JP3727400B2 - Crossing detection device - Google Patents

Description

一方、図１において、前記撮像装置１により得られた画像情報は、右カメラ画像保存メモリＭ１及び左カメラ画像保存メモリＭ２に保存される。両メモリＭ１，Ｍ２の保存データは、距離算出回路２に入力される。距離算出回路２では、縦８ドット、横８ドットを単位領域として画像がメッシュ状に分割され、右画像データを基準データ、左画像データを参照データとした相関計算により視差ｓ〔ドット〕が算出される。そして、該算出された視差ｓを前記（１）式に代入すれば画像内の距離分布が求められる。この距離分布データは距離データ保存メモリＭ３に保存される。
【００１８】
また、解析データ算出回路３は、前記右カメラ画像保存メモリＭ１に保存されている右カメラの画像データを取り込み、単位領域の縦方向の８ドット分のデータの総和（これを解析データという）を算出する。そして、解析データ算出回路３にて算出された最新の解析データは、「本サイクル解析データ」として本サイクル解析データ保存メモリＭ４に保存される。このとき、解析データは、８ドット分のデータの総和で保存されるため、データ容量は１／８に削減される。解析データ算出回路３は、上記解析データの算出処理を６６ｍｓサイクルで実施する。また、本サイクル解析データ保存メモリＭ４に保存された解析データは、６６ｍｓ毎に「前サイクル解析データ」として前サイクル解析データ保存メモリＭ５に転送される。従って、メモリＭ４，Ｍ５には、６６ｍｓ間隔の前後サイクルの解析データが保存されることとなる。
【００１９】
ここで、前記解析データ算出回路３による解析領域は、画面上において例えば図４のように設定されている。即ち、本実施の形態では、自車両の前方３０ｍ〜６０ｍの範囲に存在する横断者を検出することを目的としており、より具体的には、５０ｍ前方で地上に立っている横断者を基準として当該解析領域が設定されている。図４の画面上では、水平方向に延びる６０ドット幅の垂直領域が解析領域として設定されている。
【００２０】
また、移動量算出回路４は、前記本サイクル解析データ保存メモリＭ４及び前サイクル解析データ保存メモリＭ５に保存されている解析データを取り込み、両解析データ間の単位領域毎の画像のずれから横断者の移動量を算出する。ここでは、計算量削減のため、横方向（水平走差線方向）のみの探索を行う。つまり、横断者が移動する最大範囲を机上検討して相関計算の探索範囲を設定し、この探索範囲内で、前サイクル解析データを１ドットずつずらしながら差の絶対値の総和（相関値Ｓ（ｚ））を算出する。具体的には、次の（２）式を用いて単位領域についての相関値Ｓ（ｚ）を算出する。
【００２１】
【数２】

但し、「Ｄ0 」は本サイクル解析データ、「Ｄ1 」は前サイクル解析データ、「ｘ0 」は画像の左上座標（相関計算の探索開始点）、「ｗ」は画像内における検索範囲の横幅〔ドット〕、「ｚ」はずらし量〔ドット〕である。なお、図５は、前サイクル時と本サイクル時との間に、横断者が水平右方向に所定量だけ移動した状態を示している。
【００２２】
上記の如く求めた相関値Ｓ（ｚ）の最小値について、その時のずらし量が横断者の移動量に相当する。このとき、画像上で、横断者の移動量が「０」のとき、つまり、ずらし量が「０」のとき、横断者と衝突するケースに相当する。これは、自車両の進行方向と横断者の角度が変化しないとき、横断者との距離が０ｍになったとき衝突するためである。図６は、横断者が水平右方向に移動した場合の相関値Ｓ（ｚ）とずらし量との関係を示しており、かかる場合にはずらし量＝ｋが移動量に相当する。この際、精度向上のために、相関値の最小値とそれに隣合う２つの相関値を用いて補間計算を行い、１０分の１ドット精度で移動量を算出するようにしてもよい。
【００２３】
このようにして移動量算出回路４では、６６ｍｓ毎で取り込まれた時間的に連続な２枚の画像の相関計算から横断者の移動量が算出される。そして、この移動量は移動量データ保存メモリＭ６に保存される。
【００２４】
垂直エッジ数算出回路５は、前記右カメラ画像保存メモリＭ１に保存されている右カメラの画像データを取り込み、所定の大きさで区画された検出ウィンドウ毎に同ウィンドウ内の垂直エッジ数をカウントする。ここで、検出ウインドウは、前記図４の解析領域内における横断者の大きさに見合うように設定される。本実施の形態では、５０ｍ前方の横断者の大きさを基準として、図７に示すように縦６０ドット×横１６ドットの領域にて検出ウィンドウを設定している。
【００２５】
また、検出ウィンドウは、図１８に示すように、検出ウィンドウの横幅の半分の８ドットが隣の検出ウィンドウと重なるように配列される。この場合、画像の水平方向の画素数が５１２ドットであることから、１２７個の検出ウィンドウが設定される。
【００２６】
垂直エッジ数の算出方法を以下に記載する。即ち、解析領域の左上から本サイクルの画像データを読み込み、走査線方向に輝度変化量の大きさが所定のしきい値以上である点ｘについて、そのｘ位置に対応する垂直エッジ数データ保存メモリＭ７の位置に「１」を加算していく。これを解析領域すべてについて行う。次に検出ウィンドウの横幅について、垂直エッジ数データ保存メモリＭ７から対応するｘ位置のデータの総和を求めると、検出ウィンドウの垂直エッジ数が得られる。
【００２７】
さらに、走行路検出回路６は、前記右カメラ画像保存メモリＭ１に保存されている右カメラの画像データを取り込み、自車両の走行路を検出する。より具体的には、画像内の輝度情報から道路上の左右の白線レーンマークを検出し、このレーンマーク内を走行路として検出する。
【００２８】
マイクロコンピュータ７は、周知のＣＰＵ，メモリ，入出力回路等にて構成されており、前記距離データ保存メモリＭ３、移動量データ保存メモリＭ６、垂直エッジ数データ保存メモリＭ７、及び走行路データ保存データＭ８内の保存データを読み込んで後述する横断者検出処理を実行する。そして、マイクロコンピュータ７は、道路上における横断者の検出結果に応じて警報ブザー８を作動させる。また、マイクロコンピュータ７にはヨーレイトセンサ９が接続されており、マイクロコンピュータ７は同センサ９の出力結果から自車両のヨーイング成分を検出する。
【００２９】
なお、本実施の形態では、マイクロコンピュータ７により横断者候補ウィンドウ検出手段及び横断者検出手段が構成され、ヨーレイトセンサ９によりヨーイング検出手段が構成されている。
【００３０】
次いで、本実施の形態における一連の横断者検出手順をマイクロコンピュータ７による演算処理を中心に説明する。なお、図８は横断者検出ルーチンを示すフローチャートであり、同ルーチンはマイクロコンピュータ７により６６ｍｓ毎に実行される。
【００３１】
さて、同ルーチンがスタートすると、ステップ１０１では、検出ウィンドウ毎に前記垂直エッジ数データ保存メモリＭ７から垂直エッジ数を読み出し、その数が設定しきい値以上のウィンドウを横断者の候補ウィンドウとして検出する（図９）。
【００３２】
続いてステップ１０２では、距離データ保存メモリＭ３から距離データを読み込み、横断者候補ウィンドウに対応する距離データが存在する場合、その領域に対応する移動量データを移動量データ保存メモリＭ６から呼び込んで平均化する。また、その平均化データにより前記候補ウィンドウの物体（横断者）の移動量を算出する。このとき、同時に物体までの距離も保持しておく。
【００３３】
さらに、ステップ１０３では、前記算出された移動値から事前に計測しておいた自車両のヨーイング量を差し引き、この値を画像上での絶対的な物体移動量とする。即ち、ヨーレイトセンサ９の検出値は角度変化（ｄｅｇ／ｓｅｃ）であるため、これを画像上でのふらつき量に変換する。画角２０度で画像の水平方向の分解能が５１２ドットのとき、分解能を画角で割った値２５．６（ドット／ｄｅｇ）にヨーレイト値を掛けたものが画像上でのヨーイング量となる。これを、先に算出した物体の移動量から引くことで自車両のヨーイング成分が物体移動量から排除される。
【００３４】
一方、上記の如く算出された物体の水平方向の移動量は、画像上におけるドット数単位の移動量であるため、続くステップ１０４，１０５では、前記移動量を実空間での物体の水平方向の移動速度ｖ〔ｋｍ／ｈ〕に変換する。つまり、ステップ１０４では、次の（３）式を用い、その時の画像上における水平方向の移動量ｇ〔ドット〕から実空間での移動距離Ｌ〔ｍ〕を算出する。
【００３５】
【数３】

但し、上記（３）式において、「ｃ」はＣＣＤのセルの大きさ（ｍｍ／ドット）、「ｆ」はレンズの焦点距離〔ｍｍ〕、「ｄ１」は物体の移動前の物体と自車両との距離〔ｍ〕、「ｄ２」は物体の移動後の物体と自車両との距離〔ｍ〕である。
【００３６】
さらに、ステップ１０５では、次の（４）式を用い、６６ｍｓサイクル内における実空間での移動距離Ｌ〔ｍ〕を移動速度ｖ〔ｋｍ／ｈ〕に変換する。この移動速度ｖ〔ｋｍ／ｈ〕は、物体の移動量としてマイクロコンピュータ７内のメモリに保持される。
【００３７】
【数４】

次に、ステップ１０６では、物体の移動量（移動速度）と走行路中心との関係より横断者を仮検出する。即ち、前記走行路データ保存メモリＭ８の保存データを呼び込み、物体の距離に対応する走行路の中心位置を算出する。そして、走行路の中心位置と物体の移動方向に基づき、走行路中心に向かう物体を横断者として仮検出する。このとき、図１０に示すように、横断者は画像内において走行路中心に向かう物体として特定され、電柱や立木は画像内において走行路中心から外側に向かう物体として特定される。この仮検出結果はマイクロコンピュータ内のメモリに保存される。
【００３８】
その後、ステップ１０７では、前記仮検出結果に対して時間的な対応付け処理（以後、追跡処理という）を行うことにより、横断者が実際に存在するかどうかを判定する。以下に、追跡処理の具体的内容を説明する。
【００３９】
つまり、４サイクル分の仮検出結果を得た段階でそれを平均化し、追跡処理のための追跡データを作成する。この場合、存在するデータの処理サイクルの重み（事前に設定）の合計が所定のしきい値以上であれば、それらのデータの重みを加えて平均値を算出し、その算出値を追跡データとする。その詳細を図１１に示す。各サイクルの重み付けに際し、本サイクルでは「４」、１サイクル前では「３」、２サイクル前では「２」、３サイクル前では「１」として、重みの合計が「６」以上であれば追跡データを算出する。その結果、追跡データが存在すれば、該当する検出ウィンドウに横断者が存在する旨が検出されることとなる。かかる追跡処理では、検出ミスでデータが飛び飛びでしか存在しない場合にも当該検出ミスによるデータの不在を考慮する必要はなく、その連続性に基づいて横断者が検出される。
【００４０】
次に、追跡データマップ上の具体的な追跡処理について、それを追跡処理Ａ〜Ｄの事例に分けて説明する。なお、追跡処理Ａ〜Ｄの内容を図１２（ａ）〜（ｄ）に図示し、上記追跡処理Ａ〜Ｄを反映した一連の追跡例を図１３に示す。マップ上の数字は追跡データ（移動量）であり、実空間での横断者の移動速度〔ｋｍ／ｈ〕を表している。マイナスの移動速度は、走行路中心から外側に移動する速度値を表す。
【００４１】
追跡処理Ａ：前サイクルの追跡データ及び本サイクルの追跡データが同じ検出ウィンドウ又は隣の検出ウィンドウにある場合、それらの移動速度を比較してその差が許容範囲内（ここでは、２ｋｍ／ｈ）であれば、データの追跡が可能であるとみなす。即ち、図１２（ａ）では、追跡データのウィンドウが隣のウィンドウに移動しているが、これは移動可能なウィンドウとして処理される。
【００４２】
追跡処理Ｂ：移動可能なウィンドウ１個に対して、２個又は３個の追跡データが対応することがある。この場合、図１２（ｂ）に示すように、追跡回数の多いウィンドウを選択して処理する。但し、追跡回数が同じである場合、走行路中心より左側画像の領域では右側のデータを、右側画像の領域では左側のデータを優先する（図１３のＢ’参照）。
【００４３】
追跡処理Ｃ：図１２（ｃ）に示すように、１つの追跡データに対して２個又は３個の移動可能ウィンドウが対応する場合には、その対応する数だけ同じ追跡データを誕生させる。但し、各々の追跡データには識別ラベルを付けておく。
【００４４】
追跡処理Ｄ：図１２（ｄ）に示すように、移動可能ウィンドウがないとき、そこで追跡処理を中断する。
上記追跡データはマイクロコンピュータ７内のメモリに保存されるが、それと同時に、対応する距離データ〔ｍ〕、画像中心からの水平位置ｚｘ〔ｍ〕も保存される。保存量は１０サイクル分までとする。
【００４５】
前記水平位置ｚｘ〔ｍ〕は、画像中心からの物体の水平位置ＤＸ〔ドット〕と、距離データとから以下のようにして算出する。つまり、前記（２）式より、距離から物体の視差ｓ〔ドット〕が算出され、これはこの距離データにおいて、この視差ｓの大きさ〔ドット〕がカメラ基長線ｍ〔ｃｍ〕に相当することを意味する。従って、次の（５）式から水平位置ｚｘ〔ｍ〕が算出される。
【００４６】
【数５】

追跡データの追跡回数が１０サイクルを越えたら、過去９サイクル分と本サイクルとの合計１０サイクル分の保存データから、横断者の平均移動速度ＡＭＶ〔ｋｍ／ｈ〕及び平均水平位置ＡＸ〔ｍ〕を算出する。なお、平均水平位置ＡＸ〔ｍ〕は、画像中心から横断者までの水平距離（前記水平位置ｚｘ）と、画像中心から走行路の白線レーンマークまでの水平距離との和から求められる。この関係を示したのが図１４である。自車両が横断者と同じ距離の走行路までの道のりＤＬ〔ｍ〕及び走行路幅Ｗ〔ｍ〕は、距離データや走行路検出データに基づいて算出される。
【００４７】
以上のデータを用い、図８のステップ１０８では、上記の如く検出された横断者に対して、今現在の自車両の走行状態が危険であるか否かを判定する。本実施の形態では、その一例として、横断者が走行路に達するまでの時間ｔｍ１〔ｓ〕と、走行路を通過し終える時間ｔｍ２〔ｓ〕と、自車両が横断者と同距離の点に達する時間ｔｃ〔ｓ〕とに基づいて危険判定を実施する。つまり、ｔｍ１≦ｔｃ≦ｔｍ２であれば、自車両と横断者との衝突の危険性があるとして、マイクロコンピュータ７はステップ１０９で警報ブザー８を作動させ、横断者の検出の旨をドライバに警告する。また、ｔｃ＜ｔｍ１又はｔｃ＞ｔｍ２であれば、自車両と横断者との衝突の危険性がない、即ち安全であるとしてマイクロコンピュータ７はそのまま本ルーチンを終了する。
【００４８】
ここで、横断者が走行路に達するまでの時間ｔｍ１〔ｓ〕は、上記平均水平位置ＡＸ〔ｍ〕、平均移動速度ＡＭＶ〔ｋｍ／ｈ〕に基づいて次の（６）から算出されるようになっている。
【００４９】
ｔｍ１＝ＡＸ・３．６／ＡＭＶ ・・・（６）
走行路を通過し終える時間ｔｍ２〔ｓ〕は、時間ｔｍ１〔ｓ〕に走行路幅Ｗ〔ｍ〕を通過する時間を加算した（７）式から算出される。
【００５０】
ｔｍ２＝Ｗ・３．６／ＡＭＶ＋ｔｍ１ ・・・（７）
また、自車両が横断者と同距離の点に達する時間ｔｃ〔ｓ〕は、道のりＤＬ〔ｍ〕及び自車両の速度ＡＣＶ〔ｋｍ／ｈ〕に基づいて次の（８）式から算出される。
【００５１】
ｔｃ＝ＤＬ・３．６／ＡＣＶ ・・・（８）
以上詳述した本実施の形態によれば、以下に示す効果が得られる。
（ａ）要するに、本実施の形態では、撮像画像の垂直方向の領域（解析領域）について横断者に見合う大きさの検出ウィンドウを水平方向に隙間なく配置し、前方空間の距離データが存在する検出ウィンドウについてその中の物体の移動方向（移動量データの方向）が走行路中心に向かっていれば走行路上にいる横断者の存在を検出するようにした。上記構成によれば、道路脇に存在する横断者の検出が可能となり、その横断者が自車両の走行路を横断しようとしていること、或いは走行路上に飛び出してくること等が予測可能となる。また、検出ウィンドウという単位にて横断者を検出することにより、横断者のような比較的小さな物体に対してもその検出洩れが回避でき、検出精度を向上させることができる。その結果、自車両に対して安全性の高い撮像処理が実現できる。
【００５２】
（ｂ）白線レーンマークの検出情報から自車両の走行路データ（走行路中心の検出結果を含む）を抽出し、その走行路データから横断者の有無（危険度の判定）を行うようにした。かかる構成によれば、より安全性の高い撮像処理が実現できる。
【００５３】
（ｃ）検出ウィンドウを、隣合うウィンドウに所定量（例えば、横幅半分）だけ重ねて配置した。それにより、検出ウィンドウを並べて配置する場合に比べて、横断者に対応する検出ウィンドウが適切に得られ、横断者の動きが精度良く算出できる。
【００５４】
（ｄ）検出ウィンドウ毎に垂直エッジ数を算出し、垂直エッジ数が所定のしきい値以上のとき、その検出ウィンドウを候補ウィンドウとして検出するようにした。そして、それら候補ウィンドウについて、移動量データから道路上の横断者を検出するようにした。つまり、横断者の画像では垂直エッジ数の密度が高いことに着目し、当該垂直エッジ数に基づいて横断者の存在するウィンドウを検出することにより、その検出処理を効率的に実施できる。このとき、検出されたウィンドウを「候補ウィンドウ」とし、その後、距離分布データと移動量データとを用いて「候補ウィンドウ」の中から実際の横断者の判別を行うようにしたため、より一層確実な横断者検出が可能となる。
【００５５】
（ｅ）１サイクルでの横断者の検出結果を仮検出結果として、その仮検出された横断者の時間的な対応付け処理（追跡処理）を行い、所定サイクル以上の対応付けがなされたとき、横断者の存在を決定するようにした。従って、横断者検出の信頼性をより一層高めることができる。特に、本実施の形態では、４サイクル分の加重平均値を追跡データとして算出し、その追跡データを用いて横断者の検出処理を実施した。そのため、様々な外乱、歩行速度の不均一性等に起因した検出ミスが抑制できる。また、横断者の最初の発見から追跡処理を行い、その連続性を観測することにより、突発的な検出ミスが発生しても横断者の存在を的確に推定して、突発的な検出ミスによる悪影響を減らすことができる。
【００５６】
（ｅ）移動量データから自車両のヨーイング成分を排除し、該自車両のヨーイング成分が排除された結果から画像内の横断者を特定するようにした。この場合、横断者の正確な動きが分かり、精度高い検出結果が得られる。
【００５７】
なお、本発明は上記実施の形態の他に次の形態にて具体化できる。
（１）上記実施の形態では、走行路上の横断者を検出するために、横断者の大きさに適応した検出ウインドウを設定した（縦６０ドット×横１６ドット）。しかし、これを任意に変更してもよい。例えば走行路上を横断する自転車の大きさに適用した検出ウィンドウを設定してもよい。
【００５８】
（２）上記実施の形態では、隣合う検出ウィンドウを横幅半分だけ重ねて配置したが、これを変更してもよい。例えば、隣合う検出ウィンドウの横幅の１／３を重ねて配置したり、横幅の２／３を重ねて配置したりしてもよい。重なり合う部分が少なければ演算負荷が低減でき、重なり合う部分が多ければ検出精度が向上する。
【００５９】
（３）検出ウィンドウの仮検出結果から実際に横断者が存在するウィンドウを特定する処理を変更してもよい。例えば、９サイクル分の仮検出結果が得られた時点で６回以上検出されたウィンドウがある場合、当該ウィンドウに横断者が存在する旨を特定する。
【００６０】
（４）上記実施の形態では、距離算出回路２、解析データ算出回路３、移動量算出回路４、垂直エッジ数算出回路５及び走行路検出回路６をハードで構成したが、これらの全てを又はその一部をマイクロコンピュータ７によるソフト処理にて構成してもよい。
【００６１】
（５）上記実施の形態では、前方の距離分布データを取得するためにステレオ画像の解析（立体視）を用いた。しかし、距離分布データの取得方法はこれのみに限定されるものではなく、例えばスキャニングレーダ装置を用いてもよい。これは、光が物体に反射して戻ってくるまでの往復時間より距離を算出する手段として知られており、この光を前方の所定角度内でスキャニング（走査）することで、距離分布データを取得することができる。
【００６２】
より具体的には、光の走査角の範囲を撮像装置の画角とし、光を放射する走査角を次に述べるように設定し用いることで、本発明の実施に利用できる。ここでは、撮像装置の画角を２０度、水平方向画素数５１２ドットのとき、画像処理の立体視で画像を水平方向８ドット、垂直方向８ドットのメッシュ状に分割して距離分布を算出する場合について述べる。この場合、水平方向８ドットの領域は、０．３１３度に相当するため、画角の範囲内において、この角度（０．３１３度）単位で走査角を逐次変化させていく。そして、得られた距離データをメモリに転送し、必要時にＣＰＵから呼び出す。このようにして、検出ウィンドウに対応した距離分布データが取得できる。
【図面の簡単な説明】
【図１】発明の実施の形態における横断者検出システムの概要を示すブロック構成図。
【図２】左右カメラによるステレオ画像を示す図。
【図３】撮像装置の基本的な構成を示す概略図。
【図４】画像内に解析領域を設定した図。
【図５】横断者が水平右方向に所定量だけ移動した状態を示す図。
【図６】相関値と移動量との関係を示すグラフ。
【図７】検出ウィンドウの大きさを示す図。
【図８】横断者検出ルーチンを示すフローチャート。
【図９】画像内における横断者の候補ウィンドウを示す図。
【図１０】横断者の仮検出結果を示す図。
【図１１】追跡処理を説明するための図。
【図１２】追跡処理の事例を個々に説明するための図。
【図１３】一連の追跡処理を説明するための図。
【図１４】自車両と横断者の位置関係を示す図。
【図１５】画像内における解析領域を説明するための図。
【図１６】検出ウィンドウを説明するための図。
【図１７】検出ウィンドウにて横断者が検出された状態を示す図。
【図１８】検出ウィンドウの配列状態を示す図。
【符号の説明】
１…撮像装置、２…距離分布計測手段としての距離算出回路、４…移動量算出手段としての移動量算出回路、５…垂直エッジ数算出手段としての垂直エッジ数算出回路、６…走行路データ検出手段としての走行路検出回路、７…横断者候補ウィンドウ検出手段及び横断者検出手段としてのマイクロコンピュータ、９…ヨーイング検出手段としてのヨーレイトセンサ。[0001]
BACKGROUND OF THE INVENTION
  This inventionTraverserSpecifically, the image data captured by a CCD camera or the distance obtained by a scanning radar devicedistributionExists on the road of the vehicle using data etc.TraverserFor detectingTraverserThe present invention relates to a detection device.
[0002]
[Prior art]
As a conventional method, the distance distribution of an object existing in the front space is calculated by using stereoscopic vision of image processing or scanning with a laser radar, and obstacles including crossers are detected from the density of this distance distribution. Was.
[0003]
[Problems to be solved by the invention]
However, in the method of detecting an obstacle from the density of the distance distribution as described above, when there is a crossing person on the side of the road, before the crossing person appears on the road on the road (for example, in the white line lane mark). There was a problem that it was difficult to detect it. That is, when the crossing person is on the side of the road, the distance distribution between the crossing person and the building behind the crossing person may be integrated. In such a case, a crossing person cannot be detected at this stage, and can only be detected as a single object after the crossing person has come on the road. As a result, there is a risk that detection of a crossing person will be delayed.
[0004]
Further, as a method for detecting a moving object on a road, a method using an optical flow has been conventionally proposed. In this method, a moving object in an image is captured as a lump moving area, and, for example, a car jumping out on an intersection is detected from the area. However, although the existing technology can detect relatively large moving objects such as automobiles, detection of relatively small objects such as people and bicycles is not possible because the number of existing data in the entire image is small. The problem arises that it is possible.
[0005]
  The present invention has been made paying attention to the above-mentioned problems, and its object is relatively smallTraverserCan be detected accurately and efficiently.TraverserIt is in providing the detection apparatus.
[0006]
[Means for Solving the Problems]
  In order to achieve the above object, according to the first aspect of the present invention, the distance distribution measuring means acquires distance distribution data of an object in front space. More specifically, distance distribution data is acquired using a stereo image obtained by a CCD camera, or distance distribution data is acquired using light reflection by a scanning radar device. The movement amount calculating means acquires a plurality of pieces of image data having different times in the front space by the imaging device, and calculates the movement amount of the object.
The vertical edge number calculating means calculates the number of vertical edges in the detection window for each detection window. The crossing candidate window detecting means detects the detection window as a crossing candidate window when the number of vertical edges by the vertical edge number calculating means is not less than a predetermined threshold. Further, the crossing person detection means is obtained when the distance distribution data of the object in the front space exists in the detected candidate window, and is obtained by the movement amount calculation means in the object in the candidate window. When the direction of the moving amount is toward the center of the traveling path of the host vehicle, the crossing person is detected as a crossing person exists in the candidate window.
In other words, paying attention to the fact that the number of vertical edges is high in the image of a crossing person and detecting the window where the crossing person exists based on the number of vertical edges, the detection process can be carried out efficiently. . At this time, the detected window is set as a “candidate window”, and then the actual crossing person is identified from the “candidate window” using the distance distribution data and the movement amount data, thereby further ensuring the crossing. Person detection is possible.
[0007]
  That is, as shown in FIG. 15, for example, an area of a captured image of 30 to 60 m ahead of the host vehicle isTraverserAnd the amount of movement of the object (crossing person) is calculated in the analysis region.
Further, as shown in FIG. 16, for example, the detection window is set in a size suitable for a crossing person..This detection window isAs in the third aspect of the invention,As shown by (A), (B), and (C) in FIG. 18, it is preferable that a predetermined amount (for example, half the width) is overlapped on adjacent windows..And distancedistributionIf the movement direction of the object in the detection window where the data exists (the direction of the movement amount) is toward the center of the traveling path of the host vehicleTraverserWill be detected. in this case,As in the invention according to claim 2,For example, as shown in FIG. 17, the center of the road is detected from image information such as a lane mark indicating the road of the host vehicle..
[0008]
  According to the above configuration, it is possible to detect a crossing person existing on the side of the road, and it is possible to predict that the crossing person is about to cross the traveling road of the own vehicle or to jump out on the traveling road. In addition, by detecting crossers in the unit of detection window, it is relatively smallTraverserThe detection omission can be avoided and the detection accuracy can be improved. As a result, a highly accurate recognition process can be performed in front of the host vehicle.
[0011]
  Claims4In the invention described inTraverserThe detection means is one cycleTraverserAs a temporary detection result, the temporary detection resultTraverserWhen the association process over time is performed and association over a predetermined cycle is made,TraverserDetermine the existence of. In this case, the reliability of crossing person detection can be further enhanced.
[0012]
  Claim5In the invention described inTraverserThe detection means excludes the yawing component of the own vehicle from the movement amount data by the movement amount calculation means, and the result of eliminating the yawing component of the own vehicleTraverserIs identified. In this case, the accurate movement of the crossing person can be known, and a highly accurate detection result can be obtained.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment embodying the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing an outline of the crossing person detection system in the present embodiment. In FIG. 1, an imaging device 1 is a device that captures stereoscopic image data with a pair of left and right cameras, and the left and right cameras are configured such that their optical axes are parallel to each other and their base line length is parallel to the road surface. Has been. At this time, the imaging apparatus 1 obtains a stereo image as shown in FIG.
[0014]
A basic configuration of the optical system of the imaging apparatus 1 is shown in FIG. That is, the imaging apparatus 1 is provided with two lenses 11 and 12 facing an object (crossing person) to form two viewpoints, and the optical axes of the two lenses 11 and 12 are behind the two lenses 11 and 12, respectively. CCDs (Charge Coupled Devices) 13 and 14 are provided as image sensors that match each other.
[0015]
Here, “d” in the figure is the distance [m] between the lenses 11 and 12 and the object, “f” is the focal length [mm] of the lenses 11 and 12, and “a” and “b” are the object to lens. 11 and 12, the distance [mm] between the point focused on the CCDs 13 and 14 and the optical axis, “m” is the camera base length [cm]. In such a case, the distance d [m] is
d = f · m / (a + b)
Can be obtained. That is, since the baseline length m and the focal length f are known, the distance d to the object can be calculated if (a + b) is obtained. Note that (a + b) is obtained as the most consistent shift amount when the luminance values of the objects in the left and right images are compared while being shifted little by little.
[0016]
When (a + b) in the above equation is replaced with the parallax s expressed by the number of dots and the cell size c [mm / dot] of the CCDs 13 and 14, the following equation (1) is obtained.
[0017]
[Expression 1]

On the other hand, in FIG. 1, the image information obtained by the imaging device 1 is stored in the right camera image storage memory M1 and the left camera image storage memory M2. Data stored in both memories M1 and M2 is input to the distance calculation circuit 2. In the distance calculation circuit 2, the image is divided into meshes with 8 dots in the vertical direction and 8 dots in the horizontal direction as the unit area, and the parallax s (dot) is calculated by correlation calculation using the right image data as the standard data and the left image data as the reference data. Is done. Then, by substituting the calculated parallax s into the equation (1), the distance distribution in the image can be obtained. This distance distribution data is stored in the distance data storage memory M3.
[0018]
The analysis data calculation circuit 3 takes in the image data of the right camera stored in the right camera image storage memory M1, and sums the data of 8 dots in the vertical direction of the unit area (this is called analysis data). calculate. The latest analysis data calculated by the analysis data calculation circuit 3 is stored in the cycle analysis data storage memory M4 as “main cycle analysis data”. At this time, since the analysis data is stored as the sum of data for 8 dots, the data capacity is reduced to 1/8. The analysis data calculation circuit 3 performs the analysis data calculation process in a cycle of 66 ms. The analysis data stored in the cycle analysis data storage memory M4 is transferred to the previous cycle analysis data storage memory M5 as “previous cycle analysis data” every 66 ms. Accordingly, the analysis data of the previous and subsequent cycles at intervals of 66 ms are stored in the memories M4 and M5.
[0019]
Here, the analysis area by the analysis data calculation circuit 3 is set on the screen as shown in FIG. 4, for example. In other words, the present embodiment aims to detect crossers existing in the range of 30m to 60m ahead of the host vehicle, and more specifically, based on a crosser standing on the ground 50m ahead. The analysis area is set. On the screen of FIG. 4, a vertical region having a width of 60 dots extending in the horizontal direction is set as an analysis region.
[0020]
Further, the movement amount calculation circuit 4 takes in the analysis data stored in the cycle analysis data storage memory M4 and the previous cycle analysis data storage memory M5, and detects the crossing from the image shift for each unit area between the analysis data. Is calculated. Here, in order to reduce the amount of calculation, a search is performed only in the lateral direction (horizontal running difference direction). In other words, a search range for correlation calculation is set by desk studying the maximum range in which the traverser moves, and within this search range, the absolute value of the difference (correlation value S ( z)) is calculated. Specifically, the correlation value S (z) for the unit region is calculated using the following equation (2).
[0021]
[Expression 2]

However, “D0” is the present cycle analysis data, “D1” is the previous cycle analysis data, “x0” is the upper left coordinates of the image (correlation calculation search start point), and “w” is the width of the search range in the image [dots ], "Z" is the shift amount [dot]. FIG. 5 shows a state where the crossing person has moved by a predetermined amount in the horizontal right direction between the previous cycle and the main cycle.
[0022]
Regarding the minimum value of the correlation value S (z) obtained as described above, the shift amount at that time corresponds to the movement amount of the crossing person. At this time, when the amount of movement of the crossing person is “0” on the image, that is, when the shift amount is “0”, this corresponds to a case of colliding with the crossing person. This is because a collision occurs when the traveling direction of the host vehicle and the angle of the crossing person do not change and the distance to the crossing person becomes 0 m. FIG. 6 shows the relationship between the correlation value S (z) and the shift amount when the crosser moves in the horizontal right direction. In such a case, the shift amount = k corresponds to the shift amount. At this time, in order to improve accuracy, interpolation may be calculated using the minimum correlation value and two adjacent correlation values, and the movement amount may be calculated with 1/10 dot accuracy.
[0023]
In this way, the movement amount calculation circuit 4 calculates the movement amount of the crossing person from the correlation calculation of the two temporally continuous images captured every 66 ms. The movement amount is stored in the movement amount data storage memory M6.
[0024]
The vertical edge number calculation circuit 5 takes in the image data of the right camera stored in the right camera image storage memory M1 and counts the number of vertical edges in the same window for each detection window partitioned by a predetermined size. . Here, the detection window is set to match the size of the crossing person in the analysis region of FIG. In the present embodiment, the detection window is set in an area of 60 dots long × 16 dots wide as shown in FIG. 7 on the basis of the size of a crossing person 50 meters ahead.
[0025]
Further, as shown in FIG. 18, the detection windows are arranged so that 8 dots, which are half the width of the detection window, overlap with the adjacent detection window. In this case, since the number of pixels in the horizontal direction of the image is 512 dots, 127 detection windows are set.
[0026]
A method for calculating the number of vertical edges will be described below. That is, the image data of this cycle is read from the upper left of the analysis area, and the vertical edge number data storage memory corresponding to the x position of the point x whose magnitude of the luminance change amount is not less than a predetermined threshold in the scanning line direction. “1” is added to the position of M7. This is done for all analysis areas. Next, regarding the horizontal width of the detection window, when the total sum of the data at the corresponding x position is obtained from the vertical edge number data storage memory M7, the number of vertical edges of the detection window is obtained.
[0027]
Furthermore, the travel path detection circuit 6 takes in the image data of the right camera stored in the right camera image storage memory M1 and detects the travel path of the host vehicle. More specifically, the left and right white line lane marks on the road are detected from the luminance information in the image, and the inside of the lane mark is detected as a traveling road.
[0028]
The microcomputer 7 includes a well-known CPU, memory, input / output circuit, and the like. The distance data storage memory M3, the movement amount data storage memory M6, the vertical edge number data storage memory M7, and the travel path data storage data. The stored data in M8 is read and the crossing person detection process described later is executed. Then, the microcomputer 7 activates the alarm buzzer 8 according to the detection result of the crossing person on the road. Further, a yaw rate sensor 9 is connected to the microcomputer 7, and the microcomputer 7 detects the yawing component of the host vehicle from the output result of the sensor 9.
[0029]
  In the present embodiment, the microcomputer 7TraverserCandidate window detection means andTraverserThe detecting means is configured, and the yaw rate sensor 9 constitutes the yawing detecting means.
[0030]
Next, a series of crossing person detection procedures in the present embodiment will be described focusing on arithmetic processing by the microcomputer 7. FIG. 8 is a flowchart showing a crossing person detection routine, which is executed by the microcomputer 7 every 66 ms.
[0031]
When the routine starts, in step 101, the number of vertical edges is read from the vertical edge number data storage memory M7 for each detection window, and a window whose number is equal to or greater than a set threshold is detected as a candidate window for a traverser. (FIG. 9).
[0032]
Subsequently, in step 102, the distance data is read from the distance data storage memory M3, and if there is distance data corresponding to the crossing candidate window, the movement amount data corresponding to the area is loaded from the movement amount data storage memory M6 and averaged. Turn into. Further, the moving amount of the object (crossing person) in the candidate window is calculated from the averaged data. At this time, the distance to the object is also maintained.
[0033]
Further, in step 103, the yaw amount of the host vehicle measured in advance is subtracted from the calculated movement value, and this value is set as an absolute object movement amount on the image. That is, since the detection value of the yaw rate sensor 9 is an angle change (deg / sec), it is converted into a fluctuation amount on the image. When the angle of view is 20 degrees and the horizontal resolution of the image is 512 dots, the value 25.6 (dot / deg) obtained by dividing the resolution by the angle of view multiplied by the yaw rate value is the yawing amount on the image. By subtracting this from the previously calculated amount of movement of the object, the yaw component of the host vehicle is excluded from the amount of movement of the object.
[0034]
On the other hand, since the amount of movement of the object in the horizontal direction calculated as described above is the amount of movement in the number of dots on the image, in the subsequent steps 104 and 105, the amount of movement of the object in the horizontal direction in the real space is calculated. The travel speed is converted into v [km / h]. That is, in step 104, the movement distance L [m] in the real space is calculated from the movement amount g [dot] in the horizontal direction on the image at the time using the following equation (3).
[0035]
[Equation 3]

However, in the above equation (3), “c” is the CCD cell size (mm / dot), “f” is the focal length of the lens [mm], and “d1” is the object before moving the object and the vehicle. The distance [m], “d2” is the distance [m] between the object after moving the object and the host vehicle.
[0036]
Further, in step 105, the moving distance L [m] in the real space within the 66 ms cycle is converted into the moving speed v [km / h] using the following equation (4). This moving speed v [km / h] is held in the memory in the microcomputer 7 as the amount of movement of the object.
[0037]
[Expression 4]

Next, in step 106, a crossing person is temporarily detected from the relationship between the amount of movement (movement speed) of the object and the center of the traveling path. That is, the stored data in the travel route data storage memory M8 is loaded, and the center position of the travel route corresponding to the distance of the object is calculated. Then, based on the center position of the travel path and the moving direction of the object, an object heading to the center of the travel path is temporarily detected as a crossing person. At this time, as shown in FIG. 10, the crossing person is specified as an object heading toward the center of the traveling path in the image, and the utility pole and the standing tree are specified as objects heading outward from the center of the traveling path in the image. This temporary detection result is stored in a memory in the microcomputer.
[0038]
Thereafter, in step 107, it is determined whether or not a crossing person actually exists by performing a temporal association process (hereinafter referred to as a tracking process) on the temporary detection result. The specific contents of the tracking process will be described below.
[0039]
That is, when the provisional detection results for four cycles are obtained, the averages are averaged to create tracking data for tracking processing. In this case, if the total of the processing cycle weights (preset) of existing data is equal to or greater than a predetermined threshold value, the average value is calculated by adding the weights of the data, and the calculated value is used as tracking data. To do. The details are shown in FIG. When weighting each cycle, it is “4” in this cycle, “3” in the previous cycle, “2” in the previous cycle, “1” in the third cycle, and “1” in the previous cycle. Calculate the data. As a result, if tracking data exists, it is detected that a crossing person exists in the corresponding detection window. In such tracking processing, even when there is only a jump in data due to a detection error, it is not necessary to consider the absence of data due to the detection error, and a crossing person is detected based on the continuity.
[0040]
Next, specific tracking processing on the tracking data map will be described by dividing it into cases of tracking processing A to D. The contents of the tracking processes A to D are shown in FIGS. 12A to 12D, and a series of tracking examples reflecting the tracking processes A to D are shown in FIG. The numbers on the map are tracking data (movement amount) and represent the moving speed [km / h] of the crossing person in the real space. The negative moving speed represents a speed value that moves outward from the center of the traveling path.
[0041]
Tracking process A: If the tracking data of the previous cycle and the tracking data of this cycle are in the same detection window or the adjacent detection window, the moving speeds are compared and the difference is within an allowable range (here 2 km / h) If so, it is assumed that the data can be traced. That is, in FIG. 12A, the window of the tracking data is moved to the adjacent window, but this is processed as a movable window.
[0042]
Tracking process B: Two or three pieces of tracking data may correspond to one movable window. In this case, as shown in FIG. 12B, a window with a large number of tracking times is selected and processed. However, when the number of times of tracking is the same, priority is given to the right data in the left image area and the left data in the right image area from the center of the road (see B 'in FIG. 13).
[0043]
Tracking process C: As shown in FIG. 12C, when two or three movable windows correspond to one tracking data, the same number of tracking data is born. However, an identification label is attached to each tracking data.
[0044]
Tracking process D: As shown in FIG. 12D, when there is no movable window, the tracking process is interrupted there.
The tracking data is stored in the memory in the microcomputer 7, but at the same time, the corresponding distance data [m] and the horizontal position zx [m] from the image center are also stored. The storage amount is up to 10 cycles.
[0045]
The horizontal position zx [m] is calculated as follows from the horizontal position DX [dot] of the object from the center of the image and the distance data. In other words, the parallax s [dot] of the object is calculated from the distance from the equation (2), and in this distance data, the size [dot] of the parallax s corresponds to the camera base length m [cm]. Means. Accordingly, the horizontal position zx [m] is calculated from the following equation (5).
[0046]
[Equation 5]

If the number of times tracking data exceeds 10 cycles, the average moving speed AMV [km / h] and average horizontal position AX [m] of the traverser are obtained from the stored data of the total of 10 cycles of the past 9 cycles and this cycle. Is calculated. The average horizontal position AX [m] is obtained from the sum of the horizontal distance from the image center to the crossing person (the horizontal position zx) and the horizontal distance from the image center to the white line lane mark on the road. FIG. 14 shows this relationship. The distance DL [m] and the travel path width W [m] until the host vehicle travels the same distance as the crossing person are calculated based on the distance data and the travel path detection data.
[0047]
Using the above data, in step 108 in FIG. 8, it is determined whether or not the current traveling state of the host vehicle is dangerous for the crossing person detected as described above. In this embodiment, as an example, the time tm1 [s] until the crosser reaches the travel path, the time tm2 [s] when the traverser finishes passing the travel path, and the own vehicle are at the same distance from the crosser. The risk determination is performed based on the reaching time tc [s]. In other words, if tm1 ≦ tc ≦ tm2, the microcomputer 7 activates the alarm buzzer 8 in step 109 to warn the driver that the crossing is detected, because there is a risk of collision between the host vehicle and the crossing person. To do. If tc <tm1 or tc> tm2, the microcomputer 7 terminates the routine as it is because there is no danger of collision between the host vehicle and the crossing person, that is, it is safe.
[0048]
Here, the time tm1 [s] until the crossing person reaches the travel path is calculated from the following (6) based on the average horizontal position AX [m] and the average moving speed AMV [km / h]. It has become.
[0049]
tm1 = AX.3.6 / AMV (6)
The time tm2 [s] that has passed through the travel path is calculated from the equation (7) obtained by adding the time to pass the travel path width W [m] to the time tm1 [s].
[0050]
tm2 = W · 3.6 / AMV + tm1 (7)
The time tc [s] for the host vehicle to reach a point at the same distance as the crossing person is calculated from the following equation (8) based on the road DL [m] and the host vehicle speed ACV [km / h]. .
[0051]
tc = DL · 3.6 / ACV (8)
According to the embodiment described in detail above, the following effects can be obtained.
(A) In short, in the present embodiment, detection windows having a size suitable for a crossing person are arranged in the horizontal direction without gaps in a vertical region (analysis region) of a captured image, and detection of distance data in the front space is present. If the moving direction of the object in the window (the direction of the moving amount data) is toward the center of the travel path, the presence of a crosser on the travel path is detected. According to the above configuration, it is possible to detect a crossing person existing on the side of the road, and it is possible to predict that the crossing person is about to cross the traveling road of the own vehicle or to jump out on the traveling road. Further, by detecting a crossing person in a unit of a detection window, it is possible to avoid a detection omission of a relatively small object such as a crossing person and to improve detection accuracy. As a result, it is possible to realize imaging processing with high safety for the host vehicle.
[0052]
(B) The vehicle's travel route data (including the detection result of the travel route center) is extracted from the detection information of the white line lane mark, and the presence / absence of a crossing person (judgment level determination) is performed from the travel route data. . According to this configuration, it is possible to realize imaging processing with higher safety.
[0053]
(C) The detection window is arranged so as to overlap a predetermined amount (for example, half the width) on the adjacent window. Thereby, as compared with the case where the detection windows are arranged side by side, a detection window corresponding to the crossing person is appropriately obtained, and the movement of the crossing person can be calculated with high accuracy.
[0054]
(D) The number of vertical edges is calculated for each detection window, and when the number of vertical edges is equal to or greater than a predetermined threshold, the detection window is detected as a candidate window. And about the candidate window, the crossing person on the road was detected from movement amount data. That is, focusing on the fact that the density of the number of vertical edges is high in the crossing person's image and detecting the window in which the crossing person exists based on the number of vertical edges, the detection process can be efficiently performed. At this time, the detected window is set as a “candidate window”, and then the actual traverser is determined from the “candidate window” using the distance distribution data and the movement amount data. Crossing person detection becomes possible.
[0055]
  (E) in one cycleTraverserAs a temporary detection result, the temporary detection resultTraverserWhen the time-related association processing (tracking processing) ofTraverserThe existence of was decided. Therefore, the reliability of crossing detection can be further enhanced. In particular, in the present embodiment, a weighted average value for four cycles is calculated as tracking data, and a crossing person detection process is performed using the tracking data. Therefore, detection errors caused by various disturbances, nonuniformity of walking speed, and the like can be suppressed. In addition, by performing tracking processing from the first discovery of a crossing person and observing its continuity, even if a sudden detection error occurs, the existence of the crossing person is accurately estimated, and due to a sudden detection error Adverse effects can be reduced.
[0056]
  (E) The yawing component of the own vehicle is excluded from the movement amount data, and the result of the elimination of the yawing component of the own vehicleTraverserTo be specified. In this case, the accurate movement of the crossing person can be known, and a highly accurate detection result can be obtained.
[0057]
The present invention can be embodied in the following form in addition to the above embodiment.
(1) In the above embodiment, in order to detect a crossing person on the road, a detection window adapted to the size of the crossing person is set (vertical 60 dots × horizontal 16 dots). However, this may be changed arbitrarily. For example, you may set the detection window applied to the magnitude | size of the bicycle which crosses on a running path.
[0058]
(2) In the above-described embodiment, the adjacent detection windows are arranged so as to overlap each other by half the width, but this may be changed. For example, 1/3 of the horizontal width of the adjacent detection windows may be overlapped, or 2/3 of the horizontal width may be overlapped. If there are few overlapping parts, the calculation load can be reduced, and if there are many overlapping parts, the detection accuracy improves.
[0059]
(3) You may change the process which specifies the window where a crossing person actually exists from the temporary detection result of a detection window. For example, when there is a window detected six times or more when the provisional detection result for nine cycles is obtained, it is specified that a crossing person exists in the window.
[0060]
(4) In the above embodiment, the distance calculation circuit 2, the analysis data calculation circuit 3, the movement amount calculation circuit 4, the vertical edge number calculation circuit 5, and the travel path detection circuit 6 are configured by hardware. A part thereof may be configured by software processing by the microcomputer 7.
[0061]
(5) In the above-described embodiment, stereo image analysis (stereoscopic view) is used to acquire forward distance distribution data. However, the distance distribution data acquisition method is not limited to this, and for example, a scanning radar device may be used. This is known as a means for calculating the distance from the round trip time until the light is reflected back to the object, and the distance distribution data is obtained by scanning (scanning) the light within a predetermined angle in front. Can be acquired.
[0062]
More specifically, the range of the scanning angle of light is used as the angle of view of the imaging device, and the scanning angle for radiating light is set and used as described below, so that the present invention can be utilized. Here, when the angle of view of the imaging device is 20 degrees and the number of pixels in the horizontal direction is 512 dots, the distance distribution is calculated by dividing the image into a mesh shape of 8 dots in the horizontal direction and 8 dots in the vertical direction by stereoscopic processing of image processing. Describe the case. In this case, since the region of 8 dots in the horizontal direction corresponds to 0.313 degrees, the scanning angle is sequentially changed in units of this angle (0.313 degrees) within the field angle range. Then, the obtained distance data is transferred to the memory and called from the CPU when necessary. In this way, distance distribution data corresponding to the detection window can be acquired.
[Brief description of the drawings]
FIG. 1 is a block configuration diagram showing an outline of a crossing person detection system in an embodiment of the invention.
FIG. 2 is a diagram showing a stereo image obtained by left and right cameras.
FIG. 3 is a schematic diagram illustrating a basic configuration of an imaging apparatus.
FIG. 4 is a diagram in which an analysis area is set in an image.
FIG. 5 is a diagram showing a state in which a crossing person has moved a predetermined amount in the horizontal right direction.
FIG. 6 is a graph showing a relationship between a correlation value and a movement amount.
FIG. 7 is a diagram showing the size of a detection window.
FIG. 8 is a flowchart showing a crossing person detection routine.
FIG. 9 is a diagram showing a crossing candidate window in an image.
FIG. 10 is a diagram illustrating a temporary detection result of a crossing person.
FIG. 11 is a diagram for explaining tracking processing;
FIG. 12 is a diagram for individually explaining examples of tracking processing;
FIG. 13 is a diagram for explaining a series of tracking processes.
FIG. 14 is a diagram showing a positional relationship between the own vehicle and a crossing person.
FIG. 15 is a diagram for explaining an analysis region in an image.
FIG. 16 is a diagram for explaining a detection window;
FIG. 17 is a diagram illustrating a state in which a crossing person is detected in a detection window.
FIG. 18 is a diagram showing an arrangement state of detection windows.
[Explanation of symbols]
  DESCRIPTION OF SYMBOLS 1 ... Imaging device, 2 ... Distance calculation circuit as distance distribution measurement means, 4 ... Movement amount calculation circuit as movement amount calculation means, 5 ... Vertical edge number calculation circuit as vertical edge number calculation means, 6 ... Traveling path data Travel path detection circuit as detection means, 7...TraverserCandidate window detection means andTraverserMicrocomputer as detection means, 9... Yaw rate sensor as yawing detection means.

Claims (5)

Distance distribution measuring means for acquiring distance distribution data of an object in front space;
To obtain a plurality of image data having different times of the front space by the imaging device, and a movement amount calculating means for calculating an amount of movement of the object,
Crossing detection means for setting an analysis region where the object exists in the vertical direction of the image with respect to the image data and providing a detection window corresponding to the size of the object , and arranging the detection window in the horizontal direction without any gap When,
Vertical edge number calculating means for calculating the number of vertical edges in the detection window for each detection window;
When the number of vertical edges calculated by the means for calculating the number of vertical edges in the detection window is equal to or greater than a predetermined threshold, the crossing candidate window that detects the detection window as a candidate window that is a candidate for crossing With detection means
Wherein the cross-person detection means movement, said even if the distance distribution data of an object space in front is present in the candidate window, obtained by the movement amount calculating means in the object in the candidate window when the direction of the amount is toward the roadway center of the vehicle, transverse's detecting device comprising a benzalkonium detect the crossing person as there is cross party to the candidate window. The crossing person detecting device according to claim 1,
A road data detecting means for detecting a lane mark or the like indicating the road of the vehicle from the image data;
The crossing person detection means is a crossing person detection device that detects the center of the road based on the detection result of the road data detection means . The crossing person detection device according to claim 1 or 2,
The detection window is a crossing person detection device which is arranged by overlapping a predetermined amount on an adjacent window . The crossing person detection means performs a temporal association process of the temporarily detected crossing person as a provisional detection result of the crossing person detection result in one cycle. The crossing person detection device according to any one of claims 1 to 3, which determines the presence of a crossing person . Yawing detection means for measuring the yawing component of the traveling of the host vehicle further comprises
The crossing person detection means excludes the yawing component of the own vehicle from the movement amount data by the movement amount calculation means, and specifies the crossing person from the result of eliminating the yawing component of the own vehicle. The crossing device detection device according to claim .

Images