1288365 九、發明說明： 【發明所屬之技術領域】 記設板對位運動控制系統之對位標 效率對位標記’ _是有關於—種防干擾之高 理方法之整合則運動控騎準定位之快速影像處 【先前技術】 除此=的對位之技術應用為玻璃光罩精密對位， 許業、平面:|板_位技術的範圍甚為廣泛，如半導 二、而朵置：：益業、、印刷電路板業等領域皆為其應用領 技衍，^Ϊ讀位為上述電子產業之各類曝光機之關鍵 技㈣降低製造成本，則對提高國 際只兄f力將有莫大的助益。1288365 Nine, invention description: [Technical field of invention] Recording of the alignment efficiency of the board alignment motion control system _ is the integration of the high-precision method of anti-interference The fast image area [previous technology] In addition to the technical application of this = the precise alignment of the glass mask, Xuye, plane: | plate_bit technology has a wide range, such as semi-conducting two, but the flowering: : Yiye, the printed circuit board industry and other fields are all leading to their application. ^Ϊ Reading is the key technology of all kinds of exposure machines in the above electronic industry. (4) Reducing the manufacturing cost, it will improve the international brothers. Great help.
首先"月參照[第1A圖]所示先前技術之圓環與圓點對 位標記及其影像處衫意圖。目前f知的雙層板對位技術 如日商EP簡公司之STN-LCD生產線的光罩對準系統，是 採用-個圓環標記al及-個圓點對位標記a2進行光罩對 位，這兩組標記分別置於上下兩片不同的板子上，每組標 記又分置於上下板的兩個角落’透過兩具cCD取像及灰階 影像處理’當上下兩個標記中心a3幾乎重疊時，就代表上 下兩片板子完成對位，可進行後續之製造程序。此先前技 術之優點係標記構造及製作簡易，惟其缺點為其灰階值（水 平灰階值a4及垂直灰階值a5)之選定及光源強弱會影響 1288365 精度。 再請參照[帛a _ 於晶圓之光罩對位應用—二不先别技術之雙十字對位標記 雙十字對位的方式，預=忍圖，此種標記對位的方式是採 存於記憶體中，實際對擷曰曰圓上的十字標記bl，將之儲 標記b2與晶圓上的十字時，再將即時從光罩取得之十字 之工作。此法優點：單t Μ進行對位及運動控制補償 影像處理有邊緣干擾及叠^標記，^製作容易，而缺點則是： 續請參照[第1C、第m ^ 位標記於TFT-LCD之舞位圖所示先前技術之雙十字對 仞妒人β从、日丨# 應用示意圖。TFT-LCD在光罩對 位、膠合及檢測時一般传用… 之用十予標記作為對位之標記，這 種對位方式經常採用的是上下板雙十字ei，e2精密對位的 方式。而杈常被討論及使用的標記影像處理方法為H〇ugh Transform，由於存在疊影d等問題，將使得影像處理的速First, "month reference to the prior art ring and dot alignment mark and its image intent as shown in [Fig. 1A]. At present, the two-layer board alignment technology of the known company, such as the mask alignment system of the STN-LCD production line of the Japanese company EP Jane, uses a ring mark a and a dot alignment mark a2 for the mask alignment. The two sets of markers are placed on two different boards, each of which is placed in two corners of the upper and lower panels. 'Two images are taken through cCD and grayscale image processing'. When overlapping, it means that the upper and lower boards are aligned, and the subsequent manufacturing process can be performed. The advantages of this prior art are that the marking structure and fabrication are simple, but the disadvantage is that the selection of the grayscale value (horizontal grayscale value a4 and vertical grayscale value a5) and the intensity of the light source will affect the accuracy of 1288365. Please refer to [帛a _ wafer reticle alignment application - two non-technical double cross align mark double cross alignment method, pre-bearing map, the way of this mark alignment is to save In the memory, the actual cross mark bl on the circle, which stores the mark b2 and the cross on the wafer, will immediately work on the cross taken from the mask. The advantages of this method: single t Μ for alignment and motion control compensation image processing has edge interference and stacking marks, ^ is easy to make, but the disadvantages are: Continue to refer to [1C, m ^ bit mark on TFT-LCD The dance technique shows the prior art double cross pair 仞妒 从 β 从, 日丨# application schematic. The TFT-LCD is generally used for aligning, gluing, and detecting the reticle... It is marked with a ten mark as the alignment mark. This alignment method often uses the method of double-crossing ei and e2 precision alignment of the upper and lower plates. However, the image processing method that is often discussed and used is H〇ugh Transform, which will make the speed of image processing due to problems such as overlapping d.
度與精度受到影響。而此法之優點為：單一種標記，製作 容易，缺點則是：影像處理有邊緣干擾及疊影等問題，處 理速度較慢。 再請參閱[第1E圖]所示先前技術之多組十字標記於 雷射光纖對位應用示意圖。這是一種多組雷射光纖el對位 的應用，同樣是上層板e2、下層板e3雙層精密對位的方 式，較不同的是，使用多組的十字對位標記e4作為對位的 輔助工具及基準，此方法之優點係為採多組標記，對位程 序與精度較嚴謹，而缺點乃係影像處理協同作業較複雜， 對位時間相對較長。 1288365 【發明内容】 有鑑於此’本發明之目的係在提供 處理方:’係可避免雙十字標記之直二 …中二對稱型對位記號，並將影 小挺赴、 分開處理，並財像素（像素之 • 雙以二下1位）之精度得到其形心、（幾何中心），可符合 鲁雙曰板向精密對位之需求’並可有效提高對位速度。 影傻ΐ據上狀目的’树明提供_種靠域設計及其 =一处理方法’包含：提供—具有―對位孔且對位孔内呈 ^對位標記之複合影像，運用二值化法獲得 像 值影像，進行閉合處理將雜訊去除，填滿對位= 心像獲得對位孔影像，以取得對位孔影像之形心、座標，另 再運用二值化法，對複合影像進行處理，以獲得-低閥值 影像之後進行閉合處理，將對位孔之外的影像去除^用 =態學處理方法將對位標記邊緣區域之外的影像去除，獲 侍對位標記影像，再取得對位標記影像之形心座標，之後 將一开座標施以座標運算，分別取得兩組座標之偏移量。 • /本發明之另一目的係在於提供一種雙層板對位運動控 制系統及其應用的方法，其藉由此上述之對位記號影像處 理方法，結合一幾何計算模式與一運動控制模組，可自動 地、快速地完成雙層板的精密對位。 根據上述之目的，本發明提供一種雙層板對位運動控 制系統，包含：設置於第一層板與第二層板之兩組對位記 (ί 7 1288365 號，並以二取像裝置分別攝取對位記號之複合影像，再利 用一對位C號影像處理模組，計算每一複合影像之X軸（橫 轴）與Y軸（縱軸）之一座標偏移量，接著以一雙層板偏差量 計算模組，算得雙層板間之位置與角度偏差量再以三軸運 動控制模組，進行雙層精密對位之補償修正。 【實施方式】 兹配合圖式將本發明較佳實施例詳細說明如下。 • 參閱[第2圖]所示之本發明較佳實施例之對位記號影 像處理方法流程圖。本發明之對位記號影像處理方法S2〇 主要係為求出複合影像之偏移座標值，其至少包含下列步 驟：提供一複合影像，例如可由一影像擷取設備之取像裝 置提供複合影像，重要的是，此複合影像須具有一對位孔， 例如一圓形孔，且對位孔内具有一對位標記，例如一十字 標記（步驟S21)。利用二值化法處理上述複合影像，進行 鲁一兩於一基準閥值之處理，以進一步獲得一高閥值影像（步 驟S22a)。接續進行高閥值影像的閉合處理，將對位孔之 外的雜訊影像去除（步驟S23a)。續填滿高閥值影像内的 對位標記影像，獲得對位孔的影像（步驟S24a)。再利用 形心運算法則求得對位孔影像之形心座標，如形心座標為 (Xlc，Yic)(步驟S25a)。再次運用二值化法對前述的複合 影像進行一低於基準閥值的處理，以獲得一複合影像的低 $值衫像（步驟S22b)。進行低閥值影像的閉合處理，將 對位孔之外的雜訊影像去除（步驟S23b)。利用型態學的處 8 1288365 理技術將對位標記邊緣區域之外的影像去除，而獲得對位 標記影像（步驟S24b)。再以形心法求得對位標記影像之 形心座標，如其形心座標為（Xlm，Ylm)(步驟S25b)。將二 形〜座私施以座標運算，取得一組座標偏移量，例如此組 座標偏移量為△ XI ，△ γι( △ xl=xlm—Xle，△ Yl=Ylm-Ylc)(步驟 S26)。 上述步驟S22a至步驟S25a之取出對位孔影像之形心 座標，和上述之步驟S22b至步驟S25b之求取對位標記之 形心座標的步驟並無先後執行順序之要求，惟進行步驟S26 運算座b偏移置之如，需先將對位孔影像之形心座標與對 位標記之形心座標求出，始得進行。 請參閱[第3圖]所示之本發明較佳實施例之雙層板對 位運動控制系統之系統架構圖、及[第4圖]所繪示之雙層 板對位運動控制糸統之對位示意圖，其中雙層板對位運動 控制系統30包括··兩組對位記號31 ,係分別設置於第一 層板301與第一層板302之對應兩側之同一位置，且每組 對位圮號31係至少由一對位孔311與一對位標記312所組 成。一取像裝置32，設置於第一層板3〇1兩侧面上，取像 裝置32的設置位置可如第2圖内之方式或設置於第一層板 301與第二層板302之斜對角之方式，用以分別攝取對位 記號31之一複合影像。一對位記號影像處理模組33，利 用前述之對位記號影像處理方法，用以自取像裝置32取得 複合影像後，運用高、低二值化法取得其高、低閥值影像， 並去除雜訊後單純化，以求出每一對位孔及對位標記影像 1288365 ^心座標，以計算出每—該純合影像之u (橫轴） *軸（縱軸）之-座標偏移量。一雙層板偏差量計算模 =34’用以自對位記號影像處理模組33取得兩組對位記 =之座標偏移量之後，以幾何空間運算技術，得出第一層 ^〇1與第二層板302之雙層板間的位置與角度偏差量。 ★二軸運動控制模組35，用以取得由雙層板偏差量計算模 ί 出之位置與角度偏差量’並根據位置與角度偏差量 鲁:制第二層板或第—層板向X軸（橫軸）、Υ轴（縱軸）或 :軸（施轉軸）補償移動或轉動—偏差角度。且本發明之 二軸運動控制模組35於修正偏差量時，可固定第一層板 3〇1或第二層板逝中的任—層，而僅根據固定的一層板 來修正另一層板。 [第5Α、5Β、第6Α、6Β、第7Α、7Β及第8Α、8Β圖]係本 發明較佳實施例之雙層板對位運動控制系統之各種形狀的 、子位孔與對位標記形狀示意圖，其中左側乃分♦示對位孔 # 311與對位標記312，右側乃為其重疊後之複合影像示意， 然而上述之對位孔311形狀不必限定為圓形，亦可為橢圓或 、其他可尋求形心座標之兩個象限或四個象限對稱式之幾何 • ie型，又其對位孔311之外圍覆蓋一無亮區遮罩，此區面積 大小與影像之純度有助於取像品質及後續的影像處理。另 外，對位標記形狀也不限定一種形狀，舉凡十字形對位標 記312、方形對位標記312a、三角形對位標記312b、圓形對 位標記312c等可運用幾何原理找出一形心之形狀，皆可作 為本發明之對位標記（31、31a、31b、31c)。 1288365 上述的第一層板301或第二層板302中，若其中一層為 晶圓、STN-LCD或TFT - LCD時，另一層應為玻璃光罩，此時 取像裝置32應設於玻璃光罩之外側。 又，本發明所提之取像裝置32可用CCD、CMOS或數位攝 影機型態之電子裝置。 再請參閱[第9圖]之本發明較佳實施例之雙層板對位 運動控制方法之流程圖。本件雙層板對位運動控制方法 S40，包含下列步驟··首先需在第一層板3〇ι與一第二層板 302上提供兩組對位記號31，此對位記號31係分別設置於 第一層板301與第二層板3〇2之對應兩側之同一位置（步驟 S41)。其次分別取得兩組對位記號31之複合影像，此可藉 由。又置於帛g板3G1或第二層板側面之二取像裝置 32/刀別獲得兩組對位記號31之兩組複合影像（步驟⑽）。 兩組對位記號31之偏差量，可利用前述之對位記 理方法，執行兩組複合影像之兩組座標偏差量計 异，取付岐座標料量分_(Δχι，Λγι) 步驟S43)。續而求出雙層板間之位置與角度偏差 二層板之間位置偏得到第-層板與第 S44)。再藉由一 f、:AX，AY與角度偏差量“(步驟 二;巧動控制模組補償上述位置偏差量△ 量與二::n-標準差，當位置偏差 兩㈣人μ、 丰時回獲得兩組對位記號之 兩、、且複"邊叮步驟依序執行（步驟s46)。完成對位： 1288365 當位置偏差量與角度偏差量進入標準差時即完成對位（步 驟 S47)。 綜上所述，當知本案所揭露之專利技術已具有產業 性、新賴性與進步性，符合發明專利要件。惟以上述者， 僅為本案之一較佳實施例而已，並非用來限定本案實施之 摩色圍。即凡依本案申請專利範圍所做的均等變化與修飾， 皆為本案之專利範圍所涵蓋。 【圖式簡單說明】 第1A圖係先前技術之圓環與圓點對位標記及其影像處理示意 圖； 第1B圖係先前技術之雙十字對位標記於晶圓之光罩對位應用 示意圖； 第1C、1D圖係先前技術之雙十字對位標記於tFT—lcd之對位應 用示意圖； 第1E圖係先前技術之多組十字標記於雷射光纖對位應用示意 圑， 程 第2圖係本發輸佳實關之對位記號及其影像處理方法流 1Ξ1 · 圃， 第3圖係本發·佳實關之雙層板靠運動控㈣統之 架構圖； ” 〜 第4圖係本發明較佳實施例之雙層板對位運動控制系統之 示意圖； ~ 、 8B圖係本發明較佳實 第5A、5B、第6A、6B、第7A、7B及第8A、 1288365 施例之雙層板對位運動控制系統之各種形狀的對位孔與 對位標記形狀示意圖；及 第9圖係本發明較佳實施例之雙層板對位運動控制方法之流程 圖。 【主要元件符號說明】 [先前技術部分]Degree and accuracy are affected. The advantage of this method is that a single type of mark is easy to manufacture, and the disadvantage is that the image processing has problems such as edge interference and overlapping, and the processing speed is slow. Please refer to the schematic diagram of the application of multiple sets of cross marks in the prior art shown in [Fig. 1E] to the laser fiber alignment. This is a multi-group laser fiber el-position application, which is also a double-layer precision alignment of the upper layer e2 and the lower layer e3. The difference is that multiple sets of cross-alignment marks e4 are used as the auxiliary of the alignment. Tools and benchmarks, the advantage of this method is that the multi-group mark is adopted, the alignment procedure and precision are more rigorous, and the disadvantage is that the image processing collaborative work is more complicated and the registration time is relatively longer. 1288365 [Description of the Invention] In view of the above, the object of the present invention is to provide a processing party: 'the system can avoid the two-symmetric type alignment mark of the double cross mark, and the small shadow goes to and separately, and the money is saved. The precision of the pixel (the pixel is doubled to the next one) is the centroid, (geometric center), which can meet the requirement of the precision alignment of the Lushuang board and can effectively improve the alignment speed. According to the above-mentioned purpose, 'Shu Ming provides _ kind of domain design and its = one processing method' includes: providing - a composite image with a - aligning hole and a aligning mark in the aligning hole, using binarization The method obtains the image of the image, performs the closing process to remove the noise, fills the alignment = the image of the heart image to obtain the image of the alignment hole, and obtains the centroid and coordinates of the image of the alignment hole, and then uses the binarization method to perform the composite image. Processing, after obtaining the low threshold image, performing the closing process, removing the image outside the alignment hole, removing the image outside the edge region of the alignment mark by using the state processing method, and obtaining the image of the alignment mark, and then Obtain the centroid coordinate of the alignment mark image, and then apply a coordinate operation to the open coordinate to obtain the offset of the two sets of coordinates. • Another object of the present invention is to provide a two-layer plate alignment motion control system and a method thereof, which combines a geometric calculation mode and a motion control module by the above-described alignment mark image processing method The precise alignment of the double-layer board can be completed automatically and quickly. According to the above object, the present invention provides a two-layer plate alignment motion control system, comprising: two sets of alignment marks arranged on the first layer board and the second layer board (ί 7 1288365, and the two image taking devices respectively Take a composite image of the alignment mark, and then use a pair of C image processing modules to calculate the coordinate offset of one of the X axis (horizon axis) and the Y axis (vertical axis) of each composite image, followed by a pair The layer deviation calculation module calculates the position and angle deviation between the two layers and then uses the three-axis motion control module to perform double-layer precision alignment compensation correction. [Embodiment] The preferred embodiment is described in detail below. • Refer to the flowchart of the alignment mark image processing method according to the preferred embodiment of the present invention shown in [Fig. 2]. The alignment mark image processing method S2 of the present invention is mainly for finding a composite The offset coordinate value of the image includes at least the following steps: providing a composite image, for example, a composite image can be provided by an image capturing device of an image capturing device. Importantly, the composite image must have a pair of bit holes, such as a circle. a hole having a pair of bit marks, for example, a cross mark (step S21). The composite image is processed by a binarization method, and the processing of the reference value is performed to further obtain a high value. Threshold image (step S22a). The closing process of the high threshold image is continued, and the noise image outside the alignment hole is removed (step S23a). The image of the alignment mark in the high threshold image is continuously filled, and the pair is obtained. The image of the bit hole (step S24a). Then, the centroid coordinate of the alignment hole image is obtained by using the centroid algorithm, such as the centroid coordinate (Xlc, Yic) (step S25a). The binarization method is used again for the aforementioned The composite image performs a process lower than the reference threshold to obtain a low value shirt image of the composite image (step S22b). The low threshold image is closed to remove the noise image outside the alignment hole (step S23b). The image outside the edge region of the alignment mark is removed by the method of morphology, and the image of the alignment mark is obtained (step S24b). Then the centroid of the image of the alignment mark is obtained by the centroid method. Coordinates, such as their centroid coordinates (X Lm, Ylm) (step S25b). The two-shaped-seat is operated as a coordinate to obtain a set of coordinate offsets, for example, the offset of the set is Δ XI , Δ γι ( △ xl = xlm - Xle, △ Yl=Ylm-Ylc) (Step S26) The step of extracting the centroid coordinates of the alignment hole image in the above steps S22a to S25a, and the steps of obtaining the centroid coordinates of the alignment mark in the above steps S22b to S25b are not The order of the sequence is executed successively. However, if the step b of the calculation block b is performed in step S26, the centroid coordinate of the alignment hole image and the centroid coordinate of the alignment mark are first obtained, and the process proceeds. See [No. 3 FIG. 1 is a system architecture diagram of a two-layer board alignment motion control system according to a preferred embodiment of the present invention, and a alignment diagram of a two-layer board alignment motion control system illustrated in FIG. 4, wherein The double-layer aligning motion control system 30 includes two sets of alignment marks 31 which are respectively disposed at the same position on the corresponding two sides of the first layer board 301 and the first layer board 302, and each group aligning number 31 It is composed of at least a pair of bit holes 311 and a pair of bit marks 312. An image capturing device 32 is disposed on both sides of the first layer plate 3〇1, and the image capturing device 32 is disposed at a position as shown in FIG. 2 or obliquely disposed on the first layer plate 301 and the second layer plate 302. In a diagonal manner, a composite image of one of the alignment marks 31 is taken separately. The pair of mark image processing module 33 uses the above-described alignment mark image processing method to obtain high and low threshold images by using the high and low binarization methods after acquiring the composite image from the image capturing device 32, and After the noise is removed, the simplification is performed to find each pair of holes and the alignment mark image 1288365 ^ heart coordinates to calculate the coordinate of each of the homozygous images u (horizontal axis) * axis (vertical axis) Transfer amount. A two-layer board deviation amount calculation mode=34' is used to obtain the coordinate offset of the two sets of alignment marks from the alignment mark image processing module 33, and then obtain the first layer by geometric space operation technology. The amount of positional deviation from the angle between the two layers of the second layer 302. ★Two-axis motion control module 35, which is used to obtain the position and angular deviation amount of the double-layer plate deviation calculation and the deviation according to the position and angle: the second layer or the first layer to the X Axis (horizontal axis), Υ axis (vertical axis) or: axis (forward rotation axis) compensates for movement or rotation - deviation angle. Moreover, when the deviation amount is corrected, the two-axis motion control module 35 of the present invention can fix any layer of the first layer 3〇1 or the second layer, and correct the other layer only according to the fixed one layer. . [5th, 5th, 6th, 6th, 7th, 7th, and 8th, 8th drawings] are various shapes, sub-position holes and alignment marks of the double-layer plate alignment motion control system of the preferred embodiment of the present invention. A schematic view of the shape, wherein the left side is ♦ the alignment hole # 311 and the alignment mark 312, and the right side is a composite image after the overlap thereof. However, the shape of the alignment hole 311 is not limited to a circle, and may be an ellipse or Others can seek the two quadrants of the centroid coordinate or the four quadrant symmetry geometry, ie, the outer surface of the alignment hole 311 is covered with a non-bright area mask, and the size of the area and the purity of the image contribute to Image quality and subsequent image processing. In addition, the shape of the alignment mark is not limited to a shape. For example, the cross-shaped alignment mark 312, the square alignment mark 312a, the triangular alignment mark 312b, the circular alignment mark 312c, etc. can use a geometric principle to find a shape of a centroid. Both can be used as the alignment mark (31, 31a, 31b, 31c) of the present invention. 1288365 In the above first layer board 301 or second layer board 302, if one layer is a wafer, STN-LCD or TFT-LCD, the other layer should be a glass mask, and the image capturing device 32 should be disposed on the glass. Outside the mask. Further, the image pickup device 32 of the present invention can be used in an electronic device of a CCD, CMOS or digital camera type. Referring again to the [Fig. 9], a flow chart of the method for controlling the alignment of the two-layer board of the preferred embodiment of the present invention. The double-layer plate alignment motion control method S40 includes the following steps: First, two sets of alignment marks 31 are provided on the first layer board 3〇ι and a second layer board 302, and the alignment mark 31 is separately set. At the same position on the corresponding two sides of the first layer board 301 and the second layer board 3〇2 (step S41). Secondly, a composite image of two sets of alignment marks 31 is obtained, which can be borrowed. The two image capturing devices 32/knobs placed on the side of the 帛g plate 3G1 or the second layer plate obtain two sets of composite images of the two sets of alignment marks 31 (step (10)). The deviation between the two sets of alignment marks 31 can be performed by the above-described alignment management method, and the two sets of coordinate deviations of the two sets of composite images are performed, and the squatting amount is divided into _(Δχι, Λγι) step S43). Continued to find the position and angular deviation between the two plates. The position between the two plates is offset to obtain the first layer and the first layer (S44). Then by a f, :AX, AY and angular deviation amount "(Step 2; Qiao motion control module compensates for the above position deviation amount △ amount and two::n-standard deviation, when the position deviation is two (four) people μ, Fengshi The two sets of two pairs of alignment marks are obtained, and the double-steps are sequentially executed (step s46). The alignment is completed: 1288365 When the position deviation amount and the angular deviation amount enter the standard deviation, the alignment is completed (step S47) In summary, when it is known that the patented technology disclosed in this case has industrial, new and progressive nature, it meets the requirements of the invention patent. However, the above is only a preferred embodiment of the present case, not for use. To limit the implementation of the color of the case, that is, all the changes and modifications made in the scope of patent application in this case are covered by the patent scope of this case. [Simplified description of the drawings] Figure 1A is the ring and circle of the prior art. Point alignment mark and its image processing schematic; Fig. 1B is a schematic diagram of the prior art double cross alignment mark on the wafer reticle alignment application; 1C, 1D picture is the prior art double cross registration mark on tFT- Lcd application Fig. 1E is a schematic diagram of the application of multiple sets of cross marks in the prior art to laser fiber alignment, and the second picture is the alignment mark of the transmission and the image processing method of the present invention and the image processing method flow 1Ξ1 · 圃, Fig. 3 The architecture of the two-layer board of the present invention is based on the motion control (four) system diagram; "~ Figure 4 is a schematic diagram of the two-layer board alignment motion control system of the preferred embodiment of the present invention; ~, 8B diagram BRIEF DESCRIPTION OF THE DRAWINGS FIG. 5A, 5B, 6A, 6B, 7A, 7B and 8A, 1288365, a schematic diagram of the shape of the alignment hole and the alignment mark of various shapes of the double-layer plate alignment motion control system; 9 is a flow chart of a two-layer plate alignment motion control method according to a preferred embodiment of the present invention. [Description of main component symbols] [Prior Technical Section]
(K al 圓環標記 a2 圓點對位標記 s3 標記中心 a4 水平灰階值 a5 垂直灰階值 bl 晶圓上的十字標記 b2 光罩取得之十字標記 cl 雙十字 c2 雙十字 d 疊影 el 雷射光纖 e2 上層板 e3 下層板 e4 十字對位標記 13 1288365(K al ring mark a2 dot alignment mark s3 mark center a4 horizontal gray scale value a5 vertical gray scale value bl cross mark on wafer b2 mask obtained cross mark cl double cross c2 double cross d superimposed el Emitting fiber e2 upper board e3 lower board e4 cross alignment mark 13 1288365
[本發明部分] S20 步驟S21 步驟S22a 步驟S23a 步驟S24a 步驟S25a 步驟S22b 步驟S23b 步驟S24b 步驟S25b 步驟S26 對位記號影像處理方法 提供複合影像 取得高閥值影像 去除雜訊 單純化對位孔影像 求出對位孔影像之形心座標 (Xlc，Ylc) 取得低閥值影像 去除雜訊 單純化對位標記影像 求出對位標記影像之形心座標 (Xlm，Ylm) 取得座標偏移量ΔΧ1，ΔΥ1 (AXl=Xlm-Xlc, Δ Yl-Ylm-Ylc)[Invention Section] S20 Step S21 Step S22a Step S23a Step S24a Step S25a Step S22b Step S23b Step S24b Step S25b Step S26 Alignment Mark Image Processing Method Provides Composite Image Acquisition High Threshold Image Removal Noise Simplified Alignment Hole Image Search The centroid coordinate of the alignment hole image (Xlc, Ylc) obtains the low threshold image to remove the noise and simplifies the alignment mark image to obtain the centroid coordinate of the alignment mark image (Xlm, Ylm) to obtain the coordinate offset ΔΧ1, ΔΥ1 (AXl=Xlm-Xlc, Δ Yl-Ylm-Ylc)
30 301 302 3 卜 31a、31b、31c 雙層板對位運動控制系統 第一層板 第二層板 對位記號 311 對位孔 312、312a、312b、312c 對位標記 32 取像裝置 33 對位記號影像處理模組 34 雙層板偏差量計算模組 14 1288365 35 三軸運動控制模組 X 橫軸 Y 縱軸 θ 旋轉轴 S40 雙層板對位運動控制方法 步驟S41 提供兩組對位記號 步驟S42 取得兩組對位記號之複合影像 步驟S43 求出兩組對位記號之偏差量 步驟S44 求出雙層板間之位置與角度偏差量 步驟S45 以偏差量修正雙層板間之相對位置 步驟S46 判斷是否完成對位，否則返回步驟S42 步驟S47 完成對位30 301 302 3 Bu 31a, 31b, 31c Double-layer board alignment motion control system First layer board Second layer board alignment mark 311 Registration hole 312, 312a, 312b, 312c Alignment mark 32 Alignment device 33 Alignment Symbol image processing module 34 Double-layer deviation calculation module 14 1288365 35 Three-axis motion control module X Horizontal axis Y Vertical axis θ Rotation axis S40 Two-layer plate alignment motion control method Step S41 Provide two sets of alignment marks S42 Obtaining a composite image of two sets of registration marks. Step S43: Determining the deviation amount of the two sets of registration marks. Step S44: Calculating the position and angular deviation between the two layers. Step S45. Correcting the relative position between the two boards by the deviation amount. S46 determines whether the registration is completed, otherwise returns to step S42, step S47, and the alignment is completed.