1274869 九、發明說明： 【發明所屬之技術領域】 本發明係關於一種用以檢查形成在基板之凸塊的凸 塊檢查裝置及方法’制是關制讀查凸塊是否以預定 精密度形成在基板之凸塊檢查裝置及凸塊檢查方法。 【先前技術】 以往’作為連接乡層㈣喊基板之層㈣方法，提 案有-種以凸塊（bump)進行連接的方法。亦即，將例如圓 鐘狀的=塊形成在基板，遞接形成有凸塊之基板及其他 基板’藉此使凸塊貫通於基板，以連接基板彼此者。如此， 在基板形成凸塊時，關於形成有凸塊之位置及凸塊之形 狀，必須以良好之精密度作成凸塊，因此有必要在凸塊形 成後檢查該基板。 就才欢查升/成於基板之方法而言，有一種將檢查對象之 基板之圖像（標的圖像objectimage)與作為樣本之基板 籲圖像（主圖像）予以比較的方法（參照例如日本特開 2001_102761號公報)。在該方法中，將在上面形成有凸塊 的基板自上方進行攝像，而獲得2值圖像之標的圖像。而 且，預先從CAD資料等作成2值圖像之主圖像。然後， 藉由比較標的圖像與主圖像，而檢測出兩者之相里呷分， 根據該相異部分判斷是否以所希望之精密度形成⑽。 在此，通常係在i個基板形成多數個凸塊，因此盘盆 1個1個地個別比較凸塊以進行檢查，惟最好同時比較複 數個凸塊以進行檢查。以下，使用第12a圖至第Μ圖， 317686 5 1274869 說明同時比較複數個凸塊時之檢查處理。 在才双查處理中’首先準備主圖像及標的圖像。第以 /圖至圖係同時比較複數個凸塊時之圖像的例之圖 :第3 : ΓΑ圖至第12°圖中，係說明同時比較檢查第1 士 尾之Ν形。罘12Α圖係顯示主圖像之 之例6塊之圖像93。帛12β圖係顯示標的圖像 :在標嶋包含有第1凸塊之_、第2凸塊之 圖像95、及弟3凸塊之圖像96。 準備好主圖像及標的圖傻 圖像之對位，《便在適當主圖像與標的 ^ L 置匕車乂 口凸塊。例如，為了 進r 之王^凸塊中相異部分的量（面積）成為最小而 :、。騎對位後，進行主圖像與標的W像之比較， ::各：塊中相異之部分。例如，相異部分之面積比 因此，置偏移造成相異部=:: 凸塊、、及因未形成為正確形狀造成相異 刀 貝t預疋之基準量大的凸塊即判斷為缺陷。第 12C圖係重疊顯示主圖像與標的圖像之示意圖。在第以 使圖易於了解，係以虛線顯示主圖像之輪郭， ==標的圖像之輪郭。在第12C圖中，分別在第i 鬼才双測出相異部分。且根據所檢測 的圖像判定各凸塊是否有缺陷。 相“刀 在此’實際上凸塊即使產生某種程度之位置偏離，只 317686 6 1274869 :形，正確形狀的話料算缺陷。例如，直徑 谷許1㈣左右的位置偏離。在第12A圖至第nc :之：中，第2凸塊雖產生位置偏離，但 為沒有缺陷。另一方面，對於形狀必須比:置 ==格地進行判斷之情形較多，凸塊本身未具備預定 :狀的话即判斷為缺陷。在第12A圖至第以圖之例中， 第二凸塊因未形成正確形狀，所以應判斷為缺陷。由以上 之°兄明可知，有關凸塊之位置偏離的容許範圍可以說是比 有關凸塊之形狀的容許範圍更廣。亦即，關於凸塊之位置 偏離上迷基準置應設定為較大’而相比之，關於凸塊之 形狀’上述基準量應設定為較小。 …、而’在習知之方法中，關於位置及形狀之兩方係以 t同之基準量進行檢查，因此無法對其巾—方進行正確之 k 2亦即，將基準量設定為較小時，關於位置偏離之容 許範圍即被設定為較窄，因此雖或多或少有位置偏離，也 Φ會使原本並非缺陷之凸塊（第12A圖至第12C圖之第2凸 塊）也被判斷為缺陷。相反地，將基準量設定為較大時， 關於形狀之容許範圍即被設定為較廣，因此使原本不容許 之㈣的凸塊（第12A圖至第12C圖之第3凸塊）也會被 判斷為非缺陷。因此，在習知之檢查方法中，無法對於凸 塊之位置及形狀來正確地判斷缺陷之有無。 【發明内容】 因此，本發明之目的在於提供一種可對於凸塊之位置 及形狀正確地判斷缺陷之有無的凸塊檢查裝置及凸塊檢查 317686 7 1274869 ' 方法。 為了解決上述課題，本發明係採用以上所示之構成。 亦即，第1態樣係檢查形成在基板之凸塊的凸塊檢查裝 曰一凸塊檢查叙置係具備：對基板進行攝像之攝像部；從 頌不包含於由攝像部所得之標的圖像之凸塊的圖像中，檢 冽出具有正確凸塊形狀之圖像的形狀檢測部；將至少包含 ，由形狀檢測部檢測出之圖像的區域作為非檢查區域，並將 ⑩2他區域設定為檢查區域的檢查區域設定部；記憶作為檢 •二ί基準之主圖像的主圖像記憶部；及藉由對由檢查區域 、、设定部所設定之檢查區域比較標的圖像與主圖像，而檢查 凸塊疋否以預定之精密度形成的比較檢查部。 在第2悲樣中，檢查區域設定部係可將由形狀檢測部 所核測出之圖像區域及其周圍的區域設定為非檢查區域。 且在第3態樣中，凸塊檢查裝置亦可具備用以記憶主 圖像中顯示作為比較檢查部之比較對象的區域之資料的檢 鲁^區域記憶部。此時，比較檢查部係對於由檢查區域設定 斤口又疋之心查區域、且記憶在檢查區域記憶部之資料所 顯示之區域，比較主圖像與標的圖像。 日在第4態樣中，記憶在檢查區域記憶部之資料亦可為 頌不應形成在基板之凸塊所對應之區域者。此時，主圖像 係具有應形成在基板之凸塊的圖像有一部分缺陷的圖像。 ^再者，第5態樣係用以檢查形成在基板之凸塊的凸塊 檢查方法。在凸塊檢查方法中包含：對基板進行攝像之攝 像步驟；從顯示由攝像步驟所得之標的圖像所包含之凸塊 317686 8 1274869 ，的圖像中，檢測出具有正確凸塊形狀之圖像的形狀檢測步 驟；將至少包含由形狀檢測步驟檢測出之圖像的區域作為 非檢查區域，並將其他區域設定為檢查區域的檢查區域設 疋步驟，及藉由對由檢查區域設定步驟所設定之檢查區域 比較標的圖像與主圖像，而檢查凸塊是否以預定之精密度 形成的比較檢查步驟。 • 根據前述第1態樣及第5態樣，預先檢測出正確形狀 籲之凸塊圖像，將檢測出之凸塊圖像之區域設定為非檢查區 ‘域。而且，藉由對檢查區域進行標的圖像與主圖像之比較 -檢查，可將形狀不正確之凸塊及位置偏離超出容許量之凸 塊檢測為缺陷。在此，關於凸塊形狀之容許範圍係在檢測 正確形狀之凸塊圖像之處理中決定，而關於凸塊之位置偏 矛夕之谷才乾圍則在设疋非檢查區域之處理中決定。因此， 根據本發明，由於可將關於凸塊形狀及位置的容許範圍獨 立地分別予以適當設定，因此可對凸塊之位置及形狀正確 ⑩也判所缺之有無。此夕卜，根據本發明，由於採用習知方 法料比較檢查之方法，因此可在不使用特別之比較檢查 運算的情況下正確地判斷缺陷之有無。 根據前述第2態樣，可使關於位置偏離之容許範圍擴 大 般而a，關於凸塊之位置偏離之容許範圍即使擴大 也不會有問題，因此藉由使關於位置偏離之容許範圍擴 大’可進行依據實情之檢查。 根據前述第3態樣’對主圖像也可設定作為比較檢查 之區域因而可自由設定作為比較檢查之對象之區域。 317686 9 1274869 • 根據前述第4態樣，藉由攝像時之基板的反射I法正 確地取得標的圖料，比較檢查之結果會成為缺陷。因此， 即使僅將應形成在基板之凸塊（理想凸塊）之區域作為比 較檢查之對象時，亦可對於無法正確取得之標的圖像獲得 確貫之檢查結果。再者，由於只要對理想凸塊的區域進行 比較4查即可，因此可減低比較檢查之處理量。 • 本發明之上述效果及其他目的、特徵、態樣、功效係 與附圖對照，由以下詳細說明可更進一步了解。 •【實施方式】 .(第1實施形態） 第1圖係第1實施形態之凸塊檢查裝置的構成方塊 圖。·凸塊檢查裝置係具備台座1、CCD ( charge-coupled device’電荷麵合元件）攝像機2及圖像處理裝置凸塊 h查衣f 1係用以檢查在基板s是否正確形成凸塊者。 在第1圖中，台座1係將檢查對象之基板S載置在其 φ上面。在基板8形成有例如圓鐘狀之凸塊。形成在基板S 之凸塊-般而言係直徑1〇〇i 5〇//m，高度__左右 的大小。CCD攝像機2係攝像基板S之攝像手段的一例， k上方攝像口座丨所载置之基板s。經攝像之圖像資料 (數位資料）係使用在後述之比較檢查處理。在本實施形 態中，將由CCD攝像機2所得之圖像稱為標的圖像⑽㈣ 刪ge) ’將表示標的圖像之資料稱為標的圖像資料（_以 imag=data)，再者，若無法將基板§之全面一次攝像時， 亦可藉由使CCD攝像機2及/或台座i移動，而變更攝像 317686 10 1274869 :置並對基板s之全面進行攝像。CCD攝像機2係將所攝 像之基板S之標的圖像資料輸出至圖像處理裝置3 處理裝置3係使用由CCD攝像機2輪出之標的圖像，：查 是否在基板S正禮形成凸塊（凸塊是否以預定之精密度^ 成），並輸出檢查結果。 ^ ^ 如第1圖所示，圖像處理裝置3係具備2值化處理部 3卜標的圖像記憶部32、形狀檢測部％、凸塊位置記憶部 3:、比較檢查部36、主圖像記憶部37及檢查區域圖像記 ΙΉ 38°圖像處理裝置3典型上係由個人電腦所構成，圖 ，處理裝置之處理係藉由執行儲存於圖像處理I置3之預 ，程式而進行。圖像處理裝置之處理的—部分或全部亦可 藉由專用之處理電路而進行。 2值化處理部31係對從CCD攝像機2輸出之標的圖 像資料進行2值化處理。亦即，2值化處理部⑴系使標的 圖像雙換為以2值表現各像素(pixel)的2值圖像。在變換 後之標的圖像資料中，存在有凸塊之位置係以“丨，，表現、， 不存在凸塊之位置係以“〇，，表現。標的圖像記憶部”係 將經由2值化處理部3丨變換後所得之標的圖像資料予以呓 憶。凸塊位置記憶部34係將表示應形成在基板8之各凸D 位置的資料（凸塊位置資料）予以記憶。該凸塊位置 資料係由例如檢查者所預先作成且記憶在凸塊位置記憶部 34。又，凸塊位置資料只要是特定各凸塊之位置者，亦可 為任意者。例如，亦可為表示各凸塊之中心位置的座標資 料’亦可為圖像資料。例如，亦可將後述之主圖像資料 11 317686 1274869 —(masterimagedata)作為凸塊位置資料使用。 心：：二"33係檢測出包含於標的圖像之凸塊圖像 (表不凸塊雪圖像）中，呈古丁+ 兄口俅 實施形態中，凸塊係圓錐形石:塊形狀之圖像。在本 標的圖像中顯示為圓形區域麵成之凸塊係在 ，人， 亦即在本實施形態中，形肤 聰表示所檢测出之凸塊圖像之區域(標：: =域）的貝枓輸出至檢查區域設定部％。且 查區域35設Μ通知非正確形狀之凸塊圖像的區域。 檢查區域設定部3 5係對於標的圖像設定作為比較檢 2之對象的檢查區域、及非作為比較檢查之對象的檢查區 :。具體而言，檢查區域設定部35係藉由進行用以擴散正 確形狀之凸塊圖像之區域的處理（擴散處理），而產生表示 ^查區域及非檢查區域之圖像。更具體而言，檢查區域設 疋4 35係產生表不作為非檢查區域之圖像的圖像資料，該 非檢查區域係使正確形狀之凸塊圖像的區域僅擴展預定範 圍的區域。在本實施形態中，係將該圖像稱為第2檢查區 域圖像’將顯示第2檢查區域圖像之圖像資料稱為第2檢 查區域圖像資料。第2檢查區域圖像資料係與標的圖像資 料同樣地，以2值表現各像素的資料。例如，在第2檢查 區域圖像中，卩表現非檢查區域，卩1，’表現檢查 區域。 另方面，主圖像記憶部37係用以記憶主圖像資料。 主圖像資料係指在比較檢查處理中顯示作為基準之基板圖 像（主圖像）的資料。主圖像係理想地形成凸塊之基板圖 317686 12 1274869 像主圖像貝料係可藉由攝像良品所得之圖像資料，亦可 為由⑽㈣所作成之圖像資料。主圖像諸係與標的 圖像資料同樣地…值表現各像素的資料。在主圖像資 料中，與標的圖像資料同樣m”表現存在有凸塊之 位置’以“0”表現未存在有凸塊之位置。此外，檢查區域 圖，記憶部38係記憶表示第1檢查區域圖像的圖像資料 (第1檢查區域圖像資料)。第1檢查區域圖像資料係在主 圖像之區域中，指定作比較檢查之對象之區域的圖像資 料。第1檢查區域圖像資料係與第2檢查區域圖像資料同 ‘地’以2值表現各像素的資料，以“G，，表現 域，以“1”表現檢查區域。 一 一比較檢查部36係採用前述主圖像及前述標的圖像進 :對基板S之比較檢查處理。亦即，比較檢查部％係進 行主圖像與払的圖像之圖案匹配（paUem matching)，而檢 測出包含於兩者之凸塊圖像的相異點。比較檢查部％係根 所檢測出之相異點’判斷是否以所希望之精密度將各凸 塊$成在基板。例如，凸塊圖像之相異點的數量比預設之 預疋數少時，判斷以所希望之精密度將各凸塊形成在基 板，在该預定數以上時，判斷並未以所希望之精密度將各 凸塊形成在基板。 再者，在本實施形態中，比較檢查部36並非僅使用 主圖像與標的圖像，而使用第丨及第2檢查區域圖像遮蔽 各圖像（主圖像與標的圖像）後，進行比較處理檢查。具 —s ’主圖像之檢查區域係由第1檢查區域圖像所規 317686 13 1274869 疋，標的圖像之檢查區域係由第2檢查區域圖像所規定。 亦即，主圖像係在第1檢查區域圖像中，僅作為檢查區域 之區域成為比較檢查之對象。而且，標的圖像係在第丨檢 查區域圖像巾，僅將作為檢查區域之區域絲比較檢查^ 理之對象。結果，作為比較檢查處理之對象的區域係在第 1檢查區域圖像與第2檢查區域圖像之雙方，成為設為檢 查區域的區域。 以下，說明本實施形態之凸塊檢查裝置的動作。以 下’係以檢查第1至第4凸塊之4個凸塊是否形成在基板 的情形為例加以說明。帛2圖係表示正確形成之基板的一 例圖。如第2圖所示’在正確作成之基板形成有第i至 4凸塊41至44。 第3圖係凸塊檢查裝置之圖像處理裝置之處理流程的 2圖。在第3圖所示之處理前，咖攝像機2係預先攝 置在口 1之基板S之圖像。首先，在步驟S1中，取 #得^的圖像。具體而言，藉由攝像載置在台！上之基板s 所得之甘標的圖像資料係從CCD攝像機2在2值化處理部 31取仔。在2值化處理部31取得之標的圖像資料係變換 ^ 2值表現各像素之像素資料。變換後之標的圖像資料 係δ己憶在標的圖像記憶部32。1274869 IX. Description of the Invention: [Technical Field] The present invention relates to a bump inspection apparatus and method for inspecting a bump formed on a substrate, which is a method of checking whether a bump is formed with a predetermined precision A bump inspection device for a substrate and a bump inspection method. [Prior Art] In the past, as a method of connecting a layer of a substrate (fourth) to a substrate (fourth), there is a method of connecting with a bump. That is, for example, a block-shaped block is formed on the substrate, and the bump-formed substrate and other substrates are transferred, whereby the bumps are penetrated through the substrate to connect the substrates to each other. As described above, when the bump is formed on the substrate, it is necessary to form the bump with good precision with respect to the position at which the bump is formed and the shape of the bump. Therefore, it is necessary to inspect the substrate after the bump is formed. In order to check the method of raising/forming the substrate, there is a method of comparing the image of the substrate of the inspection object (the object image of the object to be examined) with the image of the substrate (the main image) as a sample (refer to, for example, Japanese Patent Laid-Open Publication No. 2001_102761). In this method, a substrate on which bumps are formed is imaged from above, and a target image of a binary image is obtained. Further, a main image of a binary image is prepared in advance from CAD data or the like. Then, by comparing the target image with the main image, the phase difference between the two is detected, and it is judged based on the different portion whether or not the desired precision is formed (10). Here, usually, a plurality of bumps are formed on i substrates, so that the bumps are individually compared one by one for inspection, but it is preferable to compare a plurality of bumps at the same time for inspection. Hereinafter, the inspection process when comparing a plurality of bumps simultaneously is described using FIG. 12a to FIG. 317686 5 1274869. In the double check process, the main image and the target image are first prepared. The image of the image when comparing a plurality of bumps at the same time is shown in Fig. 3: Fig. 3 to Fig. 12°, which shows the simultaneous comparison of the shape of the first tail. The 罘12Α image shows an image 93 of the example 6 of the main image. The 帛12β image shows the target image: the image contains the first bump _, the second bump image 95, and the ridge 3 image 96. Prepare the alignment of the main image and the target image silly image, and then place the car bump on the appropriate main image and the target ^ L. For example, the amount (area) of the distinct portion in the bump of the r is the smallest: . After riding the alignment, compare the main image with the standard W image, :: each: the different parts of the block. For example, the area ratio of the dissimilar parts is, therefore, the offset is caused by the dissimilar parts =:: the bumps, and the bumps which are not formed into the correct shape and have a large reference amount of the different knives are judged as defects. . Fig. 12C is a schematic diagram showing the main image and the target image superimposed. In the first place, the figure is easy to understand, and the circle of the main image is displayed by a dotted line, and the circle of the target image is ==. In Fig. 12C, the different parts are measured separately in the i-th ghost. And determining whether each bump is defective based on the detected image. The phase "knife is here" actually bumps even if there is a certain degree of positional deviation, only 317686 6 1274869: shape, the correct shape of the material is considered a defect. For example, the diameter of the valley 1 (four) about the position deviation. In the 12A to the Nc : in the middle, the second bump has a positional deviation, but there is no defect. On the other hand, the shape must be more than: === grid, and the bump itself does not have the predetermined shape: That is, it is judged to be a defect. In the example of Fig. 12A to Fig. 1, the second bump is judged to be a defect because it does not form a correct shape. From the above, the allowable range of the positional deviation of the bump is known. It can be said that it is wider than the allowable range of the shape of the bump. That is, the position of the bump is set to be larger than the position of the top of the bump, and the reference amount should be set with respect to the shape of the bump. In the conventional method, the two parts of the position and the shape are checked with the reference amount of t, so that it is impossible to perform the correct k 2 for the towel, that is, the reference amount is set. When it is smaller, about The allowable range of the deviation is set to be narrower, so even if there is more or less positional deviation, Φ will cause the bump which is not originally defective (the second bump of the 12A to 12C) is also judged as Conversely, when the reference amount is set to be large, the allowable range of the shape is set to be wide, so that the bumps of (4) which are not originally allowed (the third bumps of FIGS. 12A to 12C) are made. It is also judged to be non-defective. Therefore, in the conventional inspection method, it is impossible to accurately judge the presence or absence of the defect with respect to the position and shape of the bump. [Invention] Therefore, it is an object of the present invention to provide a bump for the bump. The bump inspection device and the bump inspection for accurately determining the presence or absence of a defect are as follows. In order to solve the above problems, the present invention adopts the above-described configuration. That is, the first aspect is formed by inspection. The bump inspection device of the bump of the substrate includes a camera portion that images the substrate, and an image that is not included in the bump of the target image obtained by the image capturing portion. a shape detecting unit that extracts an image having a correct bump shape; a region including at least an image detected by the shape detecting unit as a non-inspection region, and a 102 other region as an inspection region setting portion of the inspection region; Memorize the main image storage unit as the main image of the inspection and the reference, and check the convexity by comparing the target image and the main image with the inspection area set by the inspection area and the setting unit. A comparison inspection unit formed with a predetermined degree of precision. In the second sad sample, the inspection area setting unit can set the image area and the area around the area detected by the shape detection unit as non-inspection areas. In the third aspect, the bump inspection apparatus may further include a detection area memory unit for storing data of a region in the main image in which the comparison object of the comparison inspection unit is displayed. At this time, the comparison inspection unit compares the main image and the target image with respect to the area indicated by the data of the examination area setting and the data stored in the inspection area storage unit. In the fourth aspect of the day, the data stored in the memory of the inspection area may be such that it should not be formed in the area corresponding to the bump of the substrate. At this time, the main image has an image in which a part of the image of the bump formed on the substrate has a defect. Further, the fifth aspect is a bump inspection method for inspecting the bumps formed on the substrate. The bump inspection method includes: an image capturing step of imaging the substrate; and detecting an image having a correct bump shape from an image showing the bump 317686 8 1274869 included in the target image obtained by the image capturing step a shape detecting step; a region including at least an image detected by the shape detecting step as a non-inspecting region, and setting other regions as an inspection region setting step of the inspection region, and setting by the inspection region setting step The inspection area compares the target image with the main image, and checks whether the bump is formed with a predetermined degree of precision. • According to the first aspect and the fifth aspect, the bump image of the correct shape is detected in advance, and the area of the detected bump image is set as the non-check area ‘domain. Further, by comparing the inspection image of the inspection area with the main image - the inspection can be performed to detect the bump having an incorrect shape and the projection whose position is deviated from the allowable amount as a defect. Here, the allowable range of the shape of the bump is determined in the process of detecting the bump image of the correct shape, and the position of the bump on the position of the bump is determined in the process of setting the non-inspected area. . Therefore, according to the present invention, since the allowable range of the shape and position of the bump can be appropriately set independently, it is possible to determine the presence or absence of the position and shape of the bump. Further, according to the present invention, since the conventional method of comparative inspection is employed, it is possible to accurately judge the presence or absence of defects without using a special comparative inspection operation. According to the second aspect described above, the allowable range with respect to the positional deviation can be expanded as a, and there is no problem even if the allowable range of the positional deviation of the bump is enlarged, so that the allowable range with respect to the positional deviation is expanded. Conduct a check based on the facts. According to the third aspect described above, an area which is a comparison check can be set for the main image, and thus the area to be subjected to the comparison check can be freely set. 317686 9 1274869 • According to the fourth aspect described above, the target image is correctly obtained by the reflection I method of the substrate during imaging, and the result of the comparison inspection becomes a defect. Therefore, even if only the region of the bump (ideal bump) to be formed on the substrate is used as the object of the comparison inspection, a satisfactory inspection result can be obtained for the target image which cannot be correctly obtained. Furthermore, since it is only necessary to compare the areas of the ideal bumps, it is possible to reduce the processing amount of the comparison check. The above effects and other objects, features, aspects and functions of the present invention will be further understood from the following detailed description. [Embodiment] (First Embodiment) Fig. 1 is a block diagram showing a configuration of a bump inspection apparatus according to a first embodiment. The bump inspection device includes a pedestal 1, a CCD (charge-coupled device) camera 2, and an image processing device bump. The inspection device f1 is used to check whether or not the bump is correctly formed on the substrate s. In Fig. 1, the pedestal 1 mounts the substrate S to be inspected on top of φ. A bump of, for example, a bell shape is formed on the substrate 8. The bumps formed on the substrate S are generally of a size of 1 〇〇i 5 〇//m and a height __. The CCD camera 2 is an example of an imaging means of the imaging substrate S, and the substrate s placed on the upper imaging port holder. The image data (digital data) that has been imaged is used in the comparison check processing described later. In the present embodiment, the image obtained by the CCD camera 2 is referred to as a target image (10) (4) ge) "The data representing the target image is referred to as the target image data (_img=data), and if not, When the substrate § is fully imaged once, the CCD camera 2 and/or the pedestal i can be moved to change the image 317686 10 1274869: and the entire substrate s is imaged. The CCD camera 2 outputs the image data of the target substrate S to the image processing device 3. The processing device 3 uses the image of the target rotated by the CCD camera 2 to check whether a bump is formed on the substrate S ( Whether the bump is at a predetermined precision and outputting the inspection result. ^ ^ As shown in Fig. 1, the image processing device 3 includes an image storage unit 32, a shape detecting unit %, a bump position memory unit 3, a comparison check unit 36, and a main image of the binarization processing unit 3. The image processing device 3 is typically composed of a personal computer. The processing of the processing device is performed by executing a program stored in the image processing I. get on. Part or all of the processing of the image processing apparatus can also be performed by a dedicated processing circuit. The binarization processing unit 31 performs binarization processing on the target image data output from the CCD camera 2. In other words, the binarization processing unit (1) converts the target image into a binary image in which each pixel (pixel) is represented by two values. In the transformed image data, the position of the bump is "丨,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, The target image storage unit "reviews the target image data obtained by the binarization processing unit 3". The bump position memory portion 34 displays data indicating the position of each convex D to be formed on the substrate 8 ( The bump position data is memorized. The bump position data is prepared in advance by the examiner and stored in the bump position memory unit 34. Further, the bump position data may be any position of the specific bumps, or For example, the coordinate data 'which indicates the center position of each bump' may also be image data. For example, the main image data 11 317686 1274869 - (masterimage data) described later may be used as the bump position data. Heart:: The second "33 system detects the bump image contained in the target image (the image is not the snow image of the bump), in the form of the Guding + Xiongkou , implementation, the convex block is a conical stone: block The image of the shape is displayed in the image of the target as a convex portion of the circular area, and in the present embodiment, the skin is represented by the area of the detected bump image ( Mark:: = domain) Bellow output to the inspection area The setting unit % is provided, and the inspection area 35 is provided with an area for notifying the bump image of the incorrect shape. The inspection area setting unit 3 sets the inspection area which is the target of the comparison check 2 for the target image, and is not used as the comparison check. Inspection area of the object: Specifically, the inspection area setting unit 35 generates a map indicating the inspection area and the non-inspection area by performing processing (diffusion processing) for diffusing the area of the bump image of the correct shape. More specifically, the inspection area setting image produces image data that is not an image of the non-inspection area, and the non-inspection area is such that the area of the bump image of the correct shape extends only the area of the predetermined range. In the present embodiment, the image is referred to as a second inspection region image. The image data for displaying the second inspection region image is referred to as a second inspection region image data. The second inspection region image data system and Similarly, in the target image data, the data of each pixel is expressed by two values. For example, in the second inspection area image, 卩 indicates the non-inspection area, 卩1, 'the performance inspection area. On the other hand, the main image memory The main image data refers to the material of the substrate image (main image) displayed as a reference in the comparison inspection process. The main image is a substrate pattern in which the bump is ideally formed. 317686 12 1274869 Image data obtained by the image of the main image can be obtained by the camera, or image data made by (10) (4). The main image is the same as the image data of the target image. In the main image data, the same m" as the target image data indicates that there is a position where the bump exists, and "0" indicates that there is no position where the bump exists. Further, in the inspection area map, the memory unit 38 stores image data (first inspection area image data) indicating the image of the first inspection area. The first inspection area image data is an image area of the area to be subjected to the comparison inspection in the area of the main image. The image data of the first inspection area and the image data of the second inspection area are the same as 'ground', and the data of each pixel is represented by two values, and the inspection area is expressed by "G", the expression field, and "1". In the 36th system, the main image and the target image are used for comparison inspection processing on the substrate S. That is, the comparison inspection unit % performs pattern matching (paUem matching) of the main image and the image of the 払, and detects The difference point between the bump images included in the two. The comparison check unit % detects the different points detected by the roots to determine whether the bumps are formed on the substrate with a desired precision. For example, the bump map When the number of different points is smaller than the preset number of pre-turns, it is judged that the bumps are formed on the substrate with a desired precision, and when the predetermined number is equal to or greater than the predetermined number, it is judged that the respective precisions are not used. The bump is formed on the substrate. In the present embodiment, the comparison inspection unit 36 does not use only the main image and the target image, but masks each image using the second and second inspection region images (main image and After the target image), the comparison processing check is performed. With -s ' The inspection area of the main image is defined by the first inspection area image 317686 13 1274869 疋, and the inspection area of the target image is defined by the second inspection area image. That is, the main image is in the first inspection area. In the image, only the area to be the inspection area is the object of the comparison inspection. Moreover, the target image is the image towel in the second inspection area, and only the area of the inspection area is compared and examined. The area to be inspected is the area which is the inspection area in both the first inspection area image and the second inspection area image. Hereinafter, the operation of the bump inspection apparatus according to the present embodiment will be described. The case where the four bumps of the first to fourth bumps are formed on the substrate is described as an example. The figure 2 shows an example of a substrate that is correctly formed. As shown in Fig. 2, the substrate is formed correctly. There are ith to fourth bumps 41 to 44. Fig. 3 is a view showing a processing flow of the image processing apparatus of the bump inspection apparatus. Before the processing shown in Fig. 3, the coffee camera 2 is pre-positioned at the mouth. 1 of the substrate S First, in step S1, an image obtained by # is obtained. Specifically, the image data of the image obtained by imaging the substrate s placed on the table is from the CCD camera 2 in the binarization processing unit. 31. The target image data obtained by the binarization processing unit 31 converts the pixel data of each pixel, and the converted image data system δ is recalled to the target image storage unit 32.
弟2圖係由表示步驟S1所獲得之標的圖像的示意 ::在弟4圖所示之標的圖像中，包含有顯示第！凸塊之 =像46、顯示第2凸塊之凸塊圖像47、顯示第3凸塊 之^圖像48、顯不第4凸塊之凸塊圖像49。在基板S 317686 14 1274869 :，從第4圖之標的圖像可知，第丨凸塊係大致正常形成， 第2凸塊雖然形狀正確，但位置略偏離而形成，第3凸塊 係形成缺口形狀，第4凸塊雖然形狀正確，但位置偏離甚 夕而形成。且在本實施形態中，係以將第〗及第2凸塊判 斷為正常、將第3及第4凸塊判斷為缺陷當作目標。亦即， 谷汴如第2凸塊之位置略偏離，但如第3凸塊形狀不正確 -者、或如第4凸塊位置大幅偏離者，即判斷為缺陷。 ，接著在步驟S2中，使用標的圖像在形成於基板§之 各凸塊中檢測出正確形狀之凸塊。具體而言，形狀檢測部 33係針對包含於標的圖像之各凸塊圖像判定是否為正確 /狀（在此為圓形）。以形狀檢測部33進行之判定處理係以 二列如下述方式進行。亦則，首先根據記憶在凸塊位置記憶 P 3 4之凸塊位置資料，特定應形成有各凸塊之位置（形成 有凸塊之理想位置）。然後，判定存在於特定之位置的凸塊 圖像，狀是否為圓形。是否為圓形之判定係對預設之複數 .方向算出凸塊圖像之區域的直徑，並判定所算出之直徑各 ，是否在預定範圍内。亦即，所算出之直徑的各值在預定 f圍内日τ，判定為圓形’而所算出之直徑的各值沒有在預 疋粑圍内時，則判定為非圓形。表示該預定範圍之值最好 由檢查者自由地設定。藉由使該值變化，檢查者可使凸塊 位置相關之容許量變化。而是否為圓形之判定係分別對包 含有凸塊位置資料所示之位置的區域進行。以第4圖之例 而言，係對顯示第i至第4凸塊之各凸塊圖像進行該判定。 此外，形狀檢測冑33係、將表示判定為正確形狀之凸塊圖像 317686 15 1274869 的資料任送至檢查區域設定部35。該資料可為任一形態， 在本貝知形怨中，係没為圖像資料。具體而言，設定為 包含在標的圖像資料之各凸塊圖像中僅包含判定為正確形 狀之凸塊圖像的圖像資料。 在步驟S3中，作成將在步驟S2檢測出之凸塊圖像之 區域作為非檢查區域的遮罩。具體而言，檢查區域設定部 • 35係首先特定包含於從形狀檢測部％接收之圖像資料之 各凸塊圖像的中心位置。然後，將以特定之位置為中心的 .圓形區域作為非檢查區域，並作成表示以其他區域為檢查 • 區域之遮罩的圖像。The second drawing is indicated by the image indicating the target image obtained in step S1. :: The target image shown in the fourth drawing includes the display! The bump = image 46, the bump image 47 showing the second bump, the image 48 showing the third bump, and the bump image 49 showing the fourth bump. In the substrate S 317686 14 1274869 : from the image of the standard image in Fig. 4, the second bump is formed substantially normally, and the second bump is formed in a slightly offset shape, but the third bump is formed in a notched shape. Although the fourth bump is shaped correctly, the position is deviated. Further, in the present embodiment, the first and second bumps are judged to be normal, and the third and fourth bumps are judged to be defects. That is, the position of the valley bump is slightly deviated as in the case of the second bump, but if the shape of the third bump is not correct, or if the position of the fourth bump is largely deviated, it is judged to be a defect. Then, in step S2, bumps of the correct shape are detected in the bumps formed on the substrate § using the target image. Specifically, the shape detecting unit 33 determines whether or not the bump image included in the target image is correct/shaped (here, circular). The determination processing by the shape detecting unit 33 is performed in the following two rows as follows. In other words, firstly, according to the position data of the bumps of the memory P 3 4 in the position of the bump, the position of each bump (the ideal position where the bump is formed) should be specifically formed. Then, it is determined whether or not the bump image exists at a specific position, and whether the shape is a circle. Whether or not the circle is determined is the diameter of the region of the bump image for the predetermined plural number, and it is determined whether or not the calculated diameter is within a predetermined range. In other words, when the values of the calculated diameters are determined to be circular in the predetermined f-day τ and the values of the calculated diameters are not within the predetermined range, it is determined to be non-circular. It is preferable that the value of the predetermined range is freely set by the examiner. By changing this value, the examiner can vary the allowable amount of bump position correlation. The determination as to whether or not the circle is performed is performed on the region containing the position indicated by the bump position data. In the example of Fig. 4, this determination is made for each of the bump images showing the i-th to fourth bumps. Further, the shape detecting unit 33 transmits the data indicating the bump image 317686 15 1274869 determined to be the correct shape to the inspection region setting unit 35. This information can be in any form, and in the case of this knowledge, it is not image data. Specifically, it is set as image data including only the bump image determined to be in the correct shape among the bump images of the target image data. In step S3, a mask having the region of the bump image detected in step S2 as a non-inspection region is created. Specifically, the inspection area setting unit 35 first specifies the center position of each of the bump images included in the image data received from the shape detecting unit %. Then, a circular area centered on a specific position is used as a non-inspection area, and an image indicating that the other area is a mask of the inspection area is created.
a接著在步驟S4中，藉由擴散在步驟S3作成之圖像之 非檢查區域，而作成前述第2檢查區域圖像。亦即，檢查 區域設定部35係使在步驟S3作成之圖像資料之非檢查& 域的直徑僅增大預定長度。由此，作成第2檢查區域圖像。 而^ ’擴散非檢查區域之擴散量（上述預定長度）係預先 、疋者°亥擴政里係作為顯示凸塊位置偏離相關之容旦 =指標°因此’擴散量之值最好係可由檢查者予以變更里 第5,係表示第2檢查區域圖像的示意圖。在第$圖，表 :由弟3圖所示之標的圖像所得之第2檢查區域圖像。在 ==中,線區域係檢查區域。且第5圖之虛線係表示 為正確形狀之凸塊圖像之區域，亦即在步驟S3中# ，之非檢查區。如第5圖所示，獲得 的又 像時，表示第1凸塊、第2凸塊及第4凸塊之凸 區域及其周邊區域係非檢查區域。 " 317686 16 1274869 _ 在步驟S5中，進行標的圖像與主圖像之比較檢查。 步驟S5之處理係使用以下資料而進行··記憶在標的圖像 °己丨思°卩32之標的圖像資料；由檢查區域設定部35所產生 之第2檢查區域圖像資料；記憶在主圖像記憶部之主圖像 貝料，記憶在檢查區域圖像記憶部38之第1檢查區域圖像 貝料。第6圖係本實施形態中所使用之主圖像的示意圖。 第7圖係本實施形態中所使用之第1檢查區域圖像的示意 _圖。如第6圖所示，在本實施形態中，係將表示正確形成 -守之σ凸塊（理想凸塊）用作為主圖像（master image)。且 -如第7圖所示，本實施形態之第1檢查區域圖像資料係將 全面作為檢查區域之圖像資料。以下，詳細說明步驟S5 之比較檢查部3 6之處理。 比車父檢查部36係決定用以對標的圖像及主圖像進行 比車父檢查之區域。具體而言，標的圖像之比較檢查之對象 的區域係由第2檢查區域圖像所規定。亦即，標的圖像中 _將在第2檢查區域圖像中作為檢查區域的區域作為比較檢 查的對象。以弟4圖及弟5圖為例加以說明之，表示第4 圖所示之第1、第2及第4凸塊之凸塊圖像的區域及其周 邊區域並不成為比較檢查之對象。另一方面，在主圖像之 作為比較檢查之對象區域係由第1檢查區域圖像所規定。 在本μ知形恕中’弟1檢查區域圖像之全面係檢杳區域（東 照第7圖），因此主圖像之全面即成為比較檢查的對象。 決定用以進行比較檢查的區域以後，比較檢查部36 係對標的圖像與主圖像進行對準檢查區域内之凸塊圖像之 317686 17 1274869 .位置的處理。在對位處理中，例如，在作為比輕檢查之對 象的全區域中，係以使標的圖像與主圖像之不同點的量（面 積、亦即像素數）成為最小之方式來決定位置。不同點係 指因標的圖像與主圖像而數值（G或υ不同之像素。對位 處理後’比較檢查部3 6係進行標的圖像與主目像之比較檢 查。至於僅在標的圖像與主圖像之任一方作為比較檢查之 對象=區域，並不進行比較檢查。亦即，比較檢查部％ 一係在第1檢查區域圖像及第2檢查區域圖像之雙方就作為 檢查區域的區域進行比較檢查。第8圖係用以說明進行比 較檢查之區域的示意圖。在第8圖中，重疊記载包含於標 的圖像之凸塊圖像與包含於主圖像之凸塊圖像。在第8圖 中，虛線之内側之區域以外的區域係作為比較檢查之對 ^因此’獲得第3圖所示之標的圖像時，並不^凸塊圖 46與凸塊圖像51進行比較檢查。且對凸塊圖像〇與凸 2圖像52並不進行比較檢查。至於凸塊圖像49與凸塊圖 像54,雖然凸塊圖像49係存在於非檢查區域内，但因凸 ▲ Θ像4之邛为存在於非檢查區域的外侧，故在該 進行比較檢查。 —在比較檢查中，比較檢查部乂係對表示ι個凸塊之 ::凸塊圖像輪出檢查結果。具體而言，係在各凸塊檢 :預：=，主圖像之不同點’當不同點之數量比預設 ^即判斷該凸塊為正常，而在預定數以上時， 、、断该凸塊為缺陷。以第8圖為例加以說 凸塊圖像48及凸塊圖像53進行比較檢查，可得到= 317686 18 1274869 塊相關之檢杳έ士 I 各 女从μ —、、、°果。在此，凸塊圖像48係為阊弗夕, 有缺陷之形貼丄 你馬圓形之一部分 ’由於有缺陷之部分，盥盔缺 較，不同點之數奮“夕门 缺陷之情形相比 果係為「缺陷Γ交此，有關第3凸塊之檢查結 測部33判斷為非正確形狀之凸疋為將由形狀檢 數值。 ？狀之凸塊圖像予以判斷為缺陷的 過m〜 因此，可認為不同點之數量可超 =預定數，而第4凸塊之檢查結果為「缺陷」。如此 主圖像之凸塊®像的位置超過第2檢查區域圖像之 的範圍内時’對應該凸塊圖像之凸塊係判斷為 缺^。因此，有關凸塊位置偏離之容許量係由第2檢查區 域圖像之非檢查區域之·所決定。亦即，由前述錢量 所決定。例如，將擴散量設定為較大時，# 2檢查區域圖 像之非檢查區域的範圍會變大，因此有關凸塊位置偏離相 關之容許量會變大。相反地，將擴散量設為較小時，第2 才双查區域圖像之非檢查區域的範圍會變小，因此有關凸塊 位置偏離相關之容許量會變小。 再者，步驟S5之比較檢查的具體方法係根據標的圖 像與主圖像之不同點的方法即可，可適用習知之比較檢杳 運算法（algorism)。 再者，比較檢查部係將未成為比較檢查之對象之凸塊 判斷為「正常」。在第8圖中，第1及第2凸塊並未成為比 較檢查之對象’因此第1及第2凸塊之檢查結果為「正常」。 317686 19 1274869a Next, in step S4, the second inspection area image is created by diffusing the non-inspection area of the image created in step S3. That is, the inspection area setting unit 35 increases the diameter of the non-inspection & field of the image data created in step S3 by only a predetermined length. Thereby, the second inspection area image is created. And the diffusion amount of the non-inspection area (the above-mentioned predetermined length) is pre-existing, and the degree of diffusion is the value of the diffusion amount. The fifth change is a schematic diagram showing the image of the second inspection area. In the figure $, the table shows the image of the second inspection area obtained from the target image shown in the figure of the third figure. In ==, the line area is the inspection area. And the broken line of Fig. 5 indicates the area of the bump image of the correct shape, that is, the non-inspection area of # in step S3. As shown in Fig. 5, the obtained image shows that the convex regions of the first bump, the second bump, and the fourth bump and the peripheral region thereof are non-inspection regions. " 317686 16 1274869 _ In step S5, a comparison check of the target image and the main image is performed. The processing of step S5 is performed by using the following data: the image data of the target image recorded in the target image; the image data of the second inspection area generated by the inspection area setting unit 35; The main image of the image memory unit is stored in the image of the first inspection area of the inspection area image storage unit 38. Fig. 6 is a schematic view showing a main image used in the present embodiment. Fig. 7 is a schematic view showing an image of a first inspection region used in the present embodiment. As shown in Fig. 6, in the present embodiment, a sigma bump (ideal bump) indicating that the sigma is correctly formed is used as a master image. Further, as shown in Fig. 7, the first inspection area image data of the present embodiment is comprehensively used as image data of the inspection area. Hereinafter, the processing of the comparison checking unit 36 in step S5 will be described in detail. The area in which the target inspection unit 36 determines the target image and the main image to be compared with the vehicle parent is determined. Specifically, the area of the target of the comparison check of the target image is defined by the image of the second inspection area. That is, the area in the target image as the inspection area in the second inspection area image is used as the comparison check object. Taking the brothers 4 and 5 as an example, the region of the bump image indicating the first, second, and fourth bumps shown in Fig. 4 and the peripheral region thereof are not subject to comparison inspection. On the other hand, the target area to be compared and checked in the main image is defined by the first inspection area image. In this case, the comprehensive inspection area of the image of the area 1 is examined (the picture is shown in Fig. 7), so the overall image of the main image becomes the object of comparison inspection. After determining the area for performing the comparison check, the comparison inspection unit 36 performs the process of aligning the target image with the main image by 318686 17 1274869 in the alignment image in the inspection area. In the alignment processing, for example, in the entire area which is the object of the light inspection, the position is determined such that the amount (area, that is, the number of pixels) of the difference between the target image and the main image is minimized. . The difference point refers to the value of the target image and the main image (G or υ different pixels. After the registration process, the comparison check unit 36 performs a comparison check between the target image and the main image. As for the target image only The comparison check unit does not perform the comparison check as the object to be compared with the main image. That is, the comparison inspection unit % is used as the inspection in both the first inspection area image and the second inspection area image. The area of the area is subjected to a comparison check. Fig. 8 is a schematic view for explaining the area where the comparison check is performed. In Fig. 8, the bump image included in the target image and the bump included in the main image are overlapped. In Fig. 8, the area other than the area inside the broken line is used as the comparison check. Therefore, when the image of the target shown in Fig. 3 is obtained, the bump image 46 and the bump image are not. 51 performs a comparison check, and does not perform a comparison check on the bump image 〇 and the convex 2 image 52. As for the bump image 49 and the bump image 54, although the bump image 49 exists in the non-inspection area , but because of the convex ▲ Θ 4 like the presence of 4 in the non-inspected area On the side, the comparison check is performed. - In the comparison check, the comparison check unit 表示 indicates that the bump image is rotated by the :: bump image. Specifically, it is detected in each bump: :=, the difference of the main image 'When the number of different points is greater than the preset ^, it is judged that the bump is normal, and when it is more than a predetermined number, the bump is broken as a defect. Take the figure 8 as an example. It can be seen that the bump image 48 and the bump image 53 are compared and checked to obtain = 317686 18 1274869 blocks related to the inspection of the gentleman I each female from μ —, ,, ° fruit. Here, the bump image 48 is For the 阊 夕 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , The indentation of the third bump is determined to be an inaccurate shape. The convexity of the shape is a value of the shape of the bump. The image of the bump is judged to be a defective m~ Therefore, the number of different points can be considered Super = predetermined number, and the inspection result of the 4th bump is "defect". Thus the bump image of the main image is like When the range of the image of the second inspection area is exceeded, the bump corresponding to the bump image is judged to be absent. Therefore, the tolerance for the positional deviation of the bump is not checked by the image of the second inspection area. The area is determined by the amount of money. For example, when the amount of diffusion is set to be large, the range of the non-inspection area of the #2 inspection area image becomes large, so the position of the bump is deviated. On the other hand, when the amount of diffusion is set to be small, the range of the non-inspection area of the second double-check area image becomes small, so the tolerance for the positional deviation of the bump becomes small. Furthermore, the specific method of the comparison check in step S5 is a method based on the difference between the target image and the main image, and a conventional comparison check algorithm (algorism) can be applied. Further, the comparison inspection unit judges that the bump which is not the object of the comparison inspection is "normal". In Fig. 8, the first and second bumps are not subjected to the comparison inspection. Therefore, the inspection results of the first and second bumps are "normal". 317686 19 1274869
此外，在本實施形態巾，設比較檢查之結果為作為圖 像而，顯示裝置（未圖示）。例如，比較檢查部36亦 可將&合標的圖像與主圖像之圖像顯示在顯示裝置。此 時’利用判斷為正常之凸塊、與判斷為缺陷之凸塊變更顏 ^的話，可區別顯示正常之凸塊與缺陷之凸塊。且亦可就 變合標的圖像與主圖像之圖像中、僅顯示前述第2檢查區 域圖像之檢查區域内。此時’正常之凸塊係存在於非檢查 區域内：因此未顯示，而僅顯示缺陷之凸塊。由此，亦可 區別喊不正t之凸塊與缺陷之凸塊。比較檢查之結果的提 不方法’只要是對作業者提示各凸塊之正常或缺陷的方法 ^可，而可為任—之方法，但使作業者最後判斷是否缺陷 日守，最好是如上所述顯示為圖像之方法。 如上所述，根據本實施形態，凸塊檢查裝置係首先進 行用以識別凸塊之形狀的處理（前述步驟s2)。藉由該處 理’就形狀檢測出判斷為正常之凸塊（例如前述第i及第 鬼），、、u复凸塊檢查裝置係進行標的圖像與主圖像之 比較檢查（前述步驟S5)，而檢測出形成為正相狀之凸 塊、及偏離為可容許位置之程度而形成的凸塊。如此，藉 由個別進㈣塊之形狀相關的判斷及位置相關的判斷，可 们別地4疋凸塊之形狀相關的判斷之臨界值恤 va㈣及位置相關的判斷之臨界值。具體而言，藉由使顯 ==2之預定範圍的值變化，可使凸塊之形狀相關的 二:之位S’藉由使步驟S4之預定長度的值變化， 了使凸鬼之位置相關的容許值變化。根據本實施形態，可 317686 20 Ϊ274869 適當地設定形狀及位置相關之臨界值，因此可正確地判斷 凸塊之位置及形狀是否有缺陷。 (第2實施形態） 以下’說明本發明之第2實施形態之凸塊檢查裝置。 在第2實施形態中，除了前述第！檢查區域圖像及主圖像 之内容與第1實施形態的不同之外，其他與第1實施形態 相同。亦即，凸塊檢查裝置之基本構成（第j圖）及動作 (第3圖）係與第1實施形態相同。在以下之說明中，作 為本之基板位置及开》狀（第2圖）及所取得之標的圖 (第4圖）係與第1實施形態相同。 第9圖係表示第2實施形態中所使用之第丨檢查區域 圖像的示意圖。如第9圖所示’在第2實施形態中，僅將 應形成有凸塊之區域（形成有凸塊之理想區域）設為檢查 區域。因&，可將比較檢查處理抑制在最少限度，減輕圖 像處理裝置3之處理負擔，可在短時間進行處理。 在此，在前述第1實施形態中，包含於主圖像之凸塊 圖像係顯示理想之凸塊本身者。亦即，第i實施形態之主 圖像係表示與第9圖所示之檢查區域相同之區域者。& 此’在第2實施形態中’採用與第1實施形態相同之主圖 像時，會有無法正確地進行凸塊之檢查之虞。 圖詳細說明之。 弟 弟10圖係用於說明無法順利攝像標的圖像時的假相 的比較檢查的圖。自CCD攝像機2攝像基板S時，；於 形成在基板S之銅圖案#反射，因而有無法順利攝像凸塊 317686 21 1274869 之圖案的情形。& 士 Φ 此％，具體而言標的圖像 為“1，，（意指形成有凸塊，參考第：：貝广係成為全面 乐1。圖（3)之圖俊 種榣的圖像係在比較檢查中應判斷為缺陷。 ^而’關於該標的圖像’利用前述第}檢查區域圖像 二二：，或圖像進行比較檢查時，無法檢測出缺陷。 Γ二為“1，，之標的圖像進行前述步㈣之處理 日守*才欢測出正確形狀之凸塊圖像，故在第2檢查區域 :像中全面成為檢查區域（參照第10圖⑴之圖像)。另 一=面’第1檢查區域圖像係將形成凸塊之理想區域設為 檢一區域（翏照第10 K (e)之圖像），因此作為比較檢查 處理之對象的區域係成為第J檢查區域圖像全部的檢查區 域。在此’標的圖像係全面為“Γ，因此關於第i檢查區 域圖像之檢查區域’標的圖像之各像素的值係全部為 1 。再且，主圖像關於第1檢查區域圖像之檢查區域而 言，標的圖像之各像素之值全部為“Γ (參照第10圖⑷ 之圖像）。因此，關於第丨檢查區域圖像之檢查區域，標的 圖像與主圖像之間並無不同點（參照第10圖^㈠之圖像）， 結果，各凸塊被判斷為正常。 因此，在第2實施形態中，係採用使顯示理想凸塊之 凸塊圖像之一部分缺陷的圖像作為主圖像。第u圖係第2 貫施形悲中所使用之主圖像的示意圖。如第丨i圖所示，在 第2實施形怨之主圖像中，顯示理想凸塊之凸塊圖像的中 心部分有缺陷。再者，缺陷部分之形狀及大小亦可為任意 者。然而，有必要設定缺陷部分，以便將表示理想凸塊之 22 317686 1274869 柃杳：该凸塊圖像之-部分缺陷的主圖像加以比較 ::鱼 陷。亦即’在比較檢查顯示理想凸塊之凸塊 :差：主圖像時’為了使比較檢查之結果為有缺陷之程度 的至異，必須設定缺陷部分。 如上所述，II由在主圖像設定缺陷部分，如果對第⑺ 圖U)所示之標的圖像進行比較檢查時，關於前述中心部 分，在標的圖像與主圖像會產生差異。結果，在比較檢^Further, in the towel of the present embodiment, the result of the comparison inspection is a display device (not shown) as an image. For example, the comparison checking unit 36 may display the image of the & conjugate and the image of the main image on the display device. At this time, it is possible to distinguish between the bumps of the normal bumps and the bumps of the defects by using the bumps judged to be normal and the bumps judged to be defective. Further, in the image of the variable target image and the image of the main image, only the inspection area of the image of the second inspection area is displayed. At this time, the 'normal bumps are present in the non-inspection area: therefore, not shown, only the bumps of the defects are displayed. Therefore, it is also possible to distinguish the bumps of the bumps and the bumps of the defects. The method of comparing the results of the inspection is as long as it is a method for the operator to prompt the normal or defective of each bump, but may be a method of any, but the operator finally judges whether or not the defect is maintained, preferably as above. The method of displaying as an image. As described above, according to the present embodiment, the bump inspection apparatus first performs processing for recognizing the shape of the bump (step s2 described above). By the processing, the bumps that are determined to be normal (for example, the i-th and the ghosts) are detected in the shape, and the u-bump inspection device performs a comparison check between the target image and the main image (step S5 described above). On the other hand, the bump formed in a positive phase and the bump formed to the extent that the allowable position is deviated are detected. Thus, by the shape-related judgment and the position-related judgment of the individual (four) blocks, the critical value of the judgment threshold value va (4) and the position-related judgment of the shape of the bump can be selected. Specifically, by changing the value of the predetermined range of display ==2, the position of the bump can be made two: the position S' is changed by the value of the predetermined length of the step S4, so that the position of the convex ghost is made. Related tolerance changes. According to the present embodiment, the critical value related to the shape and the position can be appropriately set by 317686 20 Ϊ 274869, so that it is possible to accurately judge whether or not the position and shape of the bump are defective. (Second Embodiment) Hereinafter, a bump inspection device according to a second embodiment of the present invention will be described. In the second embodiment, in addition to the above! The contents of the inspection area image and the main image are the same as those of the first embodiment except for the first embodiment. That is, the basic configuration (Fig. j) and the operation (Fig. 3) of the bump inspection device are the same as those of the first embodiment. In the following description, the position of the substrate and the shape of the substrate (Fig. 2) and the obtained figure (Fig. 4) are the same as those of the first embodiment. Fig. 9 is a view showing an image of a second inspection region used in the second embodiment. As shown in Fig. 9, in the second embodiment, only the region where the bump is to be formed (the ideal region in which the bump is formed) is referred to as the inspection region. Since &, the comparison check processing can be suppressed to a minimum, and the processing load of the image processing apparatus 3 can be reduced, and processing can be performed in a short time. Here, in the first embodiment, the bump image included in the main image indicates the ideal bump itself. That is, the main image of the i-th embodiment indicates the same area as the inspection area shown in Fig. 9. < In the second embodiment, when the main image is the same as that of the first embodiment, the inspection of the bumps cannot be performed correctly. The figure illustrates in detail. The brother 10 is a diagram for explaining a comparison check of a false phase when an image of a target image cannot be smoothly captured. When the CCD camera 2 images the substrate S, it is reflected by the copper pattern # formed on the substrate S, and thus there is a case where the pattern of the bumps 317686 21 1274869 cannot be smoothly imaged. & 士 此 This %, specifically the target image is "1,, (meaning that the formation of bumps, reference to:: Beguang became a comprehensive music 1. Figure (3) map of the image of Jun It should be judged as a defect in the comparison check. ^ And 'About the target image' can not detect the defect by using the above-mentioned first inspection area image 22: or when the image is compared and checked. In the target image (4), the image of the above-mentioned step (4) is used to measure the bump image of the correct shape. Therefore, in the second inspection area: the image is comprehensively the inspection area (see the image of Fig. 10 (1)). In the image area of the first inspection area, the ideal area for forming the bump is the detection area (the image of the 10th K (e)), and therefore the area to be subjected to the comparison inspection process becomes the first J checks all the inspection areas of the area image. Here, the 'target image system is all "Γ, so the value of each pixel of the image of the inspection area of the i-th inspection area image is all 1". The main image is about the inspection area of the image of the first inspection area, and each of the target images The value of the prime is "Γ (refer to the image in Fig. 10 (4)). Therefore, there is no difference between the target image and the main image in the inspection area of the image of the second inspection area (refer to Fig. 10). (I) Image) As a result, each of the bumps is judged to be normal. Therefore, in the second embodiment, an image in which one of the bump images of the ideal bump is partially defective is used as the main image. The diagram is a schematic diagram of the main image used in the second embodiment of the sorrow. As shown in Fig. i, in the main image of the second embodiment, the central portion of the bump image of the ideal bump is displayed. Further, the shape and size of the defective portion may be any. However, it is necessary to set the defective portion so as to represent the ideal bump 22 317686 1274869 柃杳: the main part of the bump image The images are compared:: fish traps. That is, 'the bumps of the ideal bumps are displayed in the comparison check: difference: main image'. In order to make the result of the comparison check as defective, the defective portion must be set. As described above, II is set by the defective part in the main image, such as When the comparison image of the target image shown in Fig. 7 (U) is compared, there is a difference between the target image and the main image with respect to the aforementioned central portion. Result, in comparison check ^
，理時，判斷各凸塊為缺陷。因&，在第2實施形態中， 藉由在主圖像設定缺陷部分，即使在因反射等原因而無法 正確地攝像標的圖像之情形，亦可正確地判斷缺陷。 再者，在前述第2實施形態中，關於正確形成之凸塊， 由於顯示該凸塊之凸塊圖像及其周邊區域係在第2檢查區 域圖像中成為非檢查區域，因此並不進行比較檢查。結果， 该凸塊即被判斷為正常，因此即使在主圖像設定缺陷部分, 也不會造成問題。 ’ 以上洋細說明本發明，但前述之說明在各方面皆只是 本發明之例示，並非欲限定本發明之範圍者。在不脫離本 發明之範圍的情況下，當然可進行各種之改良及變形。 【圖式簡單說明】 第1圖係第1實施形態之凸塊檢查裝置的構成方塊 圖0 第2圖係正確形成之基板的一例圖。 第3圖係凸塊檢查裝置之圖像處理裝置之處理流程的 流程圖。 317686 23 1274869 第4圖係由步驟S1所獲得之標的圖像的示意圖。 弟5圖係顯示弟2檢查區域圖像的示意圖。 第6圖係本實施形態中所使用之主圖像的示意圖。 第7圖係本實施形態中所使用之第1檢查區域圖像的 示意圖。 第8圖係用以說明進行比較檢查之區域的示意圖。 第9圖係第2實施形態中所使用之第1檢查區域圖像 的示意圖。 •第10圖（a)至（e)係無法順利攝像標的圖像時的假想比 ^ 較檢查的說明圖。 第11圖係第2實施形態中所使用之主圖像的示意圖。 第12A圖至第12C圖係同時比較複數凸塊時之圖像例 的示 意圖。 【主 要元件符號說明】 1 台座 2 CCD攝像機 3 圖像處理裝置 31 2值化處理部 32 標的圖像記憶部 33 形狀檢測部 34 凸塊位置記憶部 35 檢查區域設定部 36 比較檢查部 37 主圖像記憶部 24 317686 1274869 38 檢查區域圖像記憶部 41至44第1至第4凸塊 凸塊圖像 46、47、48、49、51、52、53、54 91、 94 第1凸塊之圖像 92、 95 第2凸塊之圖像 93、 96 第3凸塊之圖像 S 基板When judging, it is judged that each bump is a defect. In the second embodiment, by setting the defective portion in the main image, even if the target image cannot be accurately captured due to reflection or the like, the defect can be accurately determined. Further, in the second embodiment, the bump image formed correctly is not subjected to the non-inspection region because the bump image indicating the bump and the peripheral region thereof are displayed in the second inspection region image. Comparative check. As a result, the bump is judged to be normal, so that even if the defective portion is set in the main image, no problem is caused. The invention is described above in detail, but is not intended to limit the scope of the invention. Various modifications and variations can of course be made without departing from the scope of the invention. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a block diagram of a bump inspection apparatus according to a first embodiment. Fig. 0 is a view showing an example of a substrate which is correctly formed. Fig. 3 is a flow chart showing the processing flow of the image processing apparatus of the bump inspection apparatus. 317686 23 1274869 Figure 4 is a schematic illustration of the subject image obtained in step S1. The brother 5 shows a schematic diagram showing the image of the brother 2 inspection area. Fig. 6 is a schematic view showing a main image used in the present embodiment. Fig. 7 is a schematic view showing an image of a first inspection region used in the present embodiment. Fig. 8 is a schematic view for explaining an area where comparative inspection is performed. Fig. 9 is a schematic view showing an image of a first inspection region used in the second embodiment. • Figure 10 (a) to (e) are illustrations of the imaginary ratio of the image that cannot be successfully captured. Fig. 11 is a schematic view showing a main image used in the second embodiment. Fig. 12A to Fig. 12C are diagrams showing an example of an image when a plurality of bumps are simultaneously compared. [Description of main component symbols] 1 pedestal 2 CCD camera 3 image processing device 31 binarization processing unit 32 target image storage unit 33 shape detection unit 34 bump position storage unit 35 inspection area setting unit 36 comparison inspection unit 37 main picture Image memory unit 24 317686 1274869 38 Check area image memory units 41 to 44 1st to 4th bump image 46, 47, 48, 49, 51, 52, 53, 54 91, 94 1st bump Image 92, 95 image of second bump 93, 96 image of third bump S substrate
25 31768625 317686