JP3658643B2 - Nondestructive inspection method using radar image processing system - Google Patents

Description

【０００６】
【発明の実施の形態】
以下、図面を用いて本発明の実施の形態を説明する。
＜イ＞非破壊検査の解析方法の概要
空隙などの目標物を有する媒質、例えば図１のように球形の空隙を有する砂層の上にコンクリート層を有する地層を非破壊的に検査する。そこで、地表の測定面に沿って探査レーダ（ＵＧ−Ｖ３３地中探査レ−ダ、（株）三井造船）を移動し、各測定位置毎に地中に探査レーダから電磁波を放射し、その反射波を受信し、画像処理をして空隙や鉄筋などの目標物の位置や形状を求める。探査レーダは例えば３素子ダイポール、３偏波モード動作のアンテナ方式であり、２０ＭＨｚ〜１ＧＨｚの広帯域周波数の高分解能型である。
【０００７】
レーダから発せられた信号は、一定のビーム幅で広がって伝搬するため、目標物から反射して戻る信号画像は、実際の目標物の形状とは異なっている。そのため、図１のように実際の空隙に対するレーダの測定結果は、図１の右側の場合、実線で示したように球形空隙の目標物から反射して戻るので、レーダ画面上では、垂直方向から受信したと見なして垂直な点線の先端部に目標物があるように表示される。
【０００８】
その結果、レーダ画面上では、図２のように目標物の球形空隙が曲線Ｂのように上方に凸の広がった曲線として現れる。即ち、斜めに入射した各信号の直線（Ｙ＝ｔａｎθｎ（Ｘ−Ｘｎ））は、円の軌跡（（Ｘ−Ｘｎ）² ＋Ｙ² ＝Ｙｎ² ）に従いレーダの測定面と直交する方向の深さとして表示される。
【０００９】
このようにして得られたレーダによる測定画像は、図３のように処理される。即ち、計算を効率よくするために、必要又は有意領域を設定する（Ｓ１）。次に、ノイズとなる表面波を除去し、測定位置を原点にするために時間軸の零点調整を行う（Ｓ２）。更にノイズを抑制するために、画像の平滑化を行う（Ｓ３）。次に、鉄筋と空隙などの目標物を分けるために、検出対象の区分や対象別区間のブロック化を行う（Ｓ４）。ここで分けられた空隙と鉄筋について各々の処理を行い、これらの結果を基にして区間別の画像の合成を行い（Ｓ５）、更に平滑化処理を行う（Ｓ６）。
【００１０】
＜ロ＞空隙形状の復元
レーダの位置が空隙の真上に近づく程、直進に近い信号の影響を強く受けるため反射信号の強度はより強くなる。即ち、画面上では移動するレ−ダと固定された空隙間の位置の変化による反射信号の強度が画像の濃淡の差として示される。したがって、逆に画像の濃淡の変化が最大になる方向は、レーダからの信号が空隙の任意面に当たって反射して戻る信号の方向を示すと考えられる。
【００１１】
このことから、図２のように得られたレーダ画像表示の任意の座標（Ｘｎ、Ｙｎ）における深さＹｎが半径となる円を測定位置の座標（Ｘｎ、０）を中心にして円を描く。これを対称区間の半分である０からｍまでの区間で画像表示座標ごとに求めると式（１）となる。つづいて、座標（Ｘｎ、Ｙｎ）において画像の濃淡強度のグラディエントベクトルＧ（Ｘｎ、Ｙｎ）を式（２）により求め、そのときのベクトルの方向（傾き）は式（３）により計算する（ただし、この場合の傾きの方向は反射波の位相がレーダの信号処理方式などによって変わる可能性があるため、符号の調整が必要な場合がある）。
【００１２】
即ち、もとの空隙の形状は式（１）の円と式（４）の直線が合う交点中で下向値（下測交点）を求めて描くと、復元が可能であり、これにより空隙の形状や位置を検出することができる。このことは、図１の左側の場合を例にとると、点線下部（反射波のサイン波形の部分）のレーダ画像表示位置を実線のようにグラディエントベクトルの方向（右方向）に折り曲げると、実際の空隙の表面の位置になる。
【００１３】
この際、解析は一定の画像強度以上の信号に対して実施する。また、道路のように多層構造では各層の誘電率の差による電磁波の屈折の影響を考慮しなければならない場合がある。この影響を考慮するため、式（５）のように屈折の影響を補正すると、復元画像の収束度がより向上する。
【００１４】
【式３】

【００１５】
【式４】

【００１６】
【式５】

【００１７】
【式６】

【００１８】
【式７】

【００１９】
以上の点を図４のフロー図で示すと、ノイズ除去方式の微分フィルタを使用して、空隙画像のグラディエントベクトル処理を行い（Ｓ１０）、一定大きさ以上の最大強度のグラディエントベクトル座標を追跡し（Ｓ１１）、式３により各座標毎にグラディエント（傾き）を算定する（Ｓ１２）。次に、各座標毎にＹ値（表面からの深さ）を半径にして測定位置（Ｘ、０）を中心とする円を描き、円変換する（Ｓ１３）。空隙形状を復元するために、測定位置（Ｘ、０）を通るステップＳ１２で求めたグラディエントベクトルの傾きの直線とステップＳ１３で求めた円との下向き交点（下測交点）を算出する（Ｓ１４）。ステップＳ１１で求めた座標について復元処理が終了すると（Ｓ１５）、地層が多重である場合、屈折の影響を補正する（Ｓ１６）。次に、異常な値を除去するために、復元座標の平均化処理を行い（Ｓ１７）、更に空隙形状を明確にするために空隙を１とし、他を０とする重み関数を用いる（Ｓ１８）。なお、ステップＳ１０では、輪郭線処理により該当座標のみに対したグラディエントベクトル処理方式も可能である。また、ステップＳ１１では、空隙からの１番目と２番目（極性が互いに反対）の反射信号中で強い信号の方を選択する（後で深さ補正を行う）。
【００２０】
＜ハ＞鉄筋位置の同定
コンクリート内の鉄筋のように間隙が狭く、特に複鉄筋の場合には反射信号の散乱及び干渉などにより各鉄筋の焦点又は反射信号の前縁が区別しにくくなるので、鉄筋の焦点部分のみを重み係数（= １）によって強調し、他の部分は重み係数（= ０）により画像の背景信号として同一化させる方法を取る。
【００２１】
例えば、表面近くに複鉄筋が配置され、その下方に矩形の空隙がある領域のレーダ画像を図９に示す。この画像で鉄筋からのピーク信号（山の頂点）を鉄筋の位置と設定すると、この頂点は肉眼では把握しにくいため、図９のデータ中で鉄筋の配筋方向と直交方向で、水平方向の画像データの配列に対して深さ（各行）毎に線図を描くと、鉄筋の存在しない部分の行線図は図６のように成り、鉄筋の存在する場合で最大のピーク信号が得られる部分の行線図は図７のように成る。なお、図６乃至図７に於いて、縦軸は、相対強度（Ｒｅｌａｔｉｖｅ Ｉｎｔｅｎｓｉｔｙ）を示し、横軸は、測定方向の距離（Ｄｉｓｔａｎｃｅ（ｃｍ））を示している。
【００２２】
このように描かれた行線図の中で、例えば図７のように一連のピーク点が最大強度となる場合の座標が各鉄筋の位置を示している。即ち、各ピークの横方向の列座標（三角表示部分）が鉄筋のピッチ（水平位置）を示し、そのときの行座標が画面上の鉄筋の被覆厚さ（垂直位置）を示している。この過程をコンピュータで演算するには、各行毎の平均値で直線を引き（図７のＡ１）、その平均線以上の値に該当する各区間（平均線上で山を形成している各部分）を区間別に設定しておく。この際、鉄筋の配筋深さの差によって２つ以上のピークが一次平均線（図７のＡ１）の上に独立せずに連結されて存在する場合があり、その時は、一次平均線以上の値に対してもう一回平均した値を二次平均線（Ａ２）として用いると、より正確に鉄筋位置を検出できる。
【００２３】
即ち、平均値を比較し最大になる行を選べば、その行が鉄筋の頂点、すなわち、行要素の配列中で最大のピーク線図をなす鉄筋の垂直（深さ）位置に相当し、この行における各区間ごとの最大値を計算すると、鉄筋の水平位置になる。
【００２４】
なお、鉄筋からの多重反射信号がピークに選ばれる可能性は各行別に平均値上で＋以上の値のみを選ぶことによって避けることができる。また、複鉄筋の場合には、両方の鉄筋の存在区間を二つのブロックに分けると良い。また、肉眼確認と計算を並行して実施することにより、より精度良く検出できる。また、正確な深さ方向の情報が必要な場合は、誘電率のデータを考慮して、代表箇所のみを選び対象媒質における実際の鉄筋の深さと画面上の鉄筋の深さを補正しておくと、正確な結果が得られる。
【００２５】
以上の処理を図５のフローで示すと、まず、各鉄筋からの画像は双曲線で表示されるので、複数配列された鉄筋の場合、複数の双曲線が合成された画像が求められる（Ｓ２１）。次に、この画像の行（一定深さ）毎の線図を図７のように求める（Ｓ２２）。ステップＳ２２で求められた各行毎の線図の一次平均線及び二次平均線を図７のＡ１、Ａ２のように求める（Ｓ２３）。更に、各行別に二次平均線以上の全区間の積分を行い（Ｓ２４）、最大積分値の行を選定し、鉄筋の深さを求める（Ｓ２５）。二次平均線以上について各区間別に最大値を算出し、鉄筋の水平位置を算出する（Ｓ２６）。各鉄筋位置を明確にするために、鉄筋を１にし、その他を０とする重み関数を用いる（Ｓ２７）。なお、ステップＳ２３、Ｓ２４、Ｓ２５において、一次、二次平均による計算中で一番多くピ−ク数がカウントされた次数を選ぶ（場合によっては、一次平均が良い場合もある）。この場合、行値の符号は正負どちらを選んでも良いが、比較時にはかならず一方の符号のみを基準にする。
【００２６】
以下に本発明の実施例を示す。
＜イ＞実施例１
図８のように砂の中に矩形の空隙（横４０ｃｍ、縦１０ｃｍのポリウレタン材料）を形成し、砂の表面から３ｃｍの深さに径が２２ｃｍの鉄筋を２０ｃｍの間隔で５本配置し、更に、砂の表面から１６ｃｍの深さに同一径の鉄筋を上方の鉄筋の間の下方に配置する。砂の面の上には厚さ５ｃｍのポリウレタン層を配置する。
【００２７】
この空隙と鉄筋の位置と大きさを測定するために、レーダをポリウレタン表面の直線上を移動して、所定の測定位置から電磁波を地中内に照射して、空隙や鉄筋など地中からの反射波を受信する。
【００２８】
＜ロ＞実施例１の測定及び解析結果
実験１の測定結果を図９に示す。この測定結果は、表面波を除去し、表面を時間軸の０に合わせて調整したものである。受信反射波の強度をプロットしたものが図９である。図９の強度分布は、縦方向の深さにある物質からの反射波を示すのではなく、測定位置で受信した種々の角度の反射波の合計を下方にプロットしたものである。この縦方向の深さ（Ｄｅｐｔｈ（ｃｍ））は、概略、反射波の伝搬距離に対応している。横方向は、測定方向の距離（Ｄｉｓｔａｎｃｅ（ｃｍ））を示している。
【００２９】
図９の測定データから空隙形状を復元し、鉄筋位置を同定したものが図１０である。図１２は、図９の矩形空隙に対するグラディエントベクトルを示しており、雑音を除去しない状態で求めたものである。実際に適用するには、目標物を絞り、ブロック化して、雑音を除去して使用する
【００３０】
図１１は、矩形空隙に対する形状の復元（変換）過程を図式化したものである。この表示は、空隙反射信号中で測定距離単位別一定強度以上の最大グラディエントベクトルが得られた座標のみを対象基準にした。画像の雑音などによって、一部変換結果に多少のばらつきが見えるが、比較的原形状に近いことが分かる。この変換過程では、Ｘ、Ｙ座標を画像表示上のスケールに合わせるように注意する。なお、図１１に於いて中央の太い直線は測定表面を示し、三角印は変換前を示し、四角印は変換後を示している。
【００３１】
以上の結果や方法を用いると、図１０において鉄筋位置の同定（白い部分）および矩形空隙の形状（黒い部分）が比較的明確に再現されていることが分かる。また、反射波の電界の極性と大きさを色で表示すると、各目標物の材料の特性を明確に区別することができる。
【００３２】
＜ハ＞実施例２
図１３のように砂中にポリウレタン材料から成る直径３０ｃｍの球形の空隙を形成し、砂の上には１０ｃｍの厚さの無筋コンクリート層を配置した領域を用いる。空隙は、空隙の上部がコンクリート上面から３０ｃｍの位置に配置される。コンクリートは、Ｗ／Ｃ５５％、スランプ５ｃｍ、圧縮強度４００ｋｇ／ｃｍ² であり、砂は静岡県富士川産砂（比重：２．６１、含水率：２〜３％）を使用した。
【００３３】
＜ニ＞実施例２の測定及び解析結果
実施例２（図１３）の測定結果を図１４と図１５に示す。図１４は、表面波及び等方性の境界からの反射波を除去した状態の結果である。この結果は、表面波を除去し、表面を時間軸の０に合わせて調整したものである。この測定データから空隙形状の復元を実施した結果を図１５に示す。図１５の結果を見ると、完全に球形の形状は現れていないが、空隙の大きさなど実際の形状に近い結果が復元されることが分かった。なお、図１４乃至図１５でも、縦方向が深さ（Ｄｅｐｔｈ（ｃｍ））を示し、横方向は、測定方向の距離（Ｄｉｓｔａｎｃｅ（ｃｍ））を示している。
【００３４】
【発明の効果】
本発明は、次のような効果を得ることができる。
＜イ＞画像のグラディエントベクトルを利用して、目標物の形状や位置を求めることが出来る。
＜ロ＞画像の行方向の強度を求め、行方向に平均化することにより、鉄筋の深さと横方向の位置を求めることが出来る。
【図面の簡単な説明】
【図１】空隙のレーダ画像表示説明
【図２】空隙の実形状の復元説明図
【図３】測定画像処理の全体のフロー図
【図４】空隙の位置と大きさを求めるフロー図
【図５】鉄筋の位置を求めるフロー図
【図６】鉄筋のない部分の横方向線図
【図７】鉄筋のある部分の横方向線図
【図８】複数鉄筋条件下の矩形空隙の配置図
【図９】図８の配置の測定画像
【図１０】図８の配置の解析結果
【図１１】空隙の現状の復元説明図
【図１２】図８のグラディエントベクトル例
【図１３】無鉄筋条件下の球形空隙の配置図
【図１４】図１３の配置の測定画像
【図１５】図１３の配置の解析結果[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a nondestructive inspection by a radar image processing system that inspects a target having a different dielectric constant in a medium, such as concrete, a reinforcing bar or a void in the ground.
[0002]
[Prior art]
Conventionally, a target such as a void in the ground or a reinforcing bar in concrete has been investigated without being destroyed by a ground exploration radar or a reinforcing bar survey radar.
[0003]
However, the conventional method has the following problems.
<B> Since the signal emitted from the radar spreads and propagates with a certain beam width, the image of the target object such as the air gap is represented as a widened image different from the actual shape. When determining the presence or absence, there are many places depending on the experience of engineers, and it has been difficult to detect the size and shape.
<B> In the case of reinforced concrete, the reinforcing bars are arranged at a narrow interval in the concrete, and the reflection from the multiple reinforcing bars is strong. Becomes difficult to distinguish. For this reason, it is difficult to detect the position of the reinforcing bar.
[0004]
[Problems to be solved by the invention]
An object of the present invention is to provide a method for easily inspecting the position and size (shape) of a target in different media.
[0005]
[Means for Solving the Problems]
The present invention provides a nondestructive inspection method using a radar image processing system that detects a target existing in a medium, radiates electromagnetic waves from a plurality of measurement positions on a straight line on the surface of the medium, receives reflected waves thereof, A measurement image in the medium is obtained, and a gradient vector having a high intensity in the intensity signal with shading from the target of the measurement image is obtained by the following equation (2). From the measurement position as a reference, the gradient vector is calculated from the reference position. A straight line that inclines in an obtained direction and a straight line that inclines in a direction determined from a gradient vector from the reference position and a circle having this distance as a radius, and a time until a reflected signal from a target is obtained as a distance. This is a nondestructive inspection method using a radar image processing system characterized in that the intersection point of the target object is obtained as a lower measurement intersection point and the shape or position of the target is obtained .
And Formula (2) is as follows.
[Formula 2]

[0006]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
<A> Outline of analysis method for nondestructive inspection A medium having a target such as a void, for example, a formation having a concrete layer on a sand layer having a spherical void as shown in FIG. 1 is nondestructively inspected. Therefore, an exploration radar (UG-V33 underground exploration radar, Mitsui Engineering & Shipbuilding Co., Ltd.) is moved along the measurement surface of the ground surface, and electromagnetic waves are radiated from the exploration radar into the ground at each measurement position. A wave is received and image processing is performed to determine the position and shape of a target such as a gap or a reinforcing bar. The exploration radar is, for example, a three-element dipole, three-polarization mode operation antenna system, and is a high-resolution type with a broadband frequency of 20 MHz to 1 GHz.
[0007]
Since the signal emitted from the radar spreads and propagates with a constant beam width, the signal image reflected and returned from the target is different from the actual shape of the target. Therefore, the radar measurement result for the actual air gap as shown in FIG. 1 is reflected back from the target of the spherical air gap as shown by the solid line in the case of the right side of FIG. It is displayed that the target is at the tip of the vertical dotted line as if it was received.
[0008]
As a result, on the radar screen, the spherical gap of the target appears as a curved line having a convex upward shape as shown by a curve B as shown in FIG. That is, the straight line (Y = tan θn (X−Xn)) of each signal incident obliquely has a depth in a direction perpendicular to the radar measurement surface according to a circular locus ((X−Xn) ² + Y ² = Yn ² ). Is displayed.
[0009]
The measurement image obtained by the radar in this way is processed as shown in FIG. That is, a necessary or significant area is set for efficient calculation (S1). Next, the zero-point adjustment of the time axis is performed in order to remove the surface wave that becomes noise and make the measurement position the origin (S2). Further, in order to suppress noise, the image is smoothed (S3). Next, in order to divide targets such as reinforcing bars and gaps, detection target sections and target-specific sections are blocked (S4). Each processing is performed on the gaps and reinforcing bars divided here, and based on these results, an image for each section is synthesized (S5), and further smoothing processing is performed (S6).
[0010]
<B> The closer the position of the gap-shaped restoration radar is to the position directly above the gap, the stronger the influence of the signal near the straight line, the stronger the reflected signal becomes. That is, on the screen, the intensity of the reflected signal due to the change in position between the moving radar and the fixed gap is shown as a difference in light and shade of the image. Therefore, on the contrary, the direction in which the change in the shade of the image becomes the maximum is considered to indicate the direction of the signal in which the signal from the radar is reflected by returning to an arbitrary surface of the air gap.
[0011]
Accordingly, a circle whose depth Yn is a radius at an arbitrary coordinate (Xn, Yn) of the radar image display obtained as shown in FIG. 2 is drawn around the coordinate (Xn, 0) of the measurement position. . When this is obtained for each image display coordinate in the section from 0 to m, which is half of the symmetrical section, Expression (1) is obtained. Subsequently, the gradient vector G (Xn, Yn) of the intensity of the image at the coordinates (Xn, Yn) is obtained by the equation (2), and the direction (slope) of the vector at that time is calculated by the equation (3) (however, In this case, since the phase of the reflected wave may change depending on the radar signal processing method or the like, the sign may need to be adjusted).
[0012]
In other words, the shape of the original gap can be restored by drawing a downward value (down-measurement intersection) at the intersection point where the circle of equation (1) and the straight line of equation (4) meet. The shape and position can be detected. For example, in the case of the left side of FIG. 1, when the radar image display position below the dotted line (the portion of the sine waveform of the reflected wave) is bent in the direction of the gradient vector (right direction) as shown by the solid line, It becomes the position of the surface of the void.
[0013]
At this time, the analysis is performed on a signal having a certain image intensity or higher. Further, in a multilayer structure such as a road, there are cases where the influence of refraction of electromagnetic waves due to the difference in dielectric constant of each layer must be taken into consideration. In order to consider this influence, if the influence of refraction is corrected as in equation (5), the degree of convergence of the restored image is further improved.
[0014]
[Formula 3]

[0015]
[Formula 4]

[0016]
[Formula 5]

[0017]
[Formula 6]

[0018]
[Formula 7]

[0019]
When the above points are shown in the flowchart of FIG. 4, the gradient vector processing of the air gap image is performed using a noise elimination type differential filter (S10), and the gradient vector coordinates of the maximum intensity exceeding a certain size are tracked. (S11) The gradient (inclination) is calculated for each coordinate according to Equation 3 (S12). Next, a circle centered on the measurement position (X, 0) is drawn with the Y value (depth from the surface) as the radius for each coordinate, and the circle is converted (S13). In order to restore the gap shape, a downward intersection (down-measurement intersection) between the gradient vector gradient line obtained in step S12 passing through the measurement position (X, 0) and the circle obtained in step S13 is calculated (S14). . When the restoration process is completed for the coordinates obtained in step S11 (S15), if the formation is multiple, the influence of refraction is corrected (S16). Next, in order to remove abnormal values, restoration coordinate averaging processing is performed (S17), and a weighting function is used in which the gap is set to 1 and the others are set to 0 to further clarify the gap shape (S18). . In step S10, a gradient vector processing method for only the corresponding coordinates by contour processing is also possible. In step S11, a stronger signal is selected from the first and second reflected signals (opposite polarities are opposite to each other) from the air gap (depth correction is performed later).
[0020]
<C> Identification of rebar position The gap is narrow like the rebar in concrete, especially in the case of double rebar, because it becomes difficult to distinguish the focal point of each rebar or the leading edge of the reflected signal due to scattering and interference of the reflected signal, etc. Only the focal portion of the reinforcing bar is emphasized by the weighting coefficient (= 1), and the other parts are made identical as the background signal of the image by the weighting coefficient (= 0).
[0021]
For example, FIG. 9 shows a radar image of a region where a double rebar is disposed near the surface and a rectangular gap is present below the double rebar. In this image, if the peak signal from the reinforcing bar (the peak of the mountain) is set as the position of the reinforcing bar, this vertex is difficult to grasp with the naked eye. Therefore, in the data of FIG. When a diagram is drawn for each depth (each row) with respect to the array of image data, the row diagram of the portion where the reinforcing bars do not exist is as shown in FIG. 6, and the maximum peak signal is obtained when the reinforcing bars are present. The row diagram of the part is as shown in FIG. 6 to 7, the vertical axis represents relative intensity (Relativity Intensity), and the horizontal axis represents distance in the measurement direction (Distance (cm)).
[0022]
In the row diagram drawn in this way, for example, as shown in FIG. 7, the coordinates when a series of peak points have the maximum intensity indicate the positions of the reinforcing bars. That is, the horizontal column coordinates (triangle display part) of each peak indicate the pitch (horizontal position) of the reinforcing bar, and the row coordinates at that time indicate the covering thickness (vertical position) of the reinforcing bar on the screen. In order to calculate this process with a computer, a straight line is drawn with the average value for each row (A1 in FIG. 7), and each section corresponding to a value equal to or higher than the average line (each portion forming a mountain on the average line). Is set for each section. At this time, two or more peaks may exist on the primary average line (A1 in FIG. 7) independently of each other due to the difference in the reinforcing bar arrangement depth. If the value averaged once more with respect to the value is used as the secondary average line (A2), the position of the reinforcing bar can be detected more accurately.
[0023]
That is, if the average value is compared and the row that becomes the maximum is selected, the row corresponds to the top of the reinforcing bar, that is, the vertical (depth) position of the reinforcing bar that forms the maximum peak diagram in the array of row elements. When the maximum value for each section in the row is calculated, the horizontal position of the reinforcing bar is obtained.
[0024]
Note that the possibility that the multiple reflection signal from the reinforcing bar is selected as a peak can be avoided by selecting only a value greater than + on the average value for each row. Moreover, in the case of a double reinforcing bar, it is good to divide the existence area of both reinforcing bars into two blocks. In addition, the detection can be performed with higher accuracy by performing the visual check and the calculation in parallel. If accurate depth information is required, considering the dielectric constant data, select only representative locations and correct the actual reinforcing bar depth in the target medium and the reinforcing bar depth on the screen. And get accurate results.
[0025]
If the above processing is shown in the flow of FIG. 5, first, images from each reinforcing bar are displayed as hyperbolic curves. Therefore, in the case of a plurality of reinforcing bars arranged, an image in which a plurality of hyperbolic curves are synthesized is obtained (S21). Next, a diagram for each row (constant depth) of this image is obtained as shown in FIG. 7 (S22). The primary average line and the secondary average line for each line obtained in step S22 are obtained as A1 and A2 in FIG. 7 (S23). Further, integration is performed for all sections over the secondary average line for each row (S24), the row with the maximum integration value is selected, and the depth of the reinforcing bar is obtained (S25). The maximum value is calculated for each section for the secondary average line or higher, and the horizontal position of the reinforcing bar is calculated (S26). In order to clarify the position of each reinforcing bar, a weighting function is used in which the reinforcing bar is set to 1 and the others are set to 0 (S27). In steps S23, S24, and S25, the order in which the number of peaks is counted the most in the calculation by the primary and secondary averages is selected (in some cases, the primary average may be good). In this case, the sign of the row value may be either positive or negative, but only one sign is used as a reference at the time of comparison.
[0026]
Examples of the present invention are shown below.
<A> Example 1
As shown in FIG. 8, a rectangular gap (polyurethane material having a width of 40 cm and a length of 10 cm) is formed in the sand, and five rebars having a diameter of 22 cm and a depth of 3 cm from the surface of the sand are arranged at intervals of 20 cm. Furthermore, reinforcing bars of the same diameter are arranged below the upper reinforcing bars at a depth of 16 cm from the sand surface. A 5 cm thick polyurethane layer is placed on the sand surface.
[0027]
In order to measure the position and size of the gap and the reinforcing bar, the radar is moved on a straight line on the polyurethane surface, and electromagnetic waves are irradiated into the ground from a predetermined measurement position. Receive reflected waves.
[0028]
<B> Measurement and analysis results of Example 1 The measurement results of Experiment 1 are shown in FIG. This measurement result is obtained by removing surface waves and adjusting the surface to zero on the time axis. FIG. 9 is a plot of received reflected wave intensity. The intensity distribution of FIG. 9 does not show the reflected wave from the material at a depth in the vertical direction, but is a plot of the sum of the reflected waves of various angles received at the measurement position below. This vertical depth (Depth (cm)) roughly corresponds to the propagation distance of the reflected wave. The horizontal direction indicates the distance in the measurement direction (Distance (cm)).
[0029]
FIG. 10 shows the gap shape restored from the measurement data of FIG. 9 and the position of the reinforcing bar identified. FIG. 12 shows a gradient vector for the rectangular gap in FIG. 9, and is obtained without removing noise. In actual application, the target is squeezed and blocked to remove noise and used.
FIG. 11 is a schematic diagram of the shape restoration (conversion) process for a rectangular gap. In this display, only the coordinates in which the maximum gradient vector of a certain intensity or more per measurement distance unit was obtained in the air gap reflection signal were used as the reference. Depending on the noise of the image, some variations in the conversion results can be seen, but it can be seen that they are relatively close to the original shape. In this conversion process, care is taken to match the X and Y coordinates to the scale on the image display. In FIG. 11, the thick straight line at the center indicates the measurement surface, the triangle mark indicates before conversion, and the square mark indicates after conversion.
[0031]
Using the above results and method, it can be seen that the identification of the reinforcing bar position (white portion) and the shape of the rectangular gap (black portion) are relatively clearly reproduced in FIG. Moreover, when the polarity and magnitude of the electric field of the reflected wave are displayed in color, the characteristics of the material of each target can be clearly distinguished.
[0032]
<C> Example 2
As shown in FIG. 13, a spherical void having a diameter of 30 cm made of polyurethane material is formed in the sand, and an area in which an unreinforced concrete layer having a thickness of 10 cm is arranged on the sand is used. The upper part of the air gap is arranged at a position 30 cm from the upper surface of the concrete. The concrete was W / C 55%, slump 5 cm, and compressive strength 400 kg / cm ² , and the sand used was Fuji River sand (specific gravity: 2.61, moisture content: 2-3%) from Shizuoka Prefecture.
[0033]
<D> Measurement and analysis results of Example 2 FIGS. 14 and 15 show the measurement results of Example 2 (FIG. 13). FIG. 14 shows the result of removing the surface wave and the reflected wave from the isotropic boundary. This result is obtained by removing the surface wave and adjusting the surface to zero on the time axis. FIG. 15 shows the result of restoring the void shape from this measurement data. From the results shown in FIG. 15, it was found that a completely spherical shape did not appear, but a result close to the actual shape such as the size of the void was restored. 14 to 15, the vertical direction indicates the depth (Depth (cm)), and the horizontal direction indicates the distance in the measurement direction (Distance (cm)).
[0034]
【The invention's effect】
The present invention can obtain the following effects.
<A> The shape and position of the target can be obtained by using the gradient vector of the image.
<B> The depth of the reinforcing bar and the position in the lateral direction can be obtained by obtaining the intensity in the row direction of the image and averaging in the row direction.
[Brief description of the drawings]
[Fig. 1] Explanation of radar image display of air gap [Fig. 2] Explanation of restoration of actual shape of air gap [Fig. 3] Flow chart of overall measurement image processing [Fig. 4] Flow chart for determining position and size of air gap [Fig. 5] Flow diagram for finding the position of reinforcing bars [Fig. 6] Horizontal diagram of the part without reinforcing bars [Fig. 7] Horizontal diagram of the part with reinforcing bars [Fig. 8] Arrangement of rectangular gaps under multiple reinforcing bar conditions [ 9 is a measurement image of the arrangement of FIG. 8. FIG. 10 is an analysis result of the arrangement of FIG. 8. FIG. 11 is an explanatory diagram of the current restoration of the void. FIG. 12 is an example of a gradient vector of FIG. Fig. 14 Measurement image of the arrangement of Fig. 13 Fig. 15 Analysis result of the arrangement of Fig. 13

Claims (2)

In a non-destructive inspection method using a radar image processing system for detecting a target existing in a medium,
Irradiate electromagnetic waves from multiple measurement positions on a straight line on the surface of the medium,
Receive the reflected wave,
Find the measurement image in the medium,
The gradient vector having a large intensity in the intensity signal with grayscale from the target of the measurement image is obtained by the following equation (2) ,
With reference to the measurement position, a straight line that extends from the reference position in a direction obtained from the gradient vector ,
Using the time until the reflected signal from the target is obtained as a distance,
Find the intersection of the straight line extending inclined to the distance from the circle and the reference position to the radial direction determined from the gradient vector as the lower measuring point of intersection,
Finding the shape or position of the target,
A non-destructive inspection method using a radar image processing system.
[Formula 1]

In the nondestructive inspection method by the radar image processing system of Claim 1,
The measurement image near the target is blocked ,
A non-destructive inspection method using a radar image processing system.

Images