16, wherein:

Images