JP4817552B2

JP4817552B2 - Phase difference detection method, phase difference detection device, distance measuring device, and imaging device

Info

Publication number: JP4817552B2
Application number: JP2001243917A
Authority: JP
Inventors: 義郎黒川
Original assignee: Seiko Precision Inc
Current assignee: Seiko Precision Inc
Priority date: 2001-08-10
Filing date: 2001-08-10
Publication date: 2011-11-16
Anticipated expiration: 2021-08-10
Also published as: JP2003057531A

Description

【０００１】
【発明の技術分野】
本発明は、位相差検出方法、位相差検出装置、測距装置および撮像装置に関する。
【０００２】
【従来の技術】
従来、自動焦点カメラにおいて、いわゆるパッシブ方式で被写体にピントを合わせる際、非ＴＴＬカメラの場合は撮影レンズを通過していない被写体像を用いて被写体距離を検出した後にその検出した被写体距離に応じて撮影レンズの位置を制御しており、ＴＴＬカメラの場合は撮影レンズを通過して得られた被写体像を用いて合焦状態からのずれ量を検出した後にその検出したずれ量に応じて撮影レンズの位置を制御している。以下、図３（ａ）を参照して上述した動作原理を説明する。
【０００３】
同図において、１対のレンズ１ａと１ｂは所定の基線長ｂだけ離して配設してあり、１対のレンズ１ａと１ｂから焦点距離ｆだけ離して配設された１対の光センサアレイ３ａと３ｂに互いに異なる光路ＡとＢを介して対象２の映像をそれぞれ結像させる。対象２は１対のレンズ１ａ、１ｂから正面方向に距離Ｌだけ離れた位置に存在するものとする。
【０００４】
対象２が無限遠の位置に存在するとき一対の光センサアレイ３ａと３ｂに結像される映像の中心は光センサアレイ３ａ、３ｂ上のレンズ１ａ、１ｂの光軸に対応する基準位置（３ａ１、３ｂ１）に結像されるが、対象２が無限遠位置よりも近づくと、これら基準位置（３ａ１、３ｂ１）からαだけずれた位置に結像される。三角測距の原理から対象２までの距離ＬはＬ＝ｂｆ／αとなる。ここで、基線長ｂと焦点距離ｆは定数なので、ずれ量αを検出すれば距離Ｌを測定できる。これが非ＴＴＬカメラに用いられているパッシブ測距（いわゆる外光三角測距）の原理である。なお、非ＴＴＬカメラでは測距装置の出力値として距離Ｌを用いるかわりにずれ量αをそのまま使う場合もある。本願では、このずれ量αを位相差と称する。
【０００５】
ＴＴＬカメラの場合は、撮像レンズ（図示せず。）を通過した光を一対のレンズ１ａ、１ｂに与えて上記と同様に左右の映像対間の位相差αを検出する。なお、この場合は合焦状態にあるときの映像の中心を各光センサアレイ３ａ、３ｂ上の基準位置とする。よって、位相差αの正負が前ピント状態か後ピント状態かをその絶対値が合焦からのずれの程度をそれぞれ示す。
【０００６】
これらはいずれも光学系により対象の映像を１対の光センサアレイ上に結像させ、１対の光センサアレイが出力する１対の映像信号の相対的な位置ずれ、すなわち位相差を１対の映像信号からそれぞれ抽出した部分映像データ群（図３（ｂ）参照）について相関演算を行うことにより検出している。なお、このような位相差検出は自動焦点カメラに限るものではなく、種々の測距装置や焦点検出装置等に用いることが可能である。
【０００７】
このような位相差を検出する方式において、被写体の背景に太陽等の強い光源がある場合などに発生するフレア等の迷光に起因する検出精度の悪化を低減する方式として、例えば特開平８−１６６２３７号公報に開示されたものがある。具体的には３つの技術が開示されており、第１の技術は１対の光センサが出力する１対のセンサデータの平均値が２つのセンサデータ間で差がなくなるように補正する技術、第２の技術は１対のセンサデータから抽出した部分映像データ群の組合せごとに両部分映像データ群間の代表値に差がなくなるように補正する技術、第３の技術は１対のセンサデータの各データ値をデータ変数番号に関する微分近似値で置き換えるように補正する技術である。
【０００８】
【発明が解決しようとする課題】
しかしながら、上記第１の技術すなわち１対の光センサが出力する１対のセンサデータの平均値が２つのセンサデータ間で差がなくなるように補正する場合は、相関演算を行う際に用いる部分映像データ群がその部分映像データ群に含まれないセンサデータ間の誤差の影響を受けてしまう。よって、例えば、補正しない状態で最も相関度の高くなる部分映像データ群の組合せがフレアによる影響を受けていなく、他の部分がフレアによる影響を受けている場合、補正後のほうが補正前より位相差検出精度が悪化してしまうという問題点を有していた。
【０００９】
また、上記第２の技術すなわち１対のセンサデータから抽出した部分映像データ群の組合せごとに両部分映像データ群間の代表値に差がなくなるように補正する場合は、部分映像データ群の組合せごとに補正量を求めなければならなくなり、部分映像データ群の組合せ数が多くなるほど処理動作が増大してしまうという問題点を有していた。
【００１０】
また、上記第３の技術すなわち１対のセンサデータの各データ値をデータ変数番号に関する微分近似値で置き換えるように補正する場合は、個々のデータに対して複雑な微分近似値置換え処理が必要となる。
【００１１】
本発明は、複雑な処理および検出精度の悪化を可及的に抑制可能な位相差検出方法、位相差検出装置、測距装置および撮像装置を提供することを目的とする。
【００１２】
【課題を解決するための手段】
第１の発明は、測定対象からの像が結像される１対の光センサアレイの出力に応じた複数の映像データからなる１対の映像データ列から相対位置が異なる部分映像データ群を抽出して１対の部分映像データ群の組合せを複数作り、当該１対の部分映像データ群の各組合せに対して複数の相関度を求める相関度取得ステップと、当該複数の相関度の中で最大相関度を示す上記１対の部分映像データ群から求められた最大相関度と所定値との差を上記複数の１対の部分映像データ群の一方の部分映像データ群の映像データの数で割って得られる補正値を求める補正値取得ステップと、上記複数の部分映像データ群の各組合せ毎に上記１対の部分映像データ群の映像データ列を上記補正値に基づき補正する補正ステップと、補正後の上記映像データに基づいて補正後の複数の相関度を求める相関度取得ステップと、補正前の最大相関度と補正後の最大相関度を比較し相関度の高い最大相関度を含む複数の相関度を有効相関度と判定する比較判定ステップと、上記有効相関度と判定された複数の相関度と相関度が求められた時の上記相対位置を基にして上記１対のセンサアレイに結像された像の相互間の位相差を検出する位相差検出ステップとを含む位相差検出方法である。このようなステップを含む方法によれば、最大相関度を示す１対の部分映像データ群の差に応じた補正値により１対のセンサアレイに結像された像の相互間の位相差を検出するために用いる映像データ間のデータ差を補正するので、もともと相関度が高い１対の部分映像データ群の差に基づいた補正が行え、もともと相関度が高くないデータの影響を受けることを低減でき、部分映像データ群の組合せごとに補正量を求めるという煩わしい処理を防止でき、個々のデータに対して複雑な微分近似値置換え処理をなくすことが可能となる。
【００１３】
第２の発明は、測定対象からの像が結像される１対のセンサアレイと、上記１対のセンサアレイの出力に応じた複数の映像データからなる１対の映像データ列から相対位置が異なる部分映像データ群を抽出して１対の部分映像データ群の組合せを複数作り、当該１対の部分映像データ群の各組合せに対して複数の相関度を求める相関演算部と、上記複数の相関度の中で最大相関度を示す上記１対の部分映像データ群から求められた最大相関度と所定値との差を上記複数の１対の部分映像データ群の一方の部分映像データ群の映像データの数で割って得られる補正値を求める補正値取得部と、上記複数の部分映像データ群の各組合せ毎に上記１対の部分映像データ群の映像データ列を上記補正値に基づき補正する補正部と、上記補正部により補正された上記映像データに基づいて補正後の複数の相関度を求める相関演算部と、補正前の最大相関度と補正後の最大相関度を比較し相関度の高い最大相関度を含む複数の相関度を有効相関度と判定する比較判定部と、上記有効相関度と判定された複数の相関度と相関度が求められた時の上記相対位置を基にして上記１対のセンサアレイに結像された像の相互間の位相差を検出する位相差検出部とを含む位相差検出装置である。このような構成によれば、最大相関度を示す１対の部分映像データ群の差に応じた補正値により１対のセンサアレイに結像された像の相互間の位相差を検出するために用いる映像データ間のデータ差を補正するので、もともと相関度が高い１対の部分映像データ群の差に基づいた補正が行え、もともと相関度が高くないデータの影響を受けることを低減でき、部分映像データ群の組合せごとに補正量を求めるという煩わしい処理を防止でき、個々のデータに対して複雑な微分近似値置換え処理をなくすことが可能となる。
【００１４】
第３の発明は、上記位相差検出装置と、上記位相差検出装置が検出する位相差に基づき上記測定対象までの距離に応じた距離データを求める距離検出部とを備えた測距装置である。かかる構成によれば、上記の効果に加え、距離データの精度が向上するので測距精度の低下が防止できる。
【００１５】
第４の発明は、上記位相差検出装置と、対物レンズと、上記対物レンズを通過した被写体像が結像される結像部と、上記位相差検出装置が求めた上記位相差に応じて上記対物レンズと上記結像部との間の合焦動作を行う合焦制御部とを含む撮像装置である。かかる構成によれば、上記の効果に加え、対物レンズの合焦精度が向上するので撮像画質の向上が図れる。
【００１６】
【発明の実施の形態】
以下、本発明の実施の形態を図面に示した一実施例を参照して説明する。図１は撮像装置に本発明を採用した例である。なお、図１において図３と同一構成のものには同一符号を附してある。
【００１７】
図１において、１対のレンズ１ａ、１ｂは上述したように対象２の映像を１対の光センサアレイ３ａ、３ｂ上にそれぞれ結像させる。光センサアレイ３ａ、３ｂはそれぞれ１６２個の画素（光電変換素子）をライン上に配置してあり、それらの各画素は当該画素上に結像された対象２の映像の光量に対応する電気信号を出力する。なお、１対の光センサアレイ３ａ、３ｂの画素の個数は適宜変更可能である。出力部４は１対の光センサアレイ３ａ、３ｂの出力をＣＰＵ５に出力する。ＣＰＵ５は入力する１対の光センサアレイ３ａ、３ｂの出力をメモリ部６に記憶してある種々の動作プログラムや各種のデータに基づき以下に示すように処理する。合焦制御部７はＣＰＵ５により制御され、対物レンズ８を両矢印Ｘ方向に動かし、対物レンズ８と結像部９との間の合焦動作を行う。なお、結像部９は銀塩フィルムでもよいし、いわゆるＣＣＤセンサやＣＭＯＳセンサ等のような光電変換素子を有する固体撮像素子でもよい。
【００１８】
次に、ＣＰＵ５の機能を中心に図１、図２を参照して動作を説明する。なお、図１ではＣＰＵ５が有する機能を説明するためＣＰＵ５においては機能ブロック図を採用している。
【００１９】
図示しないレリーズスイッチが操作されると、１対の光センサアレイ３ａ、３ｂが動作を開始する（ステップ３ａ）。上述したように１対の光センサアレイ３ａ、３ｂには１対のレンズ１ａ、１ｂにより対象２の映像が互いに異なる光路ＡとＢを介してそれぞれ結像されており、１対の光センサアレイ３ａ、３ｂから結像された映像の光量に対応する電気信号が出力される。
【００２０】
Ａ／Ｄ変換部５ａは出力部４を介して入力される１対の光センサアレイ３ａ、３ｂの出力をＡ／Ｄ変換する。メモリ部５ｂはＡ／Ｄ変換された１対の光センサアレイ３ａ、３ｂの出力を１対の映像データ列（ＩＬ、ＩＲ）として１対のメモリ領域５ｂＬ、５ｂＲに記憶する。本例ではＡ／Ｄ変換された光センサアレイ３ａの出力をメモリ領域５ｂＬに記憶し、Ａ／Ｄ変換された光センサアレイ３ｂの出力をメモリ領域５ｂＲに記憶する。また、本例では１対の光センサアレイ３ａ、３ｂの画素数がそれぞれ１６２なので、映像データ列（ＩＬ、ＩＲ）はそれぞれ１６２個のデータ（ＩＬ（１〜１６２）、ＩＲ（１〜１６２））で構成される。
【００２１】
左右差判定部５ｃはメモリ部５ｂ内のメモリ領域５ｂＬ、５ｂＲに記憶された１対の映像データ列（ＩＬ、ＩＲ）を読み出し（ステップ３ｂ）、それらの平均値の差もしくは最大値の差を求め、求めた値が所定値Ａ以上か否かを判定する（ステップ３ｃ）。すなわち、１対の映像データ列（ＩＬ、ＩＲ）の差を左右差とし、この左右差が所定値Ａ以上か否かを判定する。
【００２２】
左右差補正部５ｄは左右差判定部５ｃが１対の映像データ列（ＩＬ、ＩＲ）の平均値の差もしくは最大値の差が所定値Ａ以上と判定した場合、すなわち１対の映像データ列（ＩＬ、ＩＲ）の差が所定値Ａ以上あると判定した場合（ステップ３ｃ）、１対の映像データ列（ＩＬ、ＩＲ）の平均値の差もしくは最大値の差が所定値Ａより小さくなるように映像データ列の全てのデータを一律の補正量で補正し、補正したデータをメモリ領域５ｂＬ、５ｂＲに記憶し、１対の映像データ列（ＩＬ、ＩＲ）を更新する（ステップ３ｄ）。なお、左右差補正部５ｄによる補正は既に従来技術として説明した第１の技術とほぼ同様である。
【００２３】
ステップ３ｃにおいて１対の映像データ列（ＩＬ、ＩＲ）の差が所定値Ａより小さい場合もしくはステップ３ｄによる処理が終了したら、相関度取得部としての相関演算部５ｅによる相関演算を行う（ステップ３ｅ）。具体的な処理としては、１対の映像データ列（ＩＬ、ＩＲ）からそれぞれ部分映像データ群（ｉＬ、ｉＲ）をそれらの光センサアレイ上での相対位置が異なるように抽出し、抽出した各部分映像データ群（ｉＬ、ｉＲ）の組合せに対して相関度を求める。本例では、部分映像データ群のデータ数を２６とし、図３（ｂ）に示すように映像データ列（ＩＬ）から抽出する部分映像データ群（ｉＬ）を固定し、映像データ列（ＩＲ）から抽出する部分映像データ群（ｉＲ）を１つずつずらしていく方式を採用している。具体的には、以下の式（１）に基づき相関演算を行う。
【００２４】
【数１】

【００２５】
ステップ３ｅの相関演算が終了すると、最大相関度検出部５ｆが相関演算部５ｅが行った式（１）の演算結果に基づきＳ（ｌ）の極小値（以下、図３（ｂ）に示したＳ（ｘ）とする。）すなわち最大相関度を検出し、検出したＳ（ｘ）と相関演算関数Ｓ（ｌ）（ｌ＝１〜１３７の整数）とをメモリ部５ｇに格納する（ステップ３ｆ）。
【００２６】
メモリ部５ｇへの格納が終了すると、最大相関度判定部５ｈは最大相関度検出部５ｆが求めた極小値Ｓ（ｘ）が所定値Ｂ以上所定値Ｃ以下に含まれるか否かを判定し、１対の部分映像データ群がフレア等の迷光の影響を受けていないか判定する（ステップ３ｇ）。
【００２７】
ステップ３ｇにおいて、極小値Ｓ（ｘ）が所定値Ｂ以上所定値Ｃ以下に含まれておらず１対の部分映像データ群（ＩＬ、ＩＲ）がフレア等の迷光の影響を受けている可能性が高いと判定した場合、補正値取得部としての左右差補正量取得部５ｉは極小値Ｓ（ｘ）を部分映像データ群のデータ数（２６個）で割って１データあたりの誤差量を求め、これを左右差補正値とする（ステップ３ｈ）。
【００２８】
左右差補正値が求まると、補正部としての相関演算部５ｊは１対の映像データ列（ＩＬ、ＩＲ）からそれぞれ部分映像データ群（ｉＬ、ｉＲ）をそれらの相対位置が異なるように抽出し、抽出した各部分映像データ群（ｉＬ、ｉＲ）を左右差補正量取得部５ｉが求めた左右差補正値で補正する（ステップ３ｉ）。補正の例としては、例えば極小値Ｓ（ｘ）を示す部分映像データ群（ｉＬｘ、ｉＲｘ）の各データの総和を求めるとともにその総和の大小関係を判定し、部分映像データ群（ｉＬｘ）の総和が大きい場合、部分映像データ群（ｉＬ）を抽出した際にその部分映像データ群（ｉＬ）に含まれる各データから左右差補正値を減算する補正を行い、部分映像データ群（ｉＬｘ）の総和が部分映像データ群（ｉＲｘ）の総和より小さい場合、部分映像データ群（ｉＬ）を抽出した際にその部分映像データ群（ｉＬ）に含まれる各データから左右差補正値を加算する補正を行う。このような補正を行った後の部分映像データ群（ｉＬ）を用いて相関演算部５ｊは上述した式（１）の相関演算を行う（ステップ３ｊ）。
【００２９】
このように、最大相関度を示す１対の部分映像データ群の差に応じた補正値により部分映像データ群の各組合せごとに１対の部分映像データ群を補正し、補正後の１対の部分映像データ群の各組合せに対して再度相関演算を行って相関度を求めるので、補正前の段階で相関度が高い１対の部分映像データ群の差に基づいた補正が行え、補正に際して元来相関度が高くないデータの影響を受けることを低減でき、部分映像データ群の組合せごとに補正量を求めるという煩わしい処理を防止でき、個々のデータに対して複雑な微分近似値置換え処理をなくすことが可能となる。
【００３０】
ステップ３ｊによる相関演算が終了すると、最大相関度検出部５ｋは相関演算部５ｊが行った演算結果に基づきＳ（ｌ）の極小値（Ｓ（ｘ’）とする。）すなわち最大相関度を検出する。最大相関度検出部５ｋによりＳ（ｘ’）が検出されると、比較判定部５ｌはメモリ部５ｇに格納してあるＳ（ｘ）と最大相関度検出部５ｋが検出したＳ（ｘ’）の大小比較を行う（ステップ３ｋ）。この動作について補足すると、迷光は迷光の元となる強い光源の形状やその光源と対象２との位置関係等により複雑に変わることが考えられ、迷光の状況により上述した補正で迷光の影響が低減される場合とされない場合が生じる可能性がある。よって、ステップ３ｋは上述した補正で迷光の影響が低減されるか否かを確認するために実行される。
【００３１】
比較判定部５ｌはＳ（ｘ）とＳ（ｘ’）のうち小さい方すなわち相関度の高い方を有効データとする（ステップ３ｌ、３ｍ）。具体的には、Ｓ（ｘ）がＳ（ｘ’）より小さい場合、ステップ３ｉによる補正前のＳ（ｘ）と相関演算関数Ｓ（ｌ）（ｌ＝１〜１３７の整数）をメモリ部５ｇから読み出し補間演算部５ｍに出力し、Ｓ（ｘ’）がＳ（ｘ）より小さい場合、最大相関度検出部５ｋが用いたステップ３ｉによる補正後のＳ（ｘ’）とその相関演算関数Ｓ（ｌ）（ｌ＝１〜１３７の整数）を読み出し補間演算部５ｍに出力する。
【００３２】
このように、補正前の最大相関度と補正後の最大相関度を比較し、補正により最大相関度が悪化した場合は、補正前の最大相関度を有効データとするので、複雑な迷光による影響を補正により除去できなかった場合、その補正を無効にできる。よって、効果の上がらない補正による位相差検出精度の悪化を防止できる。
【００３３】
補間演算部５ｍは入力する極小値（Ｓ（ｘ’）またはＳ（ｘ））とその前後の相関演算関数値（Ｓ（ｘ’−１）とＳ（ｘ’＋１））またはＳ（ｘ−１）とＳ（ｘ＋１））等を用いた補間法によりｘ’またはｘを補正する（ステップ３ｎ）。この補間演算は公知の技術であるので、詳細な説明は割愛する。
【００３４】
補間演算によりｘ’またはｘが補正されると、位相差検出部５ｎは補正されたｘ’またはｘの光センサ３ｂ側に設定してある基準位置（例えば、非ＴＴＬカメラのような外光三角測距の場合は測定方向における無限遠位置の対象の映像の中心位置に対応する位置とし、ＴＴＬカメラ等で用いられる焦点検出装置の場合は撮影レンズが合焦状態にあるときの対象の映像の中心位置に対応する位置とする。）からのずれ量すなわち位相差を検出する（ステップ３ｏ）。
【００３５】
合焦制御部７は位相差検出部５ｎが検出した位相差に基づき対物レンズ８の位置を制御し、対物レンズ８と結像部９との間の合焦動作を行う。なお、非ＴＴＬカメラの場合は、上記に限らず位相差検出部５ｎで検出した位相差に基づき距離検出部５ｏで対象２までの距離データを求め、この距離データに基づき合焦制御部７が対物レンズ８の位置を制御し、対物レンズ８と結像部９との間の合焦動作を行うようにしてもよい。
【００３６】
なお、上記では相関演算を行う際に一方の部分映像データ群（ｉＬ）を固定し、他方の部分映像データ群（ｉＲ）を１つずつずらしていく例を示したが、相関演算の方式は上記に限らず適宜変更可能である。例えば、従来技術として示した特開平８−１６６２３７号公報に開示されているように両方の部分映像データ群をそれぞれの相対位置が異なるように順次ずらすようにしてもよい。
【００３７】
また、上記では各映像データ列（ＩＬ、ＩＲ）のデータ数を１６２とし、部分映像データ群のデータ数を２６としたが、これらも適宜変更可能である。
【００３８】
また、上記では撮像装置に本発明を採用した例を示したが、撮像装置に限るものではない。例えば、種々の測距装置や焦点検出装置等に用いることが可能である。
【００３９】
【発明の効果】
本発明によれば、最大相関度を示す１対の部分映像データ群の差に応じた補正値により１対のセンサアレイに結像された像の相互間の位相差を検出するために用いる映像データ間のデータ差を補正するので、もともと相関度が高い１対の部分映像データ群の差に基づいた補正が行え、もともと相関度が高くないデータの影響を受けることを低減でき、部分映像データ群の組合せごとに補正量を求めるという煩わしい処理を防止でき、個々のデータに対して複雑な微分近似値置換え処理をなくすことが可能となる。
【図面の簡単な説明】
【図１】本発明の一実施例を示したブロック回路図。
【図２】図１の動作説明のためのフローチャート。
【図３】図１の動作説明のための説明図。
【符号の説明】
３ａ光センサアレイ
３ｂ光センサアレイ
５ｅ相関度取得部
５ｉ補正値取得部
５ｊ補正部
５ｎ位相差検出部
５ｏ距離検出部
７合焦制御部
８対物レンズ
９結像部[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to a phase difference detection method, a phase difference detection device, a distance measuring device, and an imaging device.
[0002]
[Prior art]
Conventionally, when focusing on a subject in a so-called passive method in an autofocus camera, in the case of a non-TTL camera, the subject distance is detected using a subject image that has not passed through the taking lens, and then the subject distance is detected. The position of the photographic lens is controlled. In the case of a TTL camera, the amount of deviation from the in-focus state is detected using the subject image obtained by passing through the photographic lens, and then the photographic lens is selected according to the detected amount of deviation. Is controlling the position. Hereinafter, the operation principle described above will be described with reference to FIG.
[0003]
In the figure, a pair of lenses 1a and 1b are arranged apart from each other by a predetermined base line length b, and a pair of photosensor arrays arranged away from the pair of lenses 1a and 1b by a focal length f. The images of the object 2 are formed on the

light beams

3a and 3b through different optical paths A and B, respectively. The object 2 is assumed to be present at a position away from the pair of lenses 1a and 1b by a distance L in the front direction.
[0004]
When the object 2 exists at an infinite position, the center of the image formed on the pair of

photosensor arrays

3a and 3b is the reference position (3a1) corresponding to the optical axis of the lenses 1a and 1b on the

photosensor arrays

3a and 3b. 3b1), but when the object 2 is closer to the infinity position, the image is formed at a position shifted by α from these reference positions (3a1, 3b1). From the principle of triangulation, the distance L to the object 2 is L = bf / α. Here, since the base line length b and the focal length f are constants, the distance L can be measured by detecting the shift amount α. This is the principle of passive ranging (so-called external light triangular ranging) used in non-TTL cameras. In a non-TTL camera, the shift amount α may be used as it is instead of using the distance L as the output value of the distance measuring device. In the present application, this shift amount α is referred to as a phase difference.
[0005]
In the case of a TTL camera, light passing through an imaging lens (not shown) is applied to the pair of lenses 1a and 1b, and the phase difference α between the left and right image pairs is detected in the same manner as described above. In this case, the center of the image when in focus is set as a reference position on each of the

optical sensor arrays

3a and 3b. Therefore, whether the phase difference α is positive or negative is indicated by the absolute value, and the degree of deviation from the in-focus state.
[0006]
Both of these images form an image of an object on a pair of photosensor arrays by an optical system, and a pair of image signals output from the pair of photosensor arrays is set to have a relative positional shift, that is, a phase difference. Are detected by performing a correlation operation on partial video data groups (see FIG. 3B) extracted from the video signals respectively. Such phase difference detection is not limited to an autofocus camera, and can be used for various distance measuring devices, focus detection devices, and the like.
[0007]
In such a system for detecting a phase difference, as a system for reducing deterioration in detection accuracy caused by stray light such as flare that occurs when a strong light source such as the sun is present in the background of the subject, for example, JP-A-8-166237 is disclosed. There are those disclosed in the Gazette. Specifically, three techniques are disclosed, and the first technique is a technique for correcting an average value of a pair of sensor data output by a pair of optical sensors so that there is no difference between the two sensor data. The second technique is a technique for correcting a combination of partial video data groups extracted from a pair of sensor data so that there is no difference in representative values between the two partial video data groups, and the third technique is a pair of sensor data. In this technique, each data value is corrected so as to be replaced with a differential approximation value related to the data variable number.
[0008]
[Problems to be solved by the invention]
However, in the case of correcting the first technique, that is, the average value of the pair of sensor data output from the pair of photosensors so that there is no difference between the two sensor data, the partial image used when performing the correlation calculation The data group is affected by an error between sensor data not included in the partial video data group. Thus, for example, if the combination of partial video data groups that have the highest degree of correlation in the uncorrected state is not affected by flare and the other part is affected by flare, the result after correction is higher than that before correction. There was a problem that the phase difference detection accuracy deteriorated.
[0009]
When correction is made so that there is no difference in the representative values between the two partial video data groups for each combination of the partial video data groups extracted from the pair of sensor data, the combination of partial video data groups The correction amount has to be obtained every time, and the processing operation increases as the number of combinations of partial video data groups increases.
[0010]
Further, when the third technique, that is, to correct each data value of a pair of sensor data with a differential approximation value related to a data variable number, complicated differential approximation value replacement processing is required for each piece of data. Become.
[0011]
It is an object of the present invention to provide a phase difference detection method, a phase difference detection device, a distance measuring device, and an imaging device that can suppress complicated processing and deterioration in detection accuracy as much as possible.
[0012]
[Means for Solving the Problems]
1st invention extracts the partial video data group from which a relative position differs from one pair of video data sequence which consists of several video data according to the output of a pair of optical sensor array in which the image from a measuring object is formed A plurality of combinations of a pair of partial video data groups and obtaining a plurality of correlations for each combination of the pair of partial video data groups, and a maximum of the plurality of correlations The difference between the maximum correlation obtained from the pair of partial video data groups indicating the correlation and a predetermined value is divided by the number of video data in one partial video data group of the plurality of pairs of partial video data groups. A correction value obtaining step for obtaining a correction value obtained by the correction, a correction step for correcting the video data string of the pair of partial video data groups for each combination of the plurality of partial video data groups based on the correction values, Later video data A correlation degree obtaining step of obtaining a plurality of correlation corrected based, effective correlation a plurality of correlation including a high maximum degree of correlation compared correlation maximum correlation and the corrected maximum correlation before correction A plurality of correlation levels determined as the effective correlation level and the relative positions when the correlation levels are obtained, and the images formed on the pair of sensor arrays are mutually correlated. A phase difference detection method including a phase difference detection step of detecting a phase difference therebetween. According to the method including such steps, the phase difference between the images formed on the pair of sensor arrays is detected by the correction value corresponding to the difference between the pair of partial video data groups indicating the maximum correlation. Because the data difference between the video data used for the correction is corrected, the correction based on the difference between the pair of partial video data groups having a high correlation degree can be performed, and the influence of the data having a low correlation degree is reduced. In addition, it is possible to prevent a troublesome process of obtaining a correction amount for each combination of partial video data groups, and it is possible to eliminate complicated differential approximation replacement processing for individual data.
[0013]
According to a second aspect of the present invention, a relative position is determined from a pair of sensor arrays on which an image from a measurement target is formed, and a pair of video data strings composed of a plurality of video data corresponding to the outputs of the pair of sensor arrays. A plurality of combinations of a pair of partial video data groups are extracted by extracting different partial video data groups, and a correlation calculation unit for obtaining a plurality of correlations for each combination of the pair of partial video data groups ; The difference between the maximum correlation degree obtained from the pair of partial video data groups indicating the maximum correlation degree among the correlation degrees and a predetermined value is calculated for one partial video data group of the plurality of pairs of partial video data groups. A correction value acquisition unit for obtaining a correction value obtained by dividing by the number of video data, and correcting the video data string of the pair of partial video data groups based on the correction values for each combination of the plurality of partial video data groups Correction unit and the correction unit A plurality of correlation including the the the the video data on the basis of the correlation calculation unit for obtaining a plurality of correlation degree after the correction, the maximum correlation between the maximum correlation comparator correlated high degree of maximum correlation to the corrected before correction A comparison / determination unit that determines the degree as an effective degree of correlation, and a plurality of correlation degrees determined as the effective degree of correlation and an image on the pair of sensor arrays based on the relative position when the degree of correlation is obtained. And a phase difference detection unit that detects a phase difference between the detected images. According to such a configuration, in order to detect the phase difference between the images formed on the pair of sensor arrays by the correction value corresponding to the difference between the pair of partial video data groups indicating the maximum correlation. Since the data difference between the video data to be used is corrected, correction based on the difference between a pair of partial video data groups having a high degree of correlation can be performed, and the influence of data having a high degree of correlation can be reduced. The troublesome process of obtaining the correction amount for each combination of the video data groups can be prevented, and the complicated differential approximation replacement process can be eliminated for each piece of data.
[0014]
A third aspect of the invention is a distance measuring device including the phase difference detection device and a distance detection unit that obtains distance data corresponding to the distance to the measurement object based on the phase difference detected by the phase difference detection device. . According to such a configuration, in addition to the above effects, the accuracy of the distance data is improved, so that a decrease in the distance measurement accuracy can be prevented.
[0015]
According to a fourth aspect of the present invention, there is provided the phase difference detection device, the objective lens, an imaging unit on which a subject image that has passed through the objective lens is formed, and the phase difference determined by the phase difference detection device. The imaging apparatus includes a focusing control unit that performs a focusing operation between the objective lens and the imaging unit. According to such a configuration, in addition to the above effects, the focusing accuracy of the objective lens is improved, so that it is possible to improve the captured image quality.
[0016]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described with reference to an example shown in the drawings. FIG. 1 shows an example in which the present invention is applied to an imaging apparatus. In FIG. 1, the same components as those in FIG. 3 are denoted by the same reference numerals.
[0017]
In FIG. 1, as described above, the pair of lenses 1a and 1b form an image of the object 2 on the pair of

photosensor arrays

3a and 3b, respectively. Each of the

optical sensor arrays

3a and 3b has 162 pixels (photoelectric conversion elements) arranged on a line, and each of these pixels has an electric signal corresponding to the light amount of the image of the object 2 formed on the pixel. Is output. The number of pixels of the pair of

photosensor arrays

3a and 3b can be changed as appropriate. The output unit 4 outputs the output of the pair of

photosensor arrays

3a and 3b to the CPU 5. The CPU 5 processes the input outputs of the pair of

photosensor arrays

3a and 3b based on various operation programs and various data stored in the memory unit 6 as described below. The focusing control unit 7 is controlled by the CPU 5 and moves the objective lens 8 in the direction of the double arrow X to perform a focusing operation between the objective lens 8 and the imaging unit 9. The imaging unit 9 may be a silver salt film or a solid-state imaging device having a photoelectric conversion element such as a so-called CCD sensor or CMOS sensor.
[0018]
Next, the operation will be described with reference to FIGS. In FIG. 1, a functional block diagram is adopted for the CPU 5 in order to explain the functions of the CPU 5.
[0019]
When a release switch (not shown) is operated, the pair of

photosensor arrays

3a and 3b starts operating (step 3a). As described above, the image of the object 2 is formed on the pair of

optical sensor arrays

3a and 3b by the pair of lenses 1a and 1b via the different optical paths A and B, respectively. An electrical signal corresponding to the light quantity of the image formed from 3a and 3b is output.
[0020]
The A / D converter 5a performs A / D conversion on the outputs of the pair of

photosensor arrays

3a and 3b input via the output unit 4. The memory unit 5b stores the outputs of the pair of

photosensor arrays

3a and 3b subjected to A / D conversion in the pair of memory areas 5bL and 5bR as a pair of video data strings (IL and IR). In this example, the output of the A / D converted photosensor array 3a is stored in the memory area 5bL, and the output of the A / D converted photosensor array 3b is stored in the memory area 5bR. In this example, since the number of pixels of the pair of

photosensor arrays

3a and 3b is 162, the video data string (IL, IR) has 162 pieces of data (IL (1-162), IR (1-162), respectively. ).
[0021]
The left / right difference determination unit 5c reads a pair of video data strings (IL, IR) stored in the memory areas 5bL and 5bR in the memory unit 5b (step 3b), and calculates the difference between the average values or the difference between the maximum values. It is determined and it is determined whether or not the determined value is equal to or greater than a predetermined value A (step 3c). That is, the difference between the pair of video data strings (IL, IR) is defined as a left / right difference, and it is determined whether the left / right difference is equal to or greater than a predetermined value A.
[0022]
The left / right difference correction unit 5d determines that the difference between the average values or the maximum values of the pair of video data sequences (IL, IR) is equal to or greater than a predetermined value A, that is, a pair of video data sequences. When it is determined that the difference between (IL, IR) is greater than or equal to the predetermined value A (step 3c), the difference between the average values or the maximum value of the pair of video data strings (IL, IR) is smaller than the predetermined value A. As described above, all the data in the video data string is corrected with a uniform correction amount, the corrected data is stored in the memory areas 5bL and 5bR, and the pair of video data strings (IL, IR) is updated (step 3d). The correction by the left / right difference correction unit 5d is almost the same as the first technique already described as the conventional technique.
[0023]
When the difference between the pair of video data strings (IL, IR) is smaller than the predetermined value A in step 3c or when the processing in step 3d is completed, the correlation calculation is performed by the correlation calculation unit 5e as the correlation degree acquisition unit (step 3e). ). Specifically, partial video data groups (iL, iR) are extracted from a pair of video data strings (IL, IR) such that their relative positions on the photosensor array are different, and each extracted The degree of correlation is obtained for the combination of partial video data groups (iL, iR). In this example, the number of data in the partial video data group is 26, and as shown in FIG. 3B, the partial video data group (iL) extracted from the video data string (IL) is fixed, and the video data string (IR) A system is adopted in which partial video data groups (iR) extracted from are shifted one by one. Specifically, the correlation calculation is performed based on the following formula (1).
[0024]
[Expression 1]

[0025]
When the correlation calculation of step 3e is completed, the maximum correlation degree detection unit 5f is based on the calculation result of the equation (1) performed by the correlation calculation unit 5e (hereinafter, shown in FIG. 3B). S (x).) That is, the maximum correlation is detected, and the detected S (x) and the correlation operation function S (l) (l = 1 to 137) are stored in the memory unit 5g (step 3f). ).
[0026]
When the storage in the memory unit 5g ends, the maximum correlation degree determination unit 5h determines whether or not the minimum value S (x) obtained by the maximum correlation degree detection unit 5f is included in a predetermined value B or more and a predetermined value C or less. It is determined whether or not the pair of partial video data groups is affected by stray light such as flare (step 3g).
[0027]
In step 3g, there is a possibility that the minimum value S (x) is not included between the predetermined value B and the predetermined value C, and the pair of partial video data groups (IL, IR) are affected by stray light such as flare. When it is determined that the difference is high, the left-right difference correction amount acquisition unit 5i as the correction value acquisition unit divides the minimum value S (x) by the number of data of the partial video data group (26) to obtain an error amount per data. This is used as the left / right difference correction value (step 3h).
[0028]
When the left-right difference correction value is obtained, the correlation calculation unit 5j as the correction unit extracts the partial video data groups (iL, iR) from the pair of video data strings (IL, IR) so that their relative positions are different. Then, each extracted partial video data group (iL, iR) is corrected with the left / right difference correction value obtained by the left / right difference correction amount acquisition unit 5i (step 3i). As an example of the correction, for example, the sum of each data of the partial video data group (iLx, iRx) indicating the minimum value S (x) is obtained and the magnitude relation of the sum is determined, and the sum of the partial video data group (iLx) is determined. When the partial video data group (iL) is extracted, correction for subtracting the right / left difference correction value from each data included in the partial video data group (iL) is performed, and the sum of the partial video data group (iLx) is obtained. Is smaller than the sum of the partial video data group (iRx), when the partial video data group (iL) is extracted, correction is performed by adding a left / right difference correction value from each data included in the partial video data group (iL). . Using the partial video data group (iL) after such correction, the correlation calculation unit 5j performs the correlation calculation of the above-described equation (1) (step 3j).
[0029]
In this way, a pair of partial video data groups is corrected for each combination of the partial video data groups with a correction value corresponding to the difference between the pair of partial video data groups indicating the maximum degree of correlation, and the corrected pair of partial video data groups is corrected. Since the correlation calculation is performed again for each combination of the partial video data groups to obtain the correlation level, correction based on the difference between a pair of partial video data groups having a high correlation level can be performed before the correction. It is possible to reduce the influence of data that does not have a high degree of correlation, to avoid the troublesome process of calculating the correction amount for each combination of partial video data groups, and to eliminate complicated differential approximation replacement processing for individual data It becomes possible.
[0030]
When the correlation calculation in step 3j is completed, the maximum correlation degree detection unit 5k detects the minimum value of S (l) (S (x ′)), that is, the maximum correlation degree, based on the calculation result performed by the correlation calculation unit 5j. To do. When S (x ′) is detected by the maximum correlation degree detection unit 5k, the comparison / determination unit 5l detects S (x) stored in the memory unit 5g and S (x ′) detected by the maximum correlation degree detection unit 5k. Are compared (step 3k). Supplementing this operation, stray light may be complicatedly changed depending on the shape of the strong light source that is the source of stray light, the positional relationship between the light source and the target 2, and the effect of stray light is reduced by the correction described above depending on the situation of stray light. There are cases where it may or may not be. Therefore, step 3k is executed to confirm whether or not the influence of stray light is reduced by the correction described above.
[0031]
The comparison / determination unit 5l uses the smaller one of S (x) and S (x ′), that is, the one having a higher degree of correlation as effective data (steps 3l and 3m). Specifically, when S (x) is smaller than S (x ′), S (x) before correction in step 3i and the correlation operation function S (l) (l = 1 to 137) are stored in the memory unit 5g. When S (x ′) is smaller than S (x), the S (x ′) corrected by step 3i used by the maximum correlation degree detection unit 5k and its correlation calculation function S are output. (L) (l = 1 to 137) is read and output to the interpolation calculation unit 5m.
[0032]
In this way, the maximum correlation before correction is compared with the maximum correlation after correction, and if the maximum correlation deteriorates due to correction, the maximum correlation before correction is used as valid data. If the correction cannot be removed, the correction can be invalidated. Therefore, it is possible to prevent deterioration of the phase difference detection accuracy due to correction that does not increase the effect.
[0033]
The interpolation calculation unit 5m inputs the minimum value (S (x ′) or S (x)) and the correlation calculation function values (S (x′−1) and S (x ′ + 1)) or S (x−) before and after the input. X ′ or x is corrected by an interpolation method using 1) and S (x + 1)) (step 3n). Since this interpolation calculation is a known technique, a detailed description thereof is omitted.
[0034]
When x ′ or x is corrected by the interpolation calculation, the phase difference detection unit 5n sets the corrected x ′ or x reference position set on the optical sensor 3b side (for example, an external light triangle such as a non-TTL camera). In the case of distance measurement, the position corresponds to the center position of the target image at an infinite position in the measurement direction. In the case of a focus detection device used in a TTL camera or the like, the target image when the photographing lens is in focus is used. The amount of deviation from the center position, that is, the phase difference is detected (step 3o).
[0035]
The focusing control unit 7 controls the position of the objective lens 8 based on the phase difference detected by the phase difference detection unit 5n, and performs a focusing operation between the objective lens 8 and the imaging unit 9. In the case of a non-TTL camera, not limited to the above, the distance detection unit 5o obtains distance data to the object 2 based on the phase difference detected by the phase difference detection unit 5n, and the focus control unit 7 determines the distance data based on the distance data. The position of the objective lens 8 may be controlled to perform a focusing operation between the objective lens 8 and the imaging unit 9.
[0036]
In the above, an example is shown in which one partial video data group (iL) is fixed and the other partial video data group (iR) is shifted one by one when performing correlation calculation. Not limited to the above, it can be changed as appropriate. For example, as disclosed in Japanese Patent Laid-Open No. 8-166237 shown as the prior art, both partial video data groups may be sequentially shifted so that their relative positions are different.
[0037]
In the above description, the number of data of each video data string (IL, IR) is 162, and the number of data of the partial video data group is 26. However, these can be changed as appropriate.
[0038]
Moreover, although the example which employ | adopted this invention for the imaging device was shown above, it is not restricted to an imaging device. For example, it can be used for various distance measuring devices, focus detection devices, and the like.
[0039]
【The invention's effect】
According to the present invention, an image used for detecting a phase difference between images formed on a pair of sensor arrays with a correction value corresponding to a difference between a pair of partial image data groups indicating the maximum correlation. Since the data difference between the data is corrected, correction based on the difference between a pair of partial video data groups with a high degree of correlation can be performed, and the influence of data that does not have a high degree of correlation can be reduced. The troublesome process of obtaining the correction amount for each group combination can be prevented, and the complicated differential approximation replacement process can be eliminated for individual data.
[Brief description of the drawings]
FIG. 1 is a block circuit diagram showing an embodiment of the present invention.
FIG. 2 is a flowchart for explaining the operation of FIG. 1;
FIG. 3 is an explanatory diagram for explaining the operation of FIG. 1;
[Explanation of symbols]
3a Photosensor array 3b Photosensor array 5e Correlation degree acquisition unit 5i Correction value acquisition unit 5j Correction unit 5n Phase difference detection unit 5o Distance detection unit 7 Focus control unit 8 Objective lens 9 Imaging unit

Claims

A partial video data group having a different relative position is extracted from a pair of video data sequences composed of a plurality of video data corresponding to the output of a pair of photosensor arrays on which an image from a measurement object is formed, and a pair of portions A plurality of combinations of video data groups, and a correlation degree obtaining step for obtaining a plurality of correlation degrees for each combination of the pair of partial video data groups;
The difference between the maximum correlation obtained from the pair of partial video data groups showing the maximum correlation among the plurality of correlations and a predetermined value is determined as one partial video of the plurality of pairs of partial video data groups. A correction value acquisition step for obtaining a correction value obtained by dividing by the number of video data in the data group ;
A correction step of correcting the video data string of the pair of partial video data groups based on the correction values for each combination of the plurality of partial video data groups ;
A correlation degree obtaining step for obtaining a plurality of correlation degrees after correction based on the corrected video data ;
A comparison determination step of comparing the maximum correlation degree before correction with the maximum correlation degree after correction and determining a plurality of correlation degrees including a maximum correlation degree with a high correlation degree as an effective correlation degree;
A plurality of correlation levels determined as the effective correlation level and a phase difference between the images formed on the pair of sensor arrays based on the relative positions when the correlation levels are obtained. A phase difference detection method comprising: a phase difference detection step.

A pair of sensor arrays on which an image from the measurement object is formed;
Extracting partial video data groups having different relative positions from a pair of video data sequences consisting of a plurality of video data according to the output of the pair of sensor arrays to create a plurality of combinations of a pair of partial video data groups, A correlation calculation unit for obtaining a plurality of correlations for each combination of a pair of partial video data groups;
The difference between the maximum correlation obtained from the pair of partial video data groups indicating the maximum correlation among the plurality of correlations and a predetermined value is determined as one partial video of the plurality of pairs of partial video data groups. A correction value acquisition unit for obtaining a correction value obtained by dividing by the number of video data in the data group ;
A correction unit that corrects the video data string of the pair of partial video data groups for each combination of the plurality of partial video data groups based on the correction values;
A correlation calculation unit for obtaining a plurality of corrected degrees of correlation based on the video data corrected by the correction unit ;
A comparison determination unit that compares the maximum correlation degree before correction with the maximum correlation degree after correction and determines a plurality of correlation degrees including a maximum correlation degree with a high correlation degree as an effective correlation degree;
A plurality of correlation levels determined as the effective correlation level and a phase difference between the images formed on the pair of sensor arrays based on the relative positions when the correlation levels are obtained. A phase difference detection device comprising: a phase difference detection unit.

A phase difference detection device according to claim 2, and a distance detection unit that obtains distance data according to the distance to the measurement object based on the phase difference detected by the phase difference detection device. Distance device.

The phase difference detection device according to claim 2, an objective lens, an imaging unit on which a subject image that has passed through the objective lens is formed, and the objective according to the phase difference obtained by the phase difference detection device An imaging apparatus comprising: a focusing control unit that performs a focusing operation between a lens and the imaging unit.