JP3607449B2 - Image processing alignment device - Google Patents

Description

但し、Ｋは、積分範囲を表す。この正規化相関係数の計算は、例えば正規化相関回路１１０で行われる。この相関係数ｆ（ｘ）は、＋１から−１までの値を示し、＋１が基準パターンと最も一致していることを表す。
【００３９】
次に、検索領域６０２の内、Ｘ座標が２〜Ｘｒｓ＋１であり、Ｙ座標が１〜Ｙｒｓである領域の入力画像データＰ（ｘ）と、基準パターンＲ（ｘ）との間で、前述のようにして正規化相関係数ＣＣ（２，１）を計算する。
【００４０】
以下、同様に、ｉを１以上Ｘｗｓ−Ｘｒｓ＋１以下の自然数、ｊを１以上Ｙｗｓ−Ｙｒｓ＋１以下の自然数とし、検査範囲領域中でＸ座標がｉからＸｒｓ＋ｉ−１までの間にあり、Ｙ座標がｊからＹｒｓ＋ｊ−１までの間にある領域の画像データを取り出し、その領域の画像データと基準パターンとの間で正規化相関係数ＣＣ（ｉ，ｊ）を計算する。相関係数の計算は、（Ｘｗｓ−Ｘｒｓ＋１）×（Ｙｗｓ−Ｙｒｓ＋１）回行われる。
【００４１】
求めた相関係数ＣＣ（１、１）、ＣＣ（２、１）、…、ＣＣ（Ｘｗｓ−Ｘｒｓ＋１、Ｙｗｓ−Ｙｒｓ＋１）は、例えば図７に示すように２次元に配置される。この図を相関マトリクスと呼ぶ。
【００４２】
次に、これらの正規化相関係数の中で最も相関係数が高いものを求め、ずれ量を計算する。すなわち、ＣＣ（Ｘｐ、Ｙｐ）の相関係数が最も大きいとする。また、図６（ｂ）に示すように、基準パターン６０３の上述の登録位置は、検索領域６０２の左上を原点とすると（Ｘｒｐ、Ｙｒｐ）の位置にあるとする。この場合、ずれ量（ＸＤ、ＹＤ）は、位置（Ｘｐ、Ｙｐ）から位置（Ｘｒｐ、Ｙｒｐ）までの画素間距離となる。すなわち、
ＸＤ＝Ｘｐ−Ｘｒｐ
ＹＤ＝Ｙｐ−Ｙｒｐ
と算出される。
【００４３】
次に、このずれ量すなわち画素間距離が予め設定した設定値内に収まるか否かを判定する。ずれ量が設定値以上である場合、このずれ量を減らすようにＸＹステージの移動量を計算し、この補正量をＸＹステージ制御装置１１５に発行する。
【００４４】
この補正量の求め方の一例を以下に説明する。まず、画素の分解能をあらかじめ求める。例えば、Ｘ方向の１画素は１００μｍであり、Ｙ方向の１画素は９０μｍであるとする。すなわち、ＸＭＡＧ＝１００（μｍ／画素）、ＹＭＡＧ＝９０（μｍ／画素）である。また、ＸＹステージの移動量をあらかじめ設定しておく。例えば、ＸＹステージは、Ｘ方向には１パルスで５μｍ移動し、Ｙ方向には１パルスで５μｍ移動するとする。すなわち、ＸＳＴＡＧＥ＝５（μｍ／パルス）、ＹＳＴＡＧＥ＝５（μｍ／パルス）とする。したがって、ＸＹステージ制御装置に発行するステージ移動量すなわちパルス数Ｘｍｏｖ，Ｙｍｉｖは、
−Ｘｍｏｖ＝ＸＤ × （ＸＭＡＧ／ＸＳＴＡＧＥ）
−Ｙｍｏｖ＝ＹＤ × （ＹＭＡＧ／ＹＳＴＡＧＥ）
となる。
【００４５】
ＸＹステージ制御装置１１５は、供給されたパルス数に応じてＸＹステージ１１６を動かす。ＸＹステージ１１６が補正量分だけ移動すると、ＸＹステージ制御装置１１５は、移動完了コマンドを画像処理装置１０１に向けて発行する。
【００４６】
その後、移動後の位置決め対象物の低解像度の画像と高解像度の画像を、上述のように同時に画像メモリ２０７、２１６に取り込み、同様の方法で正規化相関係数とずれ量を計算する。ずれ量が設定値以上の場合は補正量を計算し、ＸＹステージを再度移動して、位置決め対象物の位置を調整する。このシーケンスをずれ量が設定値内に入るまで繰り返す。
【００４７】
ずれ量が設定値以内に入ると、低解像度の画像と同時に取り込んで高解像度用画像メモリ２１６に蓄えておいた高解像度画像データと、高解像度用基準パターンメモリ２１７に記憶されている高解像度用基準パターンを用いて、上述の低解像度画像の場合と同様に、正規化相関回路１１０で正規化相関係数を計算し、相関マトリクスを作成する。
【００４８】
次に、低解像度画像の場合と同様に、相関マトリクス中で最大の相関係数を取るものを基準にして、ずれ量を計算する。
このずれ量が、予め定めた高解像度画像用の設定値より大きい場合、ＸＹステージの移動量を低解像度の場合と同様にして算出し、その移動量をＸＹステージ制御装置１１５に送信する。
【００４９】
ＸＹステージ１１６が移動を完了し、画像処理装置１０１が移動完了コードを受信したら、再び、高解像度の画像と低解像度の画像を同時に取り込み、同様の処理を繰り返す。
【００５０】
このずれ量が設定値内に入った場合、画像処理装置１０１は位置補正完了コマンドを発行し、位置補正処理を終了する。
このように、本実施例では、はじめに低解像度の画像を用いているので広い視野の画像を得ることができ、従来例では視野からはずれて検出できないような場合でも位置決めを行うことが可能となる。また、最終的には高解像度の画像を用いて位置決めを行うため、高精度で位置決めを行うことができる。
【００５１】
さらに、低解像度の画像と高解像度の画像を同時に取り込むため、低解像度の画像から高解像度の画像へ切り換える際に生じる時間的なロスを少なくし、また画像取り込みに要する時間を減らすことができる。そのため、高速な位置決めを実現できる。
【００５２】
なお、上述の実施例において、低解像度の画像と高解像度の画像の２つの画像を用いているが、解像度の種類を２つ以上にしてもよい。
図８は、ダイボンダ装置へ適用した本発明の第２の実施例を示す。本実施例は、ペレットがダイボンダ装置によりピックアップされる位置にＸＹステージ上に載せられているペレットを移動させるものである。
【００５３】
このダイボンダ装置において、画像処理装置１０１、カメラ２０１、２１０、ＣＣＵ２０２、２１１、画像処理装置１０１、モニタ２０６、２１５、キーボード１１４、ＸＹステージ制御装置１１５、ＸＹステージ１１６は、第１の実施例と同様であり、同じ動作を行う。
【００５４】
ＸＹステージ１１６上にはダイシングされたペレットが載せられている。図８に示すように、ＸＹステージ１１６上に、例えばダイシングされ、引き伸ばされたシートに貼り付けられたウェハが設置されている。
【００５５】
また、低解像度のカメラ２０１と高解像度のカメラ２１０は、図９に示すように、ともに２視野鏡筒３０１に収められている。２視野鏡筒３０１にズーミングリング３０３を経て入射された対象物の像は、低解像度のカメラ２０１に入射され、同時に例えば像倍率拡大用リレーレンズ３０４を経て高解像度のカメラ２１０に入射される。カメラ２０１、２１０に入射される像の倍率は、ズーミングリング３０３やリレーレンズ３０４により調整することができる。
【００５６】
以下、本実施例の動作を説明する。
まず、第１の実施例と同様に、位置決めを実行する前に、基準パターンを登録する。図１０に示すように低解像度のモニタ２０６に複数のペレット７０２が映るように設定する。図１０において、６０１は画像取り込み領域、６０２は基準パターンの切り出し領域を表す。また図１１に示すように高解像度のモニタ２１５にピックアップ位置にあるペレット８０２が１個映るように設定する。７０１は画像取り込み領域、７０３は基準パターンの切り出し領域を表す。第１の実施例と同様の方法で、低解像度用の基準パターン、高解像度用の基準パターン、及びそれらの登録位置や検索領域を決める。図１１では、ペレットのコーナーの一部が高解像度基準パターンとして用いられることになる。
【００５７】
その後、図５に示した第１の実施例の処理方法と同様に、低解像度の画像を用い、低解像度用基準パターンとの間で正規化相関係数及びずれ量を計算し、１つのペレットをピックアップ位置に導く。
【００５８】
次に、第１の実施例と同様に、低解像度の画像と同時に取り込んだ高解像度の画像を用い、予め登録しておいた高解像度の基準パターンを使用して、正規化相関によるパターンマッチングを行う。ずれ量が所定の設定値より小さくなるまで、ＸＹステージの位置の補正を行う。このずれ量が所定の設定値より小さくなった場合は、画像処理装置１０１は、ピックアップ許可コマンドを発行し、ペレットのピックアップを行う。
【００５９】
本実施例によれば、第１の実施例と同様の効果を得ることができる。すなわち、はじめに低解像度の画像を用いているので広い視野の画像を得ることができ、従来例では視野からはずれて検出できないような場合でも位置決めを行うことが可能となる。また、最終的には高解像度の画像を用いて位置決めを行うため、高精度で位置決めを行うことができる。そのため、ペレットサイズが大きい場合でも所望の位置決め精度を得ることができ、ピックアップ時にピックアップに失敗したり、ペレット欠けが発生することを防止できる。
【００６０】
さらに、低解像度の画像と高解像度の画像を同時に取り込むため、低解像度の画像から高解像度の画像へ切り換える際に生じる時間的なロスを少なくし、また画像取り込みに要する時間を減らすことができる。そのため、高速な位置決めを実現することができる。
【００６１】
【発明の効果】
以上述べたように、本発明によれば、低解像度の画像を用いることで広い視野の画像が得られるため、広い範囲の位置決めが可能となる。
また、本発明によれば、最終的に高解像度の画像で位置決めを行うため、精度の高い位置決めを行うことができる。
【００６２】
さらに、本発明によれば、低解像度の画像と高解像度の画像を同時に取り込むため、画像切換や画像取り込みにおける時間的なロスがなくなり、高速な位置決めが可能となる。
【図面の簡単な説明】
【図１】本発明の第１の実施例を示すブロック図。
【図２】図１に示した実施例を示す斜視図。
【図３】図１に示した実施例における低解像度用基準パターンの取り込みを示す図。
【図４】図１に示した実施例における高解像度用基準パターンの取り込みを示す図。
【図５】図１に示した実施例の動作フローチャートを示す図。
【図６】基準パターン切り出し領域を説明する図。
【図７】相関マトリクスを示す図。
【図８】本発明の第２の実施例を示す斜視図。
【図９】図８に示した実施例に用いられる２視野鏡筒を示す図。
【図１０】図８に示した実施例における低解像度用基準パターンの取り込みを示す図。
【図１１】図８に示した実施例における高解像度用基準パターンの取り込みを示す図。
【図１２】従来の画像処理装置を示すブロック図。
【符号の説明】
１０１…画像処理装置、
２０１…低解像度画像入力用カメラ、
２０２、２１１…カメラコントロールユニット、
２０３、２１２…ローパスフィルタ、
２０４、２１３…Ａ／Ｄ変換回路、
２０５、２１４…Ｄ／Ａ変換回路、
２０６、２１５…モニタ、
２０７…低解像度用画像メモリ、
２０８…低解像度用基準パターンメモリ、
２０９…同期信号発生回路、
２１０…高解像度画像入力用カメラ、
２１６…高解像度用画像メモリ、
２１７…高解像度用基準パターンメモリ、
４０１…低解像度画像取り込み領域、
４０２…低解像度用基準パターン切り出し領域、
４０３…低解像度用基準パターン、
５０１…高解像度画像取り込み領域、
５０２…高解像度用基準パターン切り出し領域、
５０３…高解像度用基準パターン。[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing alignment apparatus used for aligning the position of an article, and particularly used for positioning a semiconductor product.
[0002]
[Prior art]
FIG. 12 shows a conventional image processing apparatus using grayscale images.
On the XY stage 116, for example, a positioning object is installed.
The camera 102 has a single resolution and photographs the positioning object.
[0003]
A camera control unit 103 (hereinafter referred to as CCU) controls the camera 102.
An image captured by the camera 102 is converted into digital image data via the CCU 103, the low-pass filter 104, and the A / D converter 105 in order.
[0004]
The image memory 108 stores the digitized image data.
The synchronization signal generation circuit 117 generates a synchronization signal and controls the CCU 103 and the A / D converter 105 based on the signal.
[0005]
In the reference pattern memory 109, a reference pattern is registered in advance.
The reduction circuit 118 reduces the input image data stored in the image memory 108 and the reference pattern stored in the reference pattern memory 109.
[0006]
The normalized correlation circuit 110 performs pattern matching between the reduced input image data and the reduced reference pattern using, for example, a method disclosed in Japanese Patent Publication No. 5-52987. Thus, the approximate position of the object captured in the image is detected.
[0007]
Thereafter, for example, the reduction ratio of the reduction circuit 118 is changed so that the image becomes larger than before, and the image pattern near the approximate position of the object is cut out from the input image data stored in the image memory 108. Similarly, the reference pattern is reduced by changing the reduction rate of the reduction circuit 118. The above-described pattern matching is performed between the reduced image pattern and the reference pattern, and more detailed position detection of the positioning object is performed.
[0008]
Using the position of the positioning object thus obtained, the CPU 111 calculates the amount of movement of the XY stage, for example, and sends the amount of movement to the XY stage controller 115 via the external interface 113. The XY stage control device 115 moves the XY stage 116 based on the amount of movement sent to move the object to a desired position.
[0009]
Details of such an image processing apparatus are disclosed in Japanese Patent Publication No. 5-52987.
In some cases, the image processing apparatus shown in FIG. 12 is used to move the pellet to a pickup position where the die bonder apparatus picks up the pellet. In this case, a plurality of pellets are set on the stage 116, an image having a resolution such that a plurality of pellets are captured is taken into the image memory 108, and one of the non-defective pellets is guided to the pickup position for pickup. Details of this die bonder apparatus are disclosed in Japanese Patent Application Laid-Open No. 59-54236.
[0010]
[Problems to be solved by the invention]
In the image processing apparatus shown in FIG. 12, the object is positioned using an image having a single resolution. For this reason, if the image is made to have a high resolution in order to obtain a predetermined positioning accuracy, the field of view of the image is narrowed, and the object to be positioned deviates from the input image, so that there is a possibility that the position cannot be detected.
[0011]
In addition, when the above-described image processing apparatus is added to the die bonder apparatus, it is necessary to project a plurality of pellets in one image. Therefore, when the pellet size is large, the resolution of the image must be lowered. In that case, since the resolution of the image becomes low, the desired positioning accuracy cannot be obtained, and there is a possibility that the pickup may fail during pickup or the chipping of the pellet may occur.
The present invention has been made in view of the above problems, and an object thereof is to perform positioning of an object on a stage with high accuracy with a small number of image captures.
[0012]
[Means for Solving the Problems]
In order to solve the above problems, an image processing alignment apparatus of the present invention is higher than a stage on which a positioning target is installed, a first imaging device that reflects the positioning target, and the first imaging device that displays the positioning target. A second imaging device having a resolution, first image data output from the first imaging device, and second image data having a higher resolution than the first image data output from the second imaging device. first capturing respectively, a second input circuit, the first, first, second image storing respectively the second of the first image data captured by the input circuit and the second image data at the same timing A memory, a first image data reference pattern corresponding to a positioning object projected by the first imaging device, its registered position, and a second image data corresponding to the positioning object projected by the second imaging device; First to advance each store use the reference pattern and the registered position, and the memory for the second reference pattern, the first image data and the first reference pattern memory for the first image memory stores stores pattern matching and said second image memory is the second image data for reference the second image data a second reference pattern memory stores that stores between the first image data for the reference pattern An image processing apparatus having an arithmetic circuit for performing pattern matching with a pattern and calculating a movement amount of the stage, and controlling the movement of the stage based on the movement amount of the stage output from the image processing apparatus And a stage control device.
[0013]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
FIG. 1 is a block diagram showing a first embodiment of the present invention. FIG. 2 is a perspective view of the image processing system shown in FIG. Hereinafter, the same components are denoted by the same reference numerals, and description thereof will not be repeated.
[0014]
A positioning object is installed on the XY stage 116.
The low resolution camera 201 captures an image of the positioning target at a low resolution. The high resolution camera 210 projects the positioning target with high resolution. Images obtained by the cameras 201 and 210 are input to the image processing apparatus 101 through the CCUs 202 and 211, respectively.
[0015]
The image processing apparatus 101 includes low-pass filters 203 and 212, A / D converters 204 and 213, a synchronization signal generation circuit 209, a low-resolution image memory 207, a low-resolution reference pattern memory 208, a high-resolution image memory 216, and a high-resolution image memory 216. A resolution reference pattern memory 217, a normalized correlation circuit 110, a CPU 111, a program memory 112, an external interface 113, and D / A converters 205 and 214 are configured.
[0016]
The CCU 202 supplies a low-resolution analog image signal captured by the camera 201 to the input terminal of the low-pass filter 203. The low pass filter 203 reduces the noise of the analog image signal and supplies the analog image signal to the input terminal of the A / D converter 204. The A / D converter 204 converts the input analog image signal into a digital image represented by, for example, 8 bits per pixel. The digital image is stored in the low resolution image memory 207.
[0017]
Similarly, the CCU 210 supplies a high-resolution analog image signal captured by the camera 210 to the input terminal of the low-pass filter 212. The low-pass filter 212 reduces the noise of the analog image signal and supplies the analog image signal to the input terminal of the A / D converter 213. The A / D converter 213 converts the input analog image signal into a digital image represented by, for example, 8 bits per pixel. The digital image is stored in the high resolution image memory 216.
[0018]
The synchronization signal generation circuit 209 generates a synchronization signal and supplies it to the CCUs 202 and 211 and the A / D converters 204 and 213. As a result, the CCU 202 and the CCU 211, the A / D converter 204 and the A / D converter 213 operate in synchronization. Therefore, the images projected by the low resolution camera 201 and the high resolution camera 210 are stored in the low resolution image memory 207 and the high resolution image memory 216, respectively, at the same timing.
[0019]
The digital image signals output from the A / D converters 204 and 213 are supplied to input terminals of the D / A converters 205 and 214, respectively. The D / A converters 205 and 214 convert the input digital image signals into analog image signals again, and supply the converted analog image signals to the monitors 206 and 215, respectively. The monitors 206 and 215 display low-resolution and high-resolution positioning target images captured by the cameras 201 and 210, respectively.
[0020]
Low resolution image memory 207, low resolution reference pattern memory 208, high resolution image memory 216, high resolution reference pattern memory 217, normalized correlation circuit 110, CPU 111, external interface 113, and program memory 112 input / output terminals Are connected to each other via a bus, and transmit and receive data and the like.
[0021]
The low resolution reference pattern memory 208 and the high resolution reference pattern memory 217 store a low resolution reference pattern and a high resolution reference pattern, which will be described later, respectively.
[0022]
The program memory 112 stores a program for performing image processing.
The CPU 111 performs arithmetic processing and control according to a program stored in the program memory 112.
[0023]
The normalized correlation circuit calculates a normalized correlation coefficient to be described later using, for example, two digital image data.
A keyboard 114 and an XY stage controller 115 are connected to the external interface 113.
[0024]
An XY stage 116 is connected to the XY stage controller 115. The XY stage control device 115 moves the XY stage 116 in the X direction and the Y direction based on the signal supplied from the image processing device 101.
[0025]
The operation of this embodiment will be described below. The reference pattern of the positioning object is registered in advance in the high-resolution and low-resolution reference pattern memories. Next, positioning of the positioning object is executed using the registered reference pattern.
[0026]
First, reference pattern registration will be described.
The positioning target is placed on the XY stage, and the XY stage is adjusted to move the positioning target to a desired position.
[0027]
Next, the low-resolution and high-resolution images of the object are displayed on the monitors 206 and 215 using the cameras 201 and 210. 3 and 4 show the screens of the monitors 206 and 215, respectively. Reference numerals 401 and 501 denote images captured from the cameras 201 and 210, that is, all low-resolution and high-resolution images, respectively.
[0028]
Further, as shown in FIG. 3, a line 402 indicating a cutout area of the reference pattern is displayed on the monitor 206 together with the low-resolution image 403 of the positioning object. While viewing the monitor 206, the user moves the line 402 on the screen by the input device 114 such as a keyboard to determine the cutout position of the reference pattern and the size of the cutout area. The low resolution reference pattern thus determined is stored in the low resolution reference pattern memory 208.
[0029]
In addition, a search area is defined in the low resolution image capturing area 401. The search area includes the reference pattern 403 and is set to be the same as or narrower than the image capture area 401. Further, for example, the coordinates of the upper left corner of the reference pattern cutout area 402 are obtained with the upper left corner of the search area as the origin and, for example, the pixel as a unit. This coordinate is called a registered position, and this registered position is stored in the memory.
[0030]
Next, as shown in FIG. 4, a high-resolution image of the positioning object and a line 502 representing a reference pattern cut-out area are displayed on the monitor 215. As in the case of the low resolution, the user moves the line 502 on the screen by the input device 114 such as a keyboard while looking at the monitor 206 to determine the cutout position and cutout area size of the high resolution reference pattern. The high resolution reference pattern thus determined is stored in the low resolution reference pattern memory 217.
[0031]
Further, a search area is defined in the high resolution image capturing area 501. The search area includes the reference pattern 503 and is set to be the same as or narrower than the image capture area 501. Further, for example, the coordinates of the upper left corner of the reference pattern cutout area 502 are obtained using, for example, the upper left corner of the search area as the origin and the pixel as a unit. This coordinate is called a registered position, and this registered position is stored in the memory.
[0032]
In FIG. 4, the reference pattern 503 is the same as the reference pattern 403 shown in FIG. 3, but it is not always necessary to have the same pattern when the areas that can be projected by the camera 201 and the camera 210 do not match. It is natural.
[0033]
Next, a method for positioning the positioning object will be described.
FIG. 5 is a flowchart showing the positioning process in the present embodiment.
First, the positioning process is started. This process is started, for example, when the external interface 113 receives a positioning request command issued by the XY stage control device 115.
[0034]
First, a low-resolution image captured by the camera 201 and a high-resolution image captured by the camera 210 are simultaneously captured into the image processing apparatus 101 using a synchronization signal. The image data in the image capturing area of the low resolution image and the high resolution image are stored in the image memories 207 and 216, respectively.
[0035]
Next, using the input image P (x) stored in the low-resolution image memory 207 and the low-resolution reference pattern image R (x) stored in the low-resolution reference pattern memory 208, a normalized correlation coefficient is obtained. To detect. Hereinafter, how to obtain the normalized correlation coefficient will be described.
[0036]
As shown in FIG. 6A, a predetermined search area 602 is provided in the image capture area 601. It is assumed that the upper left corner point of the search area 602 is the origin and coordinates are expressed in pixel units. The pixel size in the X direction of the search area 602 is Xws, the pixel size in the Y direction is Yws, the pixel size in the X direction of the reference pattern 603 is Xrs, and the pixel size in the Y direction is Yrs.
[0037]
First, in the search area 602, for example, between the input image data P (x) of the area where the X coordinate is 1 to Xrs and the Y coordinate is 1 to Yrs, and the reference pattern R (x), the following expression The correlation coefficient f (x) is obtained according to The value of f (x) is assumed to be a normalized correlation coefficient CC (1, 1).
[0038]
[Expression 2]

However, K represents an integration range. The calculation of the normalized correlation coefficient is performed by the normalized correlation circuit 110, for example. The correlation coefficient f (x) indicates a value from +1 to −1, and +1 indicates that it is the best match with the reference pattern.
[0039]
Next, between the input image data P (x) in the region where the X coordinate is 2 to Xrs + 1 and the Y coordinate is 1 to Yrs in the search region 602, and the reference pattern R (x), In this way, the normalized correlation coefficient CC (2, 1) is calculated.
[0040]
Similarly, i is a natural number greater than or equal to 1 and less than or equal to Xws-Xrs + 1, j is a natural number greater than or equal to 1 and less than or equal to Yws-Yrs + 1, the X coordinate is between i and Xrs + i-1, and the Y coordinate is Image data in an area between j and Yrs + j−1 is extracted, and a normalized correlation coefficient CC (i, j) is calculated between the image data in that area and the reference pattern. The calculation of the correlation coefficient is performed (Xws−Xrs + 1) × (Yws−Yrs + 1) times.
[0041]
The obtained correlation coefficients CC (1, 1), CC (2, 1),..., CC (Xws-Xrs + 1, Yws-Yrs + 1) are two-dimensionally arranged as shown in FIG. This figure is called a correlation matrix.
[0042]
Next, the normalized correlation coefficient having the highest correlation coefficient is obtained, and the amount of deviation is calculated. That is, it is assumed that the correlation coefficient of CC (Xp, Yp) is the largest. Further, as shown in FIG. 6B, it is assumed that the above-described registration position of the reference pattern 603 is at a position (Xrp, Yrp) where the upper left of the search area 602 is the origin. In this case, the shift amount (XD, YD) is the inter-pixel distance from the position (Xp, Yp) to the position (Xrp, Yrp). That is,
XD = Xp-Xrp
YD = Yp-Yrp
Is calculated.
[0043]
Next, it is determined whether or not the deviation amount, that is, the inter-pixel distance falls within a preset set value. When the deviation amount is equal to or larger than the set value, the movement amount of the XY stage is calculated so as to reduce the deviation amount, and this correction amount is issued to the XY stage controller 115.
[0044]
An example of how to determine this correction amount will be described below. First, the pixel resolution is obtained in advance. For example, it is assumed that one pixel in the X direction is 100 μm and one pixel in the Y direction is 90 μm. That is, XMAG = 100 (μm / pixel) and YMAG = 90 (μm / pixel). Further, the movement amount of the XY stage is set in advance. For example, it is assumed that the XY stage moves 5 μm per pulse in the X direction and 5 μm per pulse in the Y direction. That is, XSTAGE = 5 (μm / pulse) and YSTAGE = 5 (μm / pulse). Therefore, the amount of stage movement to be issued to the XY stage controller, that is, the number of pulses Xmov, Ymov is
−Xmov = XD × (XMAG / XSTAGE)
−Ymov = YD × (YMAG / YSTAGE)
It becomes.
[0045]
The XY stage controller 115 moves the XY stage 116 according to the number of supplied pulses. When the XY stage 116 moves by the correction amount, the XY stage control device 115 issues a movement completion command to the image processing apparatus 101.
[0046]
Thereafter, the low-resolution image and the high-resolution image of the positioning object after the movement are simultaneously loaded into the image memories 207 and 216 as described above, and the normalized correlation coefficient and the shift amount are calculated by the same method. When the deviation amount is equal to or larger than the set value, the correction amount is calculated, and the XY stage is moved again to adjust the position of the positioning object. This sequence is repeated until the deviation amount falls within the set value.
[0047]
When the amount of deviation falls within the set value, the high resolution image data captured simultaneously with the low resolution image and stored in the high resolution image memory 216 and the high resolution image stored in the high resolution reference pattern memory 217 are stored. Using the reference pattern, the normalized correlation coefficient is calculated by the normalized correlation circuit 110 as in the case of the low-resolution image described above, and a correlation matrix is created.
[0048]
Next, as in the case of the low-resolution image, the shift amount is calculated on the basis of the one having the largest correlation coefficient in the correlation matrix.
If this deviation amount is larger than a predetermined setting value for a high resolution image, the movement amount of the XY stage is calculated in the same manner as in the case of the low resolution, and the movement amount is transmitted to the XY stage control device 115.
[0049]
When the XY stage 116 completes the movement and the image processing apparatus 101 receives the movement completion code, the high-resolution image and the low-resolution image are simultaneously captured again, and the same processing is repeated.
[0050]
When the deviation amount falls within the set value, the image processing apparatus 101 issues a position correction completion command and ends the position correction process.
Thus, in this embodiment, since a low-resolution image is used first, it is possible to obtain an image with a wide field of view, and positioning can be performed even in the case where the conventional example cannot be detected because it is out of the field of view. . Further, since positioning is finally performed using a high-resolution image, positioning can be performed with high accuracy.
[0051]
Furthermore, since a low-resolution image and a high-resolution image are captured at the same time, a time loss that occurs when switching from a low-resolution image to a high-resolution image can be reduced, and the time required for image capture can be reduced. Therefore, high-speed positioning can be realized.
[0052]
In the above-described embodiment, two images of a low resolution image and a high resolution image are used. However, two or more types of resolutions may be used.
FIG. 8 shows a second embodiment of the present invention applied to a die bonder apparatus. In this embodiment, the pellet placed on the XY stage is moved to a position where the pellet is picked up by the die bonder device.
[0053]
In this die bonder apparatus, the image processing apparatus 101, the cameras 201 and 210, the CCUs 202 and 211, the image processing apparatus 101, the monitors 206 and 215, the keyboard 114, the XY stage control apparatus 115, and the XY stage 116 are the same as in the first embodiment. And perform the same operation.
[0054]
Dicing pellets are placed on the XY stage 116. As shown in FIG. 8, on the XY stage 116, for example, a wafer that is diced and attached to a stretched sheet is placed.
[0055]
Further, the low-resolution camera 201 and the high-resolution camera 210 are both housed in a two-field barrel 301 as shown in FIG. The image of the object that has entered the two-view lens barrel 301 via the zooming ring 303 is incident on the low-resolution camera 201, and at the same time, for example, incident on the high-resolution camera 210 via the relay lens 304 for enlarging the image magnification. The magnification of the image incident on the cameras 201 and 210 can be adjusted by the zooming ring 303 and the relay lens 304.
[0056]
The operation of this embodiment will be described below.
First, as in the first embodiment, a reference pattern is registered before positioning is executed. As shown in FIG. 10, setting is made so that a plurality of pellets 702 are displayed on the low-resolution monitor 206. In FIG. 10, reference numeral 601 denotes an image capturing area, and 602 denotes a reference pattern cut-out area. Further, as shown in FIG. 11, the high resolution monitor 215 is set so that one pellet 802 at the pickup position is displayed. Reference numeral 701 denotes an image capture area, and reference numeral 703 denotes a reference pattern cut-out area. A reference pattern for low resolution, a reference pattern for high resolution, and their registration positions and search areas are determined by the same method as in the first embodiment. In FIG. 11, a part of the corner of the pellet is used as the high resolution reference pattern.
[0057]
Thereafter, similarly to the processing method of the first embodiment shown in FIG. 5, a normalized correlation coefficient and a deviation amount are calculated with respect to the low resolution reference pattern using a low resolution image, and one pellet is obtained. To the pickup position.
[0058]
Next, as in the first embodiment, using a high-resolution image captured simultaneously with a low-resolution image, using a pre-registered high-resolution reference pattern, pattern matching by normalized correlation is performed. Do. The position of the XY stage is corrected until the deviation amount becomes smaller than a predetermined set value. When the deviation amount becomes smaller than a predetermined set value, the image processing apparatus 101 issues a pickup permission command and picks up the pellet.
[0059]
According to this embodiment, the same effect as that of the first embodiment can be obtained. That is, since a low-resolution image is used first, an image with a wide field of view can be obtained, and positioning can be performed even in the case where the conventional example cannot deviate from the field of view and cannot be detected. Further, since positioning is finally performed using a high-resolution image, positioning can be performed with high accuracy. Therefore, a desired positioning accuracy can be obtained even when the pellet size is large, and it is possible to prevent the pickup from failing at the time of pick-up or the occurrence of pellet chipping.
[0060]
Furthermore, since a low-resolution image and a high-resolution image are captured at the same time, a time loss that occurs when switching from a low-resolution image to a high-resolution image can be reduced, and the time required for image capture can be reduced. Therefore, high-speed positioning can be realized.
[0061]
【The invention's effect】
As described above, according to the present invention, an image with a wide field of view can be obtained by using a low-resolution image, so that a wide range of positioning is possible.
In addition, according to the present invention, since positioning is finally performed with a high-resolution image, positioning with high accuracy can be performed.
[0062]
Furthermore, according to the present invention, since a low-resolution image and a high-resolution image are captured at the same time, there is no time loss in image switching or image capture, and high-speed positioning is possible.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a first embodiment of the present invention.
FIG. 2 is a perspective view showing the embodiment shown in FIG.
FIG. 3 is a diagram showing capturing of a low resolution reference pattern in the embodiment shown in FIG. 1;
FIG. 4 is a diagram illustrating capturing of a high resolution reference pattern in the embodiment illustrated in FIG. 1;
FIG. 5 is a diagram showing an operation flowchart of the embodiment shown in FIG. 1;
FIG. 6 is a diagram for explaining a reference pattern cut-out region.
FIG. 7 is a diagram showing a correlation matrix.
FIG. 8 is a perspective view showing a second embodiment of the present invention.
FIG. 9 is a view showing a two-field lens barrel used in the embodiment shown in FIG. 8;
FIG. 10 is a diagram showing capturing of a low resolution reference pattern in the embodiment shown in FIG. 8;
FIG. 11 is a diagram showing capturing of a high-resolution reference pattern in the embodiment shown in FIG. 8;
FIG. 12 is a block diagram showing a conventional image processing apparatus.
[Explanation of symbols]
101: Image processing apparatus,
201 ... a low resolution image input camera,
202, 211 ... camera control unit,
203, 212 ... low pass filter,
204, 213 ... A / D conversion circuit,
205, 214 ... D / A conversion circuit,
206, 215 ... monitor,
207 ... Image memory for low resolution,
208 ... Low resolution reference pattern memory,
209 ... Synchronization signal generating circuit,
210 ... High resolution image input camera,
216 ... image memory for high resolution,
217... Reference pattern memory for high resolution,
401 ... low-resolution image capture area,
402. Reference pattern cutout region for low resolution,
403 ... Low resolution reference pattern,
501... High-resolution image capture area,
502 ... High-resolution reference pattern cutout region,
503. Reference pattern for high resolution.

Claims (4)

A stage on which a positioning target is installed;
A first imaging device that projects the positioning object;
A second imaging device having a higher resolution than the first imaging device that projects the positioning object;
First and second inputs for capturing first image data output from the first imaging device and second image data having a higher resolution than the first image data output from the second imaging device, respectively . A first image memory and a second image memory for storing the first image data and the second image data captured by the first and second input circuits at the same timing, and the first imaging device. The first image data reference pattern corresponding to the positioning object projected by the image sensor and its registration position, and the second image data reference pattern corresponding to the positioning object projected by the second imaging device and the registration position thereof are stored in advance. first, the memory for the second reference pattern, said first image memory the first image data reference pattern of said first image data a first reference pattern memory for storing stores A pattern matching between the second image data for a reference pattern that the second image data a second reference pattern memory for storing pattern matching with the second image memory stores between An image processing apparatus having an arithmetic circuit for performing the calculation of the amount of movement of the stage;
A stage control device for controlling the movement of the stage based on the movement amount of the stage output by the image processing device;
An image processing alignment apparatus comprising: The image processing apparatus includes:
The first image data output from the first imaging device and the second image data output from the second imaging device are newly captured, and the first image data and the reference for the first image data Pattern matching is performed with respect to a pattern to detect the position of the positioning target, and a shift amount between the registered position of the first image data reference pattern and the detected position is calculated. Is larger than the set value, repeats the process of issuing a correction amount to the stage control device and moving the stage until the deviation amount is equal to or less than the first set value,
When the amount of deviation is less than or equal to the first set value, the second image data captured at the same time as the first image data for the first time, and newly captured from the second imaging device for the second and subsequent times. Pattern matching is performed between the second image data and the second image data reference pattern to detect the position of the positioning target, and the registered position of the second image data reference pattern and the detection The amount of deviation from the measured position is calculated, and when the amount of deviation is larger than a second set value, a correction amount is issued to the stage control device to move the stage. The image processing alignment apparatus according to claim 1, wherein the image processing alignment apparatus is repeated until the value becomes equal to or less than the value. Pattern matching between the image data and the reference pattern is
A plurality of parts are cut out from the image data,
For each of the plurality of clipped portions, the correlation coefficient f (x) having a value in the range of +1 to 0 to −1 from the clipped image data P (x) and the reference pattern R (x). )

However, K: Calculate the integration range,
3. The method according to claim 1, wherein among the plurality of cut out portions, the one having the highest correlation coefficient is determined to match the reference pattern most closely, and x is a detection position. Image processing alignment device. The positioning object is a pellet,
Image processing alignment apparatus according to any one of claims 1 to 3 wherein the pellets are characterized by being positioned in the pickup position of the bonding apparatus.

Images