JP3885112B2 - White flaw detection device and white flaw correction device for television video signal - Google Patents

Description

【００１８】
同様に、信号が図２（ｂ）のようになるためには、次の（２）式の関係にあることが必要である。

【００１９】
さらに、信号が図２（ｃ）のようになるためには、次の（３）式の関係にあることが必要である。

【００２０】
従って、いま、入力テレビジョン映像信号の波形が図２（ａ）の波形（従って、１画素の白キズ）となっているかどうかを調べるには、上述の（１）式に含まれる各式において、例えば、右辺のＶ_T を左辺に移項した式、Ｓ₂ −Ｓ₁ −Ｖ_T ＞０、Ｓ₂ −Ｓ₃ −Ｖ_T ＞０、｜Ｓ₁ −Ｓ₃ ｜−Ｖ_T ＜０を作り、これら移項した式をそれぞれａ₁ 、ａ₂ 、ａ₃ で表すと、ａ₁ ＞０、ａ₂ ＞０、ａ₃ ＜０であるから、これら式の符号が正の場合を“１”、負の場合を“０”で表すものとすると、（ａ₁ ，ａ₂ ，ａ₃ ）＝（１，１，０）のように表すことができる。
【００２１】
従って、この（１，１，０）は上記約束のもとに１画素の白キズを符号化したものと見做すことができ、この（１，１，０）をメモリに参照符号として記憶しておき、一方では、入力テレビジョン映像信号に対して、ａ₁ 、ａ₂ 、ａ₃ の演算を実時間で行うものとする。もし、この演算によって得られた（ａ_{1 ,} ａ₂ ，ａ₃ ）が（１，１，０）になれば、それをメモリに記憶されている符号（１，１，０）と比較することによって１画素の白キズであると判断する。同様にして、入力テレビジョン映像信号がその時点において２画素、３画素の白キズであるかどうかの判断も行うことができる。
【００２２】
以上の（１）乃至（３）式で規定される関係は、各サンプル点における信号の立上がり、立下がりおよび信号レベルがどうなっているかを確認するもので、これらは、いずれも１画素前後における信号のレベル差、および１画素および／または複数画素の前後におけるレベル差の絶対値のレベルとスレッショルド信号のレベルとのレベル差に基づいた関係式となっている。
【００２３】
次に、本発明による白キズ検出回路の一実施形態を図３に示し、その回路構成および回路動作につき説明する。なお、図３に示す白キズ検出回路は、白キズによる画面の水平方向のノイズパターンを検出するように構成しているが、画素単位の遅延量を１水平走査期間の遅延量に置き替えるだけで、画面の垂直方向のノイズパターンを検出するように構成することも可能である。
【００２４】
本実施形態の白キズ検出回路は、図３に示すように、４個の減算器（ＳＵＢ１〜４）と８個のシフトレジスタ（ＳＲ１〜８）で構成されているが、これら減算器およびシフトレジスタによってディジタルテレビジョン映像信号（以下では、単にテレビジョン映像信号と言う）はそれぞれ３画素および１画素遅延するものと仮定する。また、回路の入力信号は、各サンプル点ごとのテレビジョン映像信号とスレッショルド信号Ｖ_T である。なお、テレビジョン映像信号は時間の流れの古い方から順にＳ_{1 ,} Ｓ₂ ，・・・Ｓ₁₀（Ｓ₁₀は現サンプル点での信号）で示すものとする。
【００２５】
図 3に示すように、入力端子にテレビジョン映像信号Ｓ₁₀（サンプル点Ｓ₁₀での映像信号）が入力された時点において、シフトレジスタＳＲ１およびＳＲ２からそれぞれ１画素および２画素遅れた信号（それぞれＳ₉ およびＳ₈ ）が出力されるが、それら信号（Ｓ₉ およびＳ₈ ）はそれぞれ被減数および減数信号として減算器ＳＵＢ１に供給される。減算結果は減算に伴う遅延により３画素分遅れて出力することになるため、現時点では３画素分前の信号の減算結果が絶対値｜Ｓ₆ −Ｓ₅ ｜と符号Ｓ₆ −Ｓ₅ とに別れて出力されている。
【００２６】
また減算器ＳＵＢ２において、スレッショルド信号Ｖ_T から絶対値｜Ｓ₆ −Ｓ₅ ｜の減算が行われるが、ここでも３画素遅延のため、減算器ＳＵＢ２の出力信号は現時点でＶ_T −｜Ｓ₃ −Ｓ₂ ｜（これを、ａ₃ とする）である。この信号をシフトレジスタＳＲ３により１画素分遅延して現時点出力Ｖ_T −｜Ｓ₂ −Ｓ₁ ｜（これを、ａ₁ とする）が得られる。一方、減算器ＳＵＢ１の符号出力Ｓ₆ −Ｓ₅ は、直列接続されたシフトレジスタＳＲ４〜７に供給され、シフトレジスタＳＲ６から符号出力Ｓ₃ −Ｓ₂ （これを、ａ₄ とする）が、シフトレジスタＳＲ７から符号出力Ｓ₂ −Ｓ₁ （これを、ａ₂ とする）がそれぞれ得られる。
【００２７】
さらに、減算器３において、直列接続されたシフトレジスタＳＲ１，ＳＲ２の両端の信号間の減算を行い、現時点で絶対値｜Ｓ₇ −Ｓ₅ ｜を得、これを減算器４にてスレッショールド信号Ｖ_T から差引き、シフトレジスタＳＲ８を通すことによりＶ_T −｜Ｓ₃ −Ｓ₁ ｜（これを、ａ₅ とする）が得られる。
【００２８】
以上において、ａ₁ からａ₅ までの数式を整理すると、次の（４）式で表すことができ、
ａ₁ ：Ｖ_T −｜Ｓ₂ −Ｓ₁ ｜
ａ₂ ：Ｓ₂ −Ｓ₁ （符号）
ａ₃ ：Ｖ_T −｜Ｓ₃ −Ｓ₂ ｜ (4)
ａ₄ ：Ｓ₃ −Ｓ₂ （符号）
ａ₅ ：Ｖ_T −｜Ｓ₃ −Ｓ₁ ｜
【００２９】
これと、単１画素の白キズを示す図１（ａ）とを比較すると、次の（５）式を満足するとき、
ａ₁ ：＜０
ａ₂ ：＞０
ａ₃ ：＜０ (5)
ａ₄ ：＜０
ａ₅ ：＞０
テレビジョン映像信号Ｓ₂ は白キズであったことが分かる。
【００３０】
これを具体的に実行するために、（５）式をａ₁ 〜ａ₅ のそれぞれについて符号が正のとき“１”、負のとき“０”と約束して符号で表すものとすれば、この関係は次の（６）式で表すことができる。
（ａ₁ ，ａ₂ ，ａ₃ ，ａ₄ ，ａ₅ ）＝（０，１，０，０，１） (6)
すなわち、図３に示す本発明の一実施形態の白キズ検出回路において、ａ₁ からａ₅ までの各出力信号が（６）式の関係を満足していれば、テレビジョン映像信号Ｓ₂ は白キズであるということになる。
【００３１】
従って、本発明による白キズ検出装置としては、符号列（０，１，０，０，１）を予めメモリに記憶しておき、これと図３に示す検出回路の出力（ａ₁ ，ａ₂ ，ａ₃ ，ａ₄ ，ａ₅ ）の各符号列を比較照合して、入力テレビジョン映像信号が白キズであるかどうかを判定することができるように構成するものとする。
【００３２】
以上においては、図１（ａ）に示す１画素の白キズを示す白キズ検出回路の一実施形態を示したが、図１（ｂ）または（ｃ）に示す２画素以上連続した白キズを検出する白キズ検出回路も図３の構成に倣って構成することができる。
【００３３】
また、上述した白キズ検出回路の入力信号のうち、スレッショールド信号Ｖ_T （図３参照）は、入力テレビジョン映像信号の信号レベルに対応して変化させることが望ましい。これは、通常ＣＣＤテレビジョンカメラで発生した白キズのレベルがカメラのプロセス回路を通すと映像信号のレベルが高くなるのに伴って圧縮されるため、圧縮された白キズに対しても、その白キズの検出が誤動作なく行なえるようにするためである。具体的には、白キズのレベル圧縮に伴い、スレッショールド信号Ｖ_T のレベルも小さくなるように、したがって、スレッショールド信号のレベルをｋ・ｖ_T で表したとき（Ｖ_T ＝ｋ・ｖ_T ）、入力映像信号のレベル増大に伴ってｋの値が漸減するように、例えば、図４に示す特性にする。
【００３４】
次に白キズ検出回路に入力されるテレビジョン映像信号について考える。
白キズから発生するノイズは一定レベルであるのに対し、テレビジョン映像信号に含まれる周辺のノイズはランダムノイズが主体なので、これを巡回フィルタに通すことにより時間的なランダムノイズが軽減し、白キズ検出回路における白キズの検出がより容易になる。また、テレビジョン映像信号を、フィルタまたはフレームについて時間加算することにより動解像度が低減し、白キズの検出が一層容易になる。
【００３５】
図５に、巡回フィルタの一実施形態を示している。
図５において、まず、入力テレビジョン映像信号は加算器ＡＤＤの入力端子の一方に供給される。この加算器ＡＤＤの出力信号は１フレームのフレームメモリ１に供給されて蓄積されるが、毎フレームごとに読み出された信号は加算器ＡＤＤにより現時点の入力テレビジョン映像信号に加算され、累積されたテレビジョン映像信号となって再びフレームメモリ１に蓄積される。図示の実施形態では、最大２５６枚の映像信号が蓄積されるよう、フレームメモリ１を制御するメモリコントロール２、メモリコントロールに接続されたタイミング信号発生器３及び蓄積枚数設定器４が配置されている。
【００３６】
また、ビットシフタ５は、フレームメモリ１において映像信号を累積しているため、フレームメモリ１の読出し信号のレベル（振幅）が大きくなっているのをビットシフトにより１枚の映像信号のレベルに戻すためのものである。このビットシフタ５のシフト量は、また蓄積枚数設定器４によって制御される。ビットシフタ５の出力映像信号は、例えば、図３に示した白キズ検出回路の入力端子に供給される。
【００３７】
また、図５に示すように、蓄積枚数を２ⁿ （ｎ＝１，２，３，・・・８）とすることにより、出力映像信号のビットシフトのみで入力信号レベルと同じレベルの出力映像信号レベルが得られ、特別な除算回路が不要となるため、回路規模を小さくすることができる。蓄積枚数を２５６枚とした場合、ランダムノイズを２４ｄＢ改善することができ、通常のハイビジョンカメラの映像信号（ペデスタル部分）のＳ／Ｎ約４０ｄＢを６０ｄＢ〜６４ｄＢ程度まで改善し、白キズの検出がより容易になる。また、この巡回フィルタによれば、動きのある入力テレビジョン映像信号の場合、動解像度も低減するので、白キズの検出が一層容易になる。
【００３８】
以上で、テレビジョン映像信号の白キズ検出回路の説明を終わるが、全体の装置としては、画面の水平、垂直方向に１画素乃至１画素以上連続した白キズ（図１参照）を検出するためにそれぞれの方向での白キズ検出回路を独立して設ける必要がある。
【００３９】
また、以上のようにして検出された白キズを補正する（ここで補正とは、映像信号の補間を行い、もともと白キズがなかったような自然の映像に近づけることを言う）ためには、白キズが検出された場所のテレビジョン画面上での位置（アドレス）を知る必要があるが、これは例えば、テレビジョン信号の垂直、水平同期信号をカウントし、検出した場所で出力にパルス信号を出すことによって知ることができる。
【００４０】
次に、本発明テレビジョン映像信号の白キズ補正装置について説明する。
ここでも、以下に説明する一実施形態では、画面水平方向の白キズを画素補間（以下、単に補間と言う）する場合、すなわち左右のテレビジョン映像信号を用いて補間するものとする。
【００４１】
図６は、連続する画素点Ｐ₃ ，Ｐ₄ ，Ｐ₅ ，Ｐ₆ ，Ｐ₇ のうちＰ₄ ，Ｐ₅ ，Ｐ₆ が連続する白画素（３画素にわたる白画素）である場合、これら画素点を画素点Ｐ₃ ，Ｐ₇ の重み付けした値で補間する場合の補間係数α₄ ，α₅ ，α₆ を示している。ここで、図６（ａ）は直線補間で、図６（ｂ）は曲線補間の場合であり、白キズＰ₄ ，Ｐ₅ ，Ｐ₆ の補間された値（信号レベル）は、それぞれ次の（７）式によって表される。
Ｐ₇ ×α₄ ＋Ｐ₃ ×（１−α₄ ）
Ｐ₇ ×α₅ ＋Ｐ₃ ×（１−α₅ ） (7)
Ｐ₇ ×α₆ ＋Ｐ₃ ×（１−α₆ ）
【００４２】
従って、直線補間で白キズＰ₄ ，Ｐ₅ ，Ｐ₆ を補間する場合には、この（７）式のα₄ ，α₅ ，α₆ に図６（ａ）の０．２５，０．５，０．７５をそれぞれ代入し、また曲線補間で白キズＰ₄ ，Ｐ₅ ，Ｐ₆ を補間する場合には、図６（ｂ）の０．１２５，０．５，０．８７５をそれぞれ代入することによって補間された映像のレベルが得られる。なお、曲線補間の場合の曲線の形状はこれに限らず、無数にある。従って曲線補間の補間係数α₄ ，α₅ ，α₆ はそれぞれ０．１２５，０．５，０．８７５に限るものではない。
【００４３】
また、図７は、実際に、上述した白キズの補間を行なう白キズ補間回路の一実施形態を示している。
図７において、入力テレビジョン映像信号はシフトレジスタＳＲ１，ＳＲ２，・・・，ＳＲ４により１画素ずつ遅延され、シフトレジスタＳＲ４からＰ₄ の時点の白キズ信号を出力しているものとする。この時点で、入力テレビジョン映像信号が順次供給されるラッチＬＡＴＣＨ−１および２は、それぞれタイミング信号Ｔ₁ およびＴ₂ によって出力信号Ｐ₃ およびＰ₇ をホールドし続けているものとする。
【００４４】
また、乗算器ＭＵＬＴ−１および２のこれらホールドされている信号が供給されていない側の入力端子には、図示のようにそれぞれ１−α₄ ，１−α₅ ，１−α₆ およびα₄ ，α₅ ，α₆ が供給され、タイミングに合わせてそれぞれ乗算結果を出力する。それぞれ乗算器ＭＵＬＴ−１および２による両乗算結果は加算器ＡＤＤに供給されて加算される。上述したようなシフトレジスタＳＲ４から白キズが出力されているＰ₄ の時点では、その時点の補間信号Ｐ₃ ×（１−α₄ ）＋Ｐ₇ ×α₄ が出力されている（直線補間、曲線補間は係数αの値が異なるだけで、（７）式で示される演算式は同じである。従って、図７に示す補間回路の構成も補間の仕方で代わることはない）。このとき、タイミング信号Ｔ₃ によって切り換えスイッチＳＷを制御して切り換え接点をｂ側に倒し、補間信号が出力されるようにする。なお、タイミング信号Ｔ₁ ，Ｔ₂ ，Ｔ₃ は、白キズの位置（アドレス）を示すアドレス信号から得られる。
【００４５】
最後に、以上説明したテレビジョン映像信号の白キズ検出装置および白キズ補正装置を一体に構成した一実施形態を図８に示し、これにつき説明する。
本発明装置の適用対象となるテレビジョン映像信号は、ビットレート１．４８５Ｇｂｐｓのディジタルシリアル信号の形態のハイビジョン信号である。図８において、入力テレビジョン映像信号と書かれている入力信号は、上記のハイビジョン信号がシリアル−パラレル変換され、１０ビットのパラレル信号になっているものとする。このパラレル信号は、図５に示した巡回フィルタ６と図７に示した白キズ補間回路７に供給される。前者は、ランダムノイズを軽減し、動画像の動き解像度を低減して、次段の水平および垂直各方向の白キズ検出回路８および９（図３参照）における白キズの検出を容易にするためのものである。また、後者の白キズ補間回路７では、白キズの箇所の周辺の映像レベルから直線補間または曲線補間してより自然なテレビジョン映像出力を得るようにしている。
【００４６】
またスレッショールド信号は、これが水平および垂直方向の白キズ検出回路８，９に供給されるに先立ち、スレッショールド信号レベル制御回路１０に供給され、ここで図４に示すようなレベル制御が行なわれる。具体的には、上述したように、スレッショールド信号のレベルをｋ・ｖ_T で表したとき（Ｖ_T ＝ｋ・ｖ_T）、入力映像信号（図８においては、巡回フィルタ６の出力信号）のレベル増大に伴ってｋの値が漸減するようにしたものである。また、白キズ検出回路８，９は、水平、垂直の各方向に独立してそれぞれ設ける必要があることは上述した。
【００４７】
さらに、それらいずれかの検出回路８，９で白キズが検出されたときは、その検出されたことがそれらの出力側に配置されるＯＲゲートを介してアドレス信号生成回路１１に伝達される。アドレス信号生成回路１１においては、例えば、テレビジョン信号の垂直、水平同期信号をもとに、白キズが検出された場所のアドレスを示すアドレス信号を生成し、これを白キズ補間回路７に供給する。補間回路７においては、直線補間または曲線補間を適用して白キズを補間されたレベルの画素に置き換えるが、これについては上述したとおりである。
【００４８】
【発明の効果】
本発明によれば、白キズの発生した箇所の映像レベルが時間的に変化している（動画）場合、或いは時間的には変化しないが一定の映像レベルが存在する映像信号（静止画）から、自動的に白キズの位置検出を行うことができる。また、本発明によれば、これまで困難であった白キズの位置検出を容易に行うことができ、白キズの検出と補間という一連の処理を自動的に行うことにより、短時間で精度の高い白キズ補正を行うことができる。
【００４９】
さらに、本発明によれば、入力テレビジョン映像信号を直接に白キズ検出回路に供給するのでなく、いったん巡回フィルタにおいてテレビジョン映像信号に含まれる時間的なランダムノイズを軽減させるとともに、動解像度を低下させて白キズのみを目立たせるようにした結果、白キズ検出回路における白キズ検出動作をより容易にした。
【００５０】
さらにまた、白キズ検出回路に供給されるスレッショールド信号のレベルをテレビジョン映像信号のレベルの増大に対応させて減少させたことは、本発明白キズ検出装置の動作を一層確実にした。
【００５１】
以上においては、ＣＣＤカメラで発生した白キズの検出および補正について説明したが、高感度ＨＡＲＰ(High-gain Avalanche Rushing amorphous Photoconductor)方式撮像管を使用したカメラでも同様の白キズが発生する。この場合に発生した白キズは、ＣＣＤカメラの白キズよりも面積において大きく、また、白キズによるノイズレベルも高くなる。本発明は、このような白キズに対しても有効に作用する。
【図面の簡単な説明】
【図１】 白キズのラスタ上における見え方を示している。
【図２】 白キズから発生するノイズパターンを時間的に変化する波形信号として示している。
【図３】 本発明白キズ検出装置中、とくに白キズ検出回路の一実施形態を示している。
【図４】 図３に示す白キズ検出回路に供給されるスレッショールド信号レベルの入力映像信号レベルに対する特性を示している。
【図５】 本発明白キズ検出装置中、とくに巡回フィルタの一実施形態を示している。
【図６】 本発明白キズ補正装置で使用する直線補間および曲線補間の各補間係数を示している。
【図７】 本発明白キズ補正装置の白キズ補間回路の一実施形態を示している。
【図８】 本発明白キズ検出装置と白キズ補正装置とを一体に構成した一実施形態を示している。
【符号の説明】
１ フレームメモリ
２ メモリコントロール
３ タイミング信号発生器
４ 蓄積枚数設定器
５ ビットシフタ
６ 巡回フィルタ
７ 白キズ補間回路
８ 水平方向白キズ検出回路
９ 垂直方向白キズ検出回路
１０ スレッショールド信号レベル制御回路
１１ アドレス信号生成回路
ＳＵＢ 減算器
ＡＤＤ 加算器
ＭＵＬＴ 乗算器
ＳＲ シフトレジスタ
ＬＡＴＣＨ ラッチ[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a white flaw detection device and a white flaw correction device for a television video signal, and in particular, automatically detects a white flaw from a television video signal in a normal shooting state by digital processing and corrects it. It is something to try.
[0002]
[Prior art]
In a television camera using a CCD imaging device, one or more of the devices may not perform regular photoelectric conversion due to some external factor. This appears as a defect (white flaw) having a certain level of brightness at a fixed position on the screen, and the image quality is significantly impaired.
[0003]
The conventional white defect correction for CCD television cameras is to detect white defects by closing the iris of the camera, reducing the level of the television video signal to the black level, and detecting a signal that exceeds a certain threshold level. The correction was performed.
[0004]
[Problems to be solved by the invention]
However, white scratches that occur during normal shooting
(1) The video level of the place where white scratches have changed,
(2) Due to the influence of random noise on the television video signal, it could not be detected separately from the television video signal.
Therefore, on the screen of a moving image in which the level of the television video signal changes with time, or on the screen of a still image in which there is a video signal containing a certain level of noise that does not change in time with the level of the video signal. The conventional method cannot cope with the automatic detection of white spots at fixed points (because the conventional method does not detect white scratches during normal shooting).
[0005]
An object of the present invention is when the level of a video signal in a place where white flaws occur is temporally changing (moving image), or a video signal having a constant signal level that does not change temporally (still image) It is an object of the present invention to provide a white flaw detection device and a white flaw correction device for a television video signal that automatically detect white flaws from the image and further correct the detected white flaws naturally.
According to this, automatic detection of white scratches, which has been difficult until now, can be easily performed, and by performing a series of processing for correction according to the present invention automatically, white scratches with high accuracy can be achieved in a short time. Correction can be performed.
[0006]
[Means for Solving the Problems]
In order to achieve the above-described object, a white defect detection device for a television video signal according to the present invention stores encoded white defect signals of a plurality of pixels continuous in the horizontal and / or vertical direction of a television video signal. Storage means, encoding means for supplying a television video signal, and encoding the supplied television video signal in the horizontal and / or vertical direction of the television video signal according to the encoding technique of the encoding, Comparing means for sequentially comparing the code of the output of the encoding means and each of the encoded white scratch signals of a plurality of continuous pixels read out from the storage means is provided.
[0007]
Also, in the white flaw detection apparatus for a television video signal according to the present invention, the encoding method includes a difference in signal level before and after the one pixel and a level before and after the one pixel and / or the plurality of pixels. The present invention is characterized in that the technique is performed based on the level difference between the level of the absolute value of the difference and the level of the threshold signal.
[0008]
According to another aspect of the present invention, there is provided a device for detecting a white flaw in a television video signal, wherein the level of the threshold signal changes in accordance with the signal level of the supplied television video signal. .
[0009]
According to another aspect of the present invention, there is provided a white flaw detection apparatus for a television video signal, wherein the supplied television video signal is placed in front of the encoding means before the television video signal is supplied to the encoding means. The random noise is reduced by the recursive filter , and the dynamic resolution is reduced.
[0010]
Further, the present invention provides a white defect correcting device for a television image signal, which includes a white defect detecting device for a television image signal, and a television image of left and right and / or upper and lower white defects detected by the white defect detecting device. A white scratch interpolation circuit for linearly or curvedly interpolating the white scratch using a signal is further provided.
[0011]
In the white defect correction apparatus for a television video signal according to the present invention, the linear or curved interpolation is performed such that the signal level of the pixel immediately before the white defect is Pi and the signal level of the pixel immediately after the pixel is Pk, and there are n continuous signals therebetween. If there are white pixels, Pk × α ₁ + Pi × (1−α ₁ )
Pk × α ₂ + Pi × (1−α ₂ )
・
・
・
Pk × α _n + Pi × (1−α _n )
(Where α ₁ , α ₂ ,... Α _n are interpolation coefficients)
Sequential pixel interpolation is performed according to the level expressed by the following.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, the present invention will be described in detail based on an embodiment of the invention with reference to the accompanying drawings.
The device for detecting white scratches in a television video signal according to the present invention can detect white scratches from a television video signal in a normal shooting state (not a black level video signal shot with the iris of a television camera closed as in the prior art). The white defect correction device corrects the detected white defect part into a natural shape so that it is almost indistinguishable from other image parts. Here, the television image signal and the white defect are corrected. The characteristics possessed by and are as follows.
[0013]
First, white scratches occur because one or more elements constituting a CCD image sensor (imaging plate) do not perform normal photoelectric conversion. When white scratches are regarded as noise, the noise level is constant. Is unchanged. On the other hand, the television video signal itself is not only constantly changing (in the case of a moving image), but also random noise is superimposed as ambient noise. The same applies to the superposition of random noise in the case of still images.
[0014]
Further, white scratches are most often generated in only one pixel, but in rare cases, depending on the defective portion of the CCD image sensor, as shown in FIGS. Appear in the horizontal direction, the vertical direction, and white scratches continuous in the horizontal and vertical directions. In FIG. 1, the length of the white scratch is exaggerated, but the continuous white scratch is limited to about 3 pixels at most.
[0015]
When a noise pattern generated from white scratches that are continuous with only one pixel, two pixels, and three pixels is shown as a time-varying waveform signal, they are shown in FIGS. 2 (a), 2 (b), and 2 (c), respectively. It becomes like this. As can be seen from the figure, both waveforms are convex in the positive direction.
[0016]
In the present invention, when white scratches are automatically detected, the level relationship for each sample point of the digitized input television video signal is examined in real time, and this is stored in advance in a memory. When the level relationship of the white scratches that are continuous for 1 to 3 pixels coincides, the video signal is determined to be white scratches.
[0017]
This will be described more specifically.
First, as shown in the figure, the white scratch of only one pixel shown in FIG. 2A has signal levels S ₁ , S _{2, and} S _{3 at} sequential sample points. When the threshold level V _T for determining whether or not there is a significant level difference is set, in order for the signal to be as shown in FIG. is required.

[0018]
Similarly, in order for the signal to become as shown in FIG. 2B, it is necessary to have a relationship of the following equation (2).

[0019]
Furthermore, in order for the signal to become as shown in FIG. 2C, it is necessary to have the relationship of the following equation (3).

[0020]
Therefore, now, in order to check whether the waveform of the input television video signal is the waveform of FIG. 2 (a) (therefore, a white defect of one pixel), in each equation included in the above equation (1), For example, an expression in which V _T on the right side is shifted to the left side, S ₂ −S ₁ −V _T > 0, S ₂ −S ₃ −V _T > 0, | S ₁ −S ₃ | −V _T <0 When these transferred expressions are expressed as a ₁ , a ₂ , and a ₃ , respectively, a ₁ > 0, a ₂ > 0, and a ₃ <0. Therefore, when the signs of these expressions are positive, “1” If the negative case is represented by “0”, it can be represented as (a ₁ , a ₂ , a ₃ ) = ( ₁ , ₁ , 0).
[0021]
Therefore, this (1, 1, 0) can be regarded as an encoded white defect of one pixel based on the above promise, and this (1, 1, 0) is stored as a reference code in the memory. On the other hand, it is assumed that the operations of a ₁ , a ₂ , and a ₃ are performed in real time on the input television video signal. If (a _1, a ₂ , a ₃ ) obtained by this operation becomes ( _{1, 1, 0} ), compare it with the code ( _{1, 1, 0} ) stored in the memory. Based on the above, it is determined that there is a white defect of one pixel. Similarly, it can be determined whether or not the input television video signal is a white defect of 2 pixels or 3 pixels at that time.
[0022]
The relations defined by the above equations (1) to (3) are for confirming the rise, fall, and signal level of the signal at each sample point. The relational expression is based on the signal level difference and the level difference between the level of the absolute value of the level difference before and after one pixel and / or a plurality of pixels and the level of the threshold signal.
[0023]
Next, an embodiment of the white flaw detection circuit according to the present invention is shown in FIG. 3, and its circuit configuration and circuit operation will be described. Note that the white defect detection circuit shown in FIG. 3 is configured to detect a horizontal noise pattern on the screen due to white defects, but it simply replaces the delay amount in pixel units with the delay amount in one horizontal scanning period. Thus, it is possible to configure to detect a noise pattern in the vertical direction of the screen.
[0024]
As shown in FIG. 3, the white scratch detection circuit according to the present embodiment includes four subtractors (SUB1 to SUB4) and eight shift registers (SR1 to SR8). It is assumed that a digital television video signal (hereinafter simply referred to as a television video signal) is delayed by 3 pixels and 1 pixel by the register. The input signals of the circuit are a television video signal and a threshold signal V _T for each sample point. It is assumed that the television video signal is indicated by S _1, S ₂ ,... S ₁₀ (S ₁₀ is a signal at the current sample point) in order from the oldest time flow.
[0025]
As shown in FIG. 3, when the television video signal S ₁₀ (video signal at the sampling point S ₁₀ ) is input to the input terminal, signals that are delayed by one pixel and two pixels from the shift registers SR 1 and SR 2 (respectively, S ₉ and S ₈ ) are output, and these signals (S ₉ and S ₈ ) are supplied to the subtractor SUB1 as a subtracted signal and a reduced signal, respectively. Since the subtraction result is output with a delay of 3 pixels due to the delay associated with the subtraction, at present, the subtraction result of the signal 3 pixels before becomes the absolute value | S ₆ -S ₅ | and the sign S ₆ -S ₅ . It is output separately.
[0026]
In the subtractor SUB2, the absolute value | S ₆ -S ₅ | is subtracted from the threshold signal V _T. However, because of the three-pixel delay, the output signal of the subtractor SUB2 is currently V _T − | S _3. −S ₂ | (this is a ₃ ). This signal is delayed by one pixel by the shift register SR3 to obtain a current output V _T − | S ₂ −S ₁ | (this is a ₁ ). On the other hand, the sign outputs S ₆ -S ₅ of the subtractor SUB 1 are supplied to the shift registers SR 4 to SR 7 connected in series, and the sign outputs S ₃ -S ₂ (referred to as a ₄ ) from the shift register SR 6 are: Code outputs S ₂ -S ₁ (this is a ₂ ) are obtained from the shift register SR7.
[0027]
Further, the subtracter 3 performs subtraction between signals at both ends of the shift registers SR1 and SR2 connected in series to obtain an absolute value | S ₇ −S ₅ | at the present time. By subtracting from the signal V _T and passing through the shift register SR8, V _T − | S ₃ −S ₁ | (referred to as a ₅ ) is obtained.
[0028]
In the above, when formulas from a ₁ to a ₅ are arranged, they can be expressed by the following formula (4):
a ₁ : V _T − | S ₂ −S ₁ |
a _2: S ₂ -S ₁ (code)
a ₃ : V _T − | S ₃ −S ₂ | (4)
a _4: S ₃ -S ₂ (code)
a ₅ : V _T − | S ₃ −S ₁ |
[0029]
When this is compared with FIG. 1A showing white scratches of a single pixel, when the following equation (5) is satisfied,
a ₁ : <0
a ₂ :> 0
a ₃ : <0 (5)
a ₄ : <0
a ₅ :> 0
It can be seen that the television video signal S ₂ was white.
[0030]
In order to specifically execute this, if the expression (5) is represented by a sign with a promise of “1” when the sign is positive and “0” when negative for each of a _{1 to} a ₅ , This relationship can be expressed by the following equation (6).
(A ₁ , a ₂ , a ₃ , a ₄ , a ₅ ) = (0, ₁ , 0, 0, ₁ ) (6)
That is, in the white flaw detection circuit of the embodiment of the present invention shown in FIG. 3, if each output signal from a ₁ to a ₅ satisfies the relationship of equation (6), the television video signal S ₂ is It means that there are white scratches.
[0031]
Therefore, in the white flaw detection apparatus according to the present invention, the code string (0, ₁ , 0, 0, ₁ ) is stored in the memory in advance, and the output (a ₁ , a ₂ ) of the detection circuit shown in FIG. , A ₃ , a ₄ , a ₅ ) are compared and collated to determine whether the input television video signal is white or not.
[0032]
In the above, one embodiment of the white flaw detection circuit showing the white flaw of one pixel shown in FIG. 1A has been shown. However, the white flaw that is continuous for two or more pixels shown in FIG. 1B or FIG. The white defect detection circuit to be detected can also be configured following the configuration of FIG.
[0033]
Of the input signals of the white defect detection circuit described above, the threshold signal V _T (see FIG. 3) is preferably changed in accordance with the signal level of the input television video signal. This is because the level of white scratches usually generated in a CCD television camera is compressed as the level of the video signal increases when it passes through the camera's process circuit. This is so that white scratches can be detected without malfunction. More specifically, the level of the threshold signal V _T is reduced as the level of white scratches is reduced. Therefore, when the level of the threshold signal is expressed by k · v _T (V _T = k · v _T ), for example, the characteristics shown in FIG. 4 are set so that the value of k gradually decreases as the level of the input video signal increases.
[0034]
Next, consider the television video signal input to the white flaw detection circuit.
While the noise generated from white scratches is a constant level, the surrounding noise included in the television video signal is mainly random noise. By passing this through a cyclic filter , temporal random noise is reduced and white noise is reduced. Detection of white scratches in the scratch detection circuit becomes easier. In addition, the time resolution of the television video signal is added to the filter or frame to reduce the dynamic resolution, and the detection of white flaws is further facilitated.
[0035]
FIG. 5 shows an embodiment of a recursive filter .
In FIG. 5, first, an input television video signal is supplied to one of input terminals of an adder ADD. The output signal of the adder ADD is supplied to and stored in the frame memory 1 of one frame, but the signal read out every frame is added to the current input television video signal by the adder ADD and accumulated. The television video signal is stored again in the frame memory 1. In the illustrated embodiment, a memory control 2 for controlling the frame memory 1, a timing signal generator 3 connected to the memory control, and a stored number setting unit 4 are arranged so that a maximum of 256 video signals are stored. .
[0036]
Further, since the bit shifter 5 accumulates the video signal in the frame memory 1, the bit signal shifts the level (amplitude) of the read signal of the frame memory 1 back to the level of one video signal. belongs to. The shift amount of the bit shifter 5 is also controlled by the stored number setting unit 4. The output video signal of the bit shifter 5 is supplied to, for example, the input terminal of the white flaw detection circuit shown in FIG.
[0037]
In addition, as shown in FIG. 5, by setting the number of stored images to 2 ⁿ (n = 1, 2, 3,... 8), output video having the same level as the input signal level only by bit shift of the output video signal. Since the signal level is obtained and a special division circuit is not required, the circuit scale can be reduced. When the number of stored images is 256, random noise can be improved by 24 dB, and the S / N of about 40 dB of the image signal (pedestal part) of a normal high-definition camera is improved to about 60 dB to 64 dB, thereby detecting white scratches. It becomes easier. Further, according to this cyclic filter , in the case of a moving input television video signal, the dynamic resolution is also reduced, so that it is easier to detect white flaws.
[0038]
This is the end of the description of the white flaw detection circuit for television video signals. However, the entire apparatus detects white flaws (see FIG. 1) that are continuous in one or more pixels in the horizontal and vertical directions of the screen. It is necessary to provide a white scratch detection circuit in each direction independently.
[0039]
In addition, in order to correct the white defect detected as described above (here, the correction means interpolation of the video signal to bring it closer to a natural image without white defect). It is necessary to know the position (address) on the television screen where the white scratch is detected. For example, this counts the vertical and horizontal sync signals of the television signal, and outputs a pulse signal at the detected location. You can know by issuing.
[0040]
Next, a white defect correcting apparatus for a television video signal according to the present invention will be described.
Here again, in one embodiment described below, white scratches in the horizontal direction of the screen are interpolated using pixels (hereinafter simply referred to as interpolation), that is, interpolation is performed using the left and right television video signals.
[0041]
6, the pixel point P _{3 consecutive, P 4, P 5, P} 6, P 4 of the P _7, P _5, if P ₆ is white continuous pixels (white pixels over three pixels), the pixel Interpolation coefficients α ₄ , α ₅ , and α ₆ when the points are interpolated with weighted values of the pixel points P ₃ and P ₇ are shown. Here, FIG. 6A shows the case of linear interpolation, and FIG. 6B shows the case of curve interpolation. The interpolated values (signal levels) of the white scratches P ₄ , P ₅ , and P ₆ are as follows. It is expressed by equation (7).
P ₇ × α ₄ + P ₃ × (1-α ₄ )
P ₇ × α ₅ + P ₃ × (1-α ₅ ) (7)
P ₇ × α ₆ + P ₃ × (1-α ₆ )
[0042]
Accordingly, when white scratches P ₄ , P ₅ , and P ₆ are interpolated by linear interpolation, 0.25, 0.5 in FIG. 6A is added to α ₄ , α ₅ , and α ₆ in this equation (7). , 0.75, respectively, and when white scratches P ₄ , P ₅ , P ₆ are interpolated by curve interpolation, 0.125, 0.5, 0.875 in FIG. 6B are respectively substituted. By doing so, an interpolated video level is obtained. In addition, the shape of the curve in the case of curve interpolation is not limited to this, and is innumerable. Therefore, the interpolation coefficients α ₄ , α ₅ and α _{6 for} curve interpolation are not limited to 0.125, 0.5 and 0.875, respectively.
[0043]
FIG. 7 shows an embodiment of a white defect interpolation circuit that actually performs the above-described white defect interpolation.
7, an input television picture signal the shift register SR1, SR2, · · ·, are delayed by one pixel by SR4, to the shift register SR4 and that outputs the white flaw signal point in the P _4. At this point, the latch LATCH-1 and 2 input television video signals are sequentially supplied, it is assumed that each continues to hold the output signals P ₃ and P ₇ by the timing signals T ₁ and T _2.
[0044]
Further, as shown in the figure, 1-α ₄ , 1-α ₅ , 1-α ₆ and α ₄ are connected to the input terminals of the multipliers MULTI-1 and 2 where the held signals are not supplied. , Α ₅ , α ₆ are supplied, and the multiplication results are output according to the timing. Both multiplication results by the multipliers MULT-1 and 2 are supplied to the adder ADD and added. At the time point P ₄ when white scratches are output from the shift register SR4 as described above, the interpolation signal P ₃ × (1−α ₄ ) + P ₇ × α ₄ at that time is output (linear interpolation, curve The interpolation is different only in the value of the coefficient α, and the arithmetic expression shown by the equation (7) is the same, so the configuration of the interpolation circuit shown in FIG. In this case, tilting the switching contacts b side by controlling the changeover switch SW by the timing signal T _3, so that the interpolation signal is outputted. Note that the timing signals T ₁ , T ₂ , and T ₃ are obtained from an address signal indicating the position (address) of a white defect.
[0045]
Finally, FIG. 8 shows an embodiment in which the above-described television image signal white defect detection device and white defect correction device are integrated, which will be described.
The television video signal to which the present invention is applied is a high-definition signal in the form of a digital serial signal having a bit rate of 1.485 Gbps. In FIG. 8, an input signal written as an input television video signal is assumed to be a 10-bit parallel signal obtained by serial-parallel conversion of the high-definition signal. This parallel signal is supplied to the recursive filter 6 shown in FIG. 5 and the white defect interpolation circuit 7 shown in FIG. The former reduces random noise, reduces the motion resolution of moving images, and facilitates detection of white flaws in the white flaw detection circuits 8 and 9 (see FIG. 3) in the horizontal and vertical directions of the next stage. belongs to. Further, the latter white defect interpolation circuit 7 obtains a more natural television image output by linear interpolation or curve interpolation from the video level around the white defect portion.
[0046]
Further, the threshold signal is supplied to the threshold signal level control circuit 10 prior to being supplied to the white defect detection circuits 8 and 9 in the horizontal and vertical directions. Here, the level control as shown in FIG. Done. Specifically, as described above, when the level of the threshold signal is expressed by k · v _T (V _T = k · v _T ), the input video signal (in FIG. 8, the output signal of the cyclic filter 6). ), The value of k gradually decreases as the level increases. As described above, the white defect detection circuits 8 and 9 need to be provided independently in each of the horizontal and vertical directions.
[0047]
Further, when a white flaw is detected by any one of the detection circuits 8 and 9, the detection is transmitted to the address signal generation circuit 11 via an OR gate arranged on the output side thereof. In the address signal generation circuit 11, for example, an address signal indicating the address of the place where the white defect is detected is generated based on the vertical and horizontal synchronization signals of the television signal, and this is supplied to the white defect interpolation circuit 7. To do. In the interpolation circuit 7, white interpolation is replaced with interpolated level pixels by applying linear interpolation or curve interpolation, as described above.
[0048]
【The invention's effect】
According to the present invention, when the video level of a spot where white flaws have occurred is changed temporally (moving image), or from a video signal (still image) that does not change temporally but has a constant video level. The position of white scratches can be automatically detected. Further, according to the present invention, it is possible to easily detect the position of white scratches, which has been difficult until now, and by automatically performing a series of processes of white scratch detection and interpolation, it is possible to achieve accurate accuracy in a short time. High white scratch correction can be performed.
[0049]
Further, according to the present invention, the input television video signal is not directly supplied to the white defect detection circuit, but once the cyclic filter reduces the temporal random noise included in the television video signal, the dynamic resolution is reduced. As a result of lowering so that only white scratches are conspicuous, the white scratch detection operation in the white scratch detection circuit is made easier.
[0050]
Furthermore, the reduction of the level of the threshold signal supplied to the white flaw detection circuit in response to the increase in the level of the television video signal further ensures the operation of the white flaw detection device of the present invention.
[0051]
In the above, the detection and correction of white scratches generated in the CCD camera have been described. However, similar white scratches occur in a camera using a high-sensitivity HARP (High-gain Avalanche Rushing amorphous Photoconductor) imaging tube. The white scratches generated in this case are larger in area than the white scratches of the CCD camera, and the noise level due to the white scratches is also increased. The present invention effectively works against such white scratches.
[Brief description of the drawings]
FIG. 1 shows how white scratches appear on a raster.
FIG. 2 shows a noise pattern generated from white scratches as a waveform signal that changes with time.
FIG. 3 shows an embodiment of a white scratch detection circuit in the white scratch detection device of the present invention.
4 shows the characteristics of the threshold signal level supplied to the white flaw detection circuit shown in FIG. 3 with respect to the input video signal level.
FIG. 5 shows an embodiment of a recursive filter in the white flaw detection apparatus of the present invention.
FIG. 6 shows interpolation coefficients for linear interpolation and curve interpolation used in the white defect correcting apparatus of the present invention.
FIG. 7 shows an embodiment of a white defect interpolation circuit of the white defect correction apparatus of the present invention.
FIG. 8 shows an embodiment in which a white flaw detection device and a white flaw correction device of the present invention are configured integrally.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 1 Frame memory 2 Memory control 3 Timing signal generator 4 Accumulation number setting device 5 Bit shifter 6 Recursive filter 7 White defect interpolation circuit 8 Horizontal white defect detection circuit 9 Vertical white defect detection circuit 10 Threshold signal level control circuit 11 Address Signal generation circuit SUB subtracter ADD adder MULTI multiplier SR shift register LATCH latch

Claims (6)

Storage means for storing encoded white flaw signals of a plurality of pixels continuous in the horizontal and / or vertical direction of the television video signal, the television video signal supplied thereto, and the supplied television video signal Encoding means for encoding a television video signal in the horizontal and / or vertical direction according to the encoding technique of the encoding, and encoding of a plurality of continuous pixels read from the output of the encoding means and the storage means A white flaw detection apparatus for a television video signal, comprising comparison means for sequentially comparing codes of each white flaw signal. 2. The white flaw detection apparatus according to claim 1, wherein the encoding method includes: a level difference of signals before and after the one pixel and an absolute value of a level difference before and after the one pixel and / or the plurality of pixels. A white flaw detection device for a television video signal, characterized in that the method is performed based on a level difference between the level of the video signal and the level of the threshold signal. 3. The white scratch detection apparatus according to claim 2, wherein the level of the threshold signal is changed in accordance with the signal level of the supplied television video signal. Detection device. 4. The white scratch detection apparatus according to claim 1, wherein the supplied television video signal is transmitted to the encoding unit before the television video signal is supplied to the encoding unit. A white flaw detection apparatus for a television video signal, characterized in that random noise is reduced and dynamic resolution is reduced by a preliminary cyclic filter . 5. A television image signal white flaw detection device according to claim 1 is provided, and the left and right and / or upper and lower television image signals of the white flaw detected by the white flaw detection device are used. A white flaw correction device for a television video signal, further comprising a white flaw interpolation circuit for linearly or curvedly interpolating the white flaw. 6. The apparatus for correcting a white defect in a television video signal according to claim 5, wherein the linear or curved interpolation is performed such that a signal level of a pixel immediately before the white defect is Pi and a signal level of a pixel immediately after the pixel is Pk, and n signals therebetween. When there are continuous white pixels, Pk × α ₁ + Pi × (1−α ₁ )
Pk × α ₂ + Pi × (1−α ₂ )
・
・
・
Pk × α _n + Pi × (1−α _n )
(Where α ₁ , α ₂ ,... Α _n are interpolation coefficients)
A white defect correcting apparatus characterized in that sequential pixel interpolation is performed according to the level represented by

Images