4 4200408269 玖、發明說明： 【發明所屬之技術領域】 發明領域 本發明係有關於一種圖形缺陷檢查裝置及圖形缺陷檢 5查方法，特別係有關於一種可自動地檢查塗布於電漿顯示 器等之玻璃基板之螢光體之塗布缺陷之圖形缺陷檢查裝置 及圖形缺陷檢查方法。 先前技術 · 10 眾所週知，目前已有用以檢查塗布或印刷於電漿顯示 裔等之玻璃基板之螢光體之塗布或印刷缺陷之圖形缺陷檢 查裝置，而該裝置係例如藉紫外線照明光源照射紫外線於 螢光體形成條紋狀之電漿顯示器等之玻璃基板，使已形成 之螢光體發光者。該發光影像藉如線感測器（一維感測器） 15 之攝影部攝影。形成之螢光體為紅色(R)、綠色(G)、藍色(B) 之螢光體，因此由攝影部攝影時，於攝影部安裝分別對應 R、G、B之螢光體之濾色器，用以攝影分別對應之螢光體 ® 之影像。該裝置係藉輸出攝影部攝影之影像至影像處理 部，例如藉顯示於顯示裝置，檢查塗布於玻璃基板之螢光 20 體之圖形缺陷。 、 第15圖係藉前述之圖形缺陷檢測裝置攝影塗布於電漿 ， 顯示器之玻璃基板之條紋狀之螢光體之攝影畫面。於第15 圖中，91表示玻璃基板之一部份，且顯示出於玻璃基板91 上R、G、B各色之螢光體塗布成條紋狀之狀態。92係表示 5 紅色（R)螢光體上之如針孔之缺陷部。又，於玻璃基板91之 周邊部顯示之波形，係表示由攝影部得到之影像之亮度信 號值。93、94係表示亮度信號之0值。95係表示以單點虛線 A表示之紅色（R)螢光體之亮度信號值。缺陷部92之部分 中，例如因針孔缺陷，紅色(R)螢光體沒有發光，亮度信號 值降低。96係顯示以早點虛線B表示之部分，即，沒有塗布 螢光體之間隙部分之亮度信號值。98係表示以單點虛線C 表示之部分之党度彳έ號值’可知道缺陷部92之部分亮度信 號值降低。 97、99係表示亮度信號值之閾值，即缺陷判定值，一 般係設定為最大亮度信號值之5〇%左右。然而，亦可由與 缺陷檢測之精度之關係適當調節，或者以實驗來制定。 如前述’雖可由亮度信號值測出螢光體之塗布缺陷， 但由第15圖之亮度信號值可清楚的知道，不僅螢光體塗布 缺陷部92之亮度信號值較閾值還低，沒有塗布螢光體之間 隙部分之亮度信號值亦較閾值還低，因此不能自動地判定 係螢光體塗布缺陷部，或_部分。此外，乍看之下若考 慮位置資料時，可狀輕域塗布缺陷㈣，或間隙部 分，然而實際上塗布於電_示_之玻璃基板之螢光體 之寬度係200’25—，而間隙之寬度為約綱”，係 極細微之榮光面，因此無法由位置資料測出。 又，檢查細微之如電極之圖形之缺陷時，已知有使用 相鄰比較檢查法（參照日本專利公開公報第麵_55817號之 弟2〜3頁，第1圖）。此方法係用以檢測諸如像電衆顯示器之 200408269 電極之細微之圖形之缺陷者，該方法係將多數電極分組， 並比較其組中之1個電極與其他組中之1電極，|反覆以檢 查全部電極之缺陷。然而，依據該相鄰比較檢查法時，為 了比較組間之電極，必需高精度地進行對位，然而如前述， 5電漿顯示器之螢光體極細微，因此若要對位，則亦必需考 慮圖形之形狀等，且為了要正確地進行對位，需要有極高 精度之對位裝置’不易實現低成本之缺陷檢查裝置。 t 明内溶1 3 發明概要 本么明之目的係提供可自動地檢測塗布於電衆顯示器 等之基板之螢光體之塗布缺陷之圖形缺陷檢查裝置及圖形 缺陷檢查方法。 一时^ 的係提供一種可輕易地設置於 15 204 4200408269 Description of the invention: [Technical field to which the invention belongs] FIELD OF THE INVENTION The present invention relates to a pattern defect inspection device and a method for inspecting pattern defects, and particularly relates to a method for automatically inspecting a coating applied to a plasma display, etc. A pattern defect inspection device and a pattern defect inspection method for coating defects of a phosphor on a glass substrate. Prior art · 10 As we all know, there are currently pattern defect inspection devices for inspecting coating or printing defects of phosphors coated or printed on glass substrates such as plasma displays, and the device is, for example, irradiating ultraviolet rays with ultraviolet light sources The phosphor forms a glass substrate such as a plasma display with a stripe shape, so that the formed phosphor emits light. The light-emitting image is photographed by the photographing section of the line sensor (one-dimensional sensor) 15. The formed phosphors are red (R), green (G), and blue (B) phosphors. Therefore, when photographing by the photography department, install filters corresponding to R, G, and B phosphors in the photography department. Color device for photographing images of the corresponding phosphor®. The device outputs the image captured by the photographing section to the image processing section, for example, by displaying on the display device, inspecting the pattern defects of the fluorescent 20 body coated on the glass substrate. Fig. 15 is a photograph of a stripe-shaped phosphor coated on a plasma substrate of a display using a graphic defect detection device as described above. In FIG. 15, 91 indicates a part of the glass substrate, and shows a state where the phosphors of each color of R, G, and B on the glass substrate 91 are coated in a stripe shape. 92 indicates 5 pinhole defects on red (R) phosphors. The waveform displayed on the periphery of the glass substrate 91 indicates the luminance signal value of the image obtained by the imaging section. 93 and 94 are zero values of the luminance signal. 95 indicates the luminance signal value of the red (R) phosphor indicated by a single dotted line A. In the portion of the defective portion 92, for example, due to a pinhole defect, the red (R) phosphor does not emit light, and the luminance signal value decreases. The reference numeral 96 indicates a luminance signal value of a portion indicated by an early dotted line B, that is, a gap portion where no phosphor is applied. 98 indicates that the value of the squareness of the portion indicated by the single-dotted dotted line C 'indicates that the value of the luminance signal of the defective portion 92 has decreased. 97 and 99 indicate the threshold value of the brightness signal, that is, the defect judgment value, and are generally set to about 50% of the maximum brightness signal value. However, it can also be adjusted by the relationship with the accuracy of the defect detection, or it can be established through experiments. As mentioned above, although the coating defect of the phosphor can be measured from the brightness signal value, it can be clearly known from the brightness signal value of FIG. 15 that not only the brightness signal value of the phosphor coating defect portion 92 is lower than the threshold value, but also there is no coating. The brightness signal value of the gap portion of the phosphor is also lower than the threshold value, so it cannot be automatically determined as the phosphor coating defect portion or the _ portion. In addition, at first glance, if the position data is considered, the defect area or gap can be coated in a light area, but the width of the phosphor coated on the glass substrate is 200'25— "The width is approximate," which is a very delicate glare surface, so it cannot be measured from the position data. In addition, it is known to use an adjacent comparison inspection method when inspecting small defects such as electrode patterns (refer to Japanese Patent Publication) Brother _55817 (Page 2 ~ 3, Figure 1). This method is used to detect defects such as subtle patterns of electrodes like the 200408269 electrode of the electric display. This method is to group most electrodes and compare them. One electrode in the group and one electrode in the other group are repeatedly checked for defects in all the electrodes. However, according to the adjacent comparison inspection method, in order to compare the electrodes between the groups, it is necessary to perform high-precision alignment. However, As mentioned above, the phosphor of the 5 plasma display is very fine, so if you want to align, you must also consider the shape of the graphics, etc., and in order to perform the alignment correctly, you need an extremely high-precision alignment device. Achieving a low-cost defect inspection device. TMei Nyoi 1 3 Summary of the invention The purpose of this Meming is to provide a pattern defect inspection device and pattern defect inspection that can automatically detect coating defects of phosphors coated on substrates such as electronic displays. Method. The Temporary system provides an easy setting at 15 20
等之為面板之製造線，且實;見高速之缺陷檢查、並 且便宜之圖形缺陷檢錢置及_缺陷檢查方法。 部、公x月之圖①缺1^檢查裝置包含有攝#彡部、移動機構 ’“象處理部、顯示部、及控制部。該攝影部係用以攝 部二:::上之螢光體之條紋狀圖形者，前述移動機構 係==影部沿前述圖形移動者，該影像處理部 J入；自别述攝影部之影像 顯示前述影像處理部之輸 ^者’料顯示部係用以 述移動機構部與前 ，而該控制部係用以控制前 包括影像處理部、差:者’ a ’前述影像處理部 影像輪入部係用部、及缺陷檢測部，前述 71 乂檢測前述螢夹 变九體之條紋狀圖形之方向， 408269 4差兴影像檢測部係用以比較與前述圖 性之至少2處影像資料者，而該缺陷檢測部向=有相關 結果檢測前述圖形之缺陷。 “艮據所述比較 5 10 15 又，本發明之圖形缺陷檢查裝置 機構部、影像處理部、顯示部、及控^有攝影部、移動 用輯影形成於基板上之格子狀螢光體;布=攝影部係 。亥移動機構部係用以使前述攝影部沿前述圖开，、圖形者， ，，像處理部係輸入來自前述攝影部之影像信號夕:者丄，前 不部係心顯*前述影 ^，垓顯 用以控制前述軸腦μ , 而該控制部係 私動機構部與前述影像處理部 _ 像處理部包括影像輸人部 又’則逑影 部，前述影像於人撼^ 像卜及缺陷檢測 之圖形之二=;!以算旦出前述格子狀螢光體塗布膜 格子間距之整數倍之大小之至少2領域之影像較= 上之ϋ光二發B:之圖形缺陷檢查方法包括攝影形成於基板 ^體之條紋狀圖形之步驟、藉由前述攝影得到之影 像貢料檢測前述螢光體之條紋狀圖形之方向之于 述影像資料比較已依據前述 ”别 少2處影像資料之步驟、及二比=位置關係之至 之缺陷之步驟。及根據刚速比車父結果檢測前述圖形 二本發明之圖形缺陷檢查方法包括攝影形成於基板 上之格子狀榮光體塗布膜之圖形之步驟、藉由前述攝影得 20 到之影像資料算出前述格子狀勞光體塗布膜之圖形之格子 間距之步驟、由前述影像資料比較至少2領域之具有前述格 子間距之整數倍大奴影像資料之步驟、及根據前述比較 〜果檢測别述圖形之缺陷之步驟。 C 方包方式j 較佳實施例之詳細說明 第1圖係顯示本發明之圖形缺陷檢查裝置之一實施例 之圖。於第1圖中，1係·顯示器等之玻璃基板之載置台， 2係電漿顯示器等之玻璃基板，3係R、G、6之螢光體（螢光 體塗布膜）’ 4係用以使螢光體3發光之紫外線照明用光源， 5係安裝有透鏡及R、G、B之渡色器之光學系統，6係攝影 用線感測照相機等之攝影部，7係用以使攝影部6及光源4沿 破螭基板2移動以於玻璃基板2上進行掃描之移動機構部，$ 係用以檢測針孔等缺陷之影像處理部，9係顯示或列印檢杳 結果之彩色螢幕、印表機等顯示部，1〇係用以驅動移:機 構部7之驅動部，n係用以控制影像處理部8及驅動部的 控制部，而15係操作部，且係用以進行本檢查裝置之择作 之部分。又，如後述，影像處理部8係由影像輸入部12、差 異影像檢測部13及缺陷檢測部14構成。此外，光源4只要曰 使螢光體塗布膜發光之光源即可，並不限為紫外線發疋 源，亦可為其他電磁波之珈瑪射線或X線等粒子線。 第2圖係顯示第1圖所示之圖形缺陷檢查裝置之栽置a 1、玻璃基板2及攝影部6之放大圖，且對於與第1圖相同= 附上同一標號。載置台1於檢查時載置玻璃基板2， V、電漿 頭示面板之玻璃基板塗布諸如紅色(R)之螢光體時，為了广 查其塗布狀態，朝箭號指示之方向搬送塗布有紅色(R)螢光 體塗布膜之玻璃基板，並固定於第2圖所示之預定位置，以 檢查有沒有缺陷。塗布綠色（G)之螢光體塗布膜之線，及塗 布藍色(B)之螢光體之線亦進行相同之檢查。此外，本實施 例中’玻璃基板之大小係146〇mmx 1030mm，但並不限定 於此。 21係移動機構7之一部份，且為用以支持攝影部6及紫 外線知、明用光源4之支持構件。攝影部6為了檢查一片玻璃 基板，如圖示般構成為成列地配置4台線感測照相機，以遮 蔽寬度1030mm之玻璃基板。χ台線感測照相機之攝影寬度 約260mm，且線感測照相機間之視野範圍構成為一部份重 ®。藉來自紫外線照明用光源4之紫外線22激發螢光體3， 例如，將激發出之紅色化)之螢光體3之像透過光學系統5由 攝影部6攝影。該支持構件21朝紅色(R)之螢光體3之γ方向 等速地由右端向左端移動，掃描玻璃基板整面。 以下，針對該動作詳細地說明。藉紫外線照明光源4 照射紫外線22於電漿顯示器等之玻璃基板2，使已塗布（亦 包含藉印刷之塗布）之螢光體3發光。藉攝影部6攝影其發光 衫像。此時，可依據應檢測之螢光體之種類(R、G、B)於攝 衫部6安裝對應各個顏色之螢光體之濾色器。藉攝影部6攝 影之影像送出至影像處理部8。 第3圖係說明本發明之圖形缺陷檢查裝置之檢測針孔 專之圖形缺陷之原理之圖。第3圖係顯示於玻璃基板2週期 200408269 性地塗布有條紋狀之R、G、B各色之螢光體3之情況。以下 之說明係針對塗布有R、G、B各色之螢光體3之玻璃基板作 說明，然而實際之製造線當然如前述，係於每次依序塗布 各色之螢光體時進行檢查，並於測出缺陷時，停止其次之 5螢光體塗布或印刷步驟，而將有缺陷之玻璃基板洗淨，並 再次重新塗布或印刷螢光體，係具有不無謂地塗布或印刷 之優點。又，由於必需配合製造線之工作時序進行檢查， 因此必需為檢查速度快速之圖形缺陷檢查裝置。 攝影部6攝影之影像資料送至影像處理部8，再輸入影 1〇像輸入部12，且為了求出差異影像係記憶於記憶部（圖未 示）。影像輪入部12係測出螢光體之條紋狀圖形之方向，並 將記憶之影像資料分割成多數區塊。例如，切成一般週知 之4像素X 4像素之區塊（以下，表示為4χ 4區塊等）、8χ 8 區塊、或32χ 32區塊31及32，並輸出至差異影像檢測部13。 15區塊31及32係用以測出缺陷之最適當之單位區塊，該大小 係由檢查速度、處理速度及缺陷檢測精度等實驗地適當設 定。 差異影像檢測部13中，藉比較區塊31與區塊32，至少 比較2處之影像資料。比較方法係例如藉比較區塊^與區塊 2〇 32为別之像素之焭度信號值，進行區塊31與區塊32之差異 影像檢測33。發光體3上有針孔等缺陷34(第3圖係顯钟之 瑩光體具有缺陷34)時，於差異影像35測出亮度信號值之差 之缺陷36。差異影像檢測部13之輸出於缺陷檢測部14比較 差異影像與事先設定之判定值（閾值），超過判定值時，測出 11 200408269 有缺陷。該差異影像35直接顯示於顯示部9，或者得到2值 化之影像37及2值化之缺陷38之信號，因此可自動地測出缺 陷。 因此，使該等區塊31與區塊32依序移動，以針對玻璃 5 基板整體進行差異影像之檢測，藉此可檢查條紋狀之榮光 體全部之缺陷。此外，藉將該等檢查資料記憶於記憶部（圖 未示），並分析檢查資料，亦可有助於製造上之品質管理。 又’前述比較方法係說明藉亮度信號值之比較，但並不限 疋於此’藉影像信號之亮度統計圖之比較等當然亦可檢測 1〇差異影像。 且，本發明之圖形缺陷檢查裝置中，區塊31與區塊32 之位置關係必需位於螢光體之長向（於第3圖中係上下之位 置關係）’又，於使區塊31與區塊32移動時，亦必需保持該 關係。有關於該理由，利用第4圖作說明。 15 第4圖中，於玻璃基板2上橫向地塗布螢光體3，且區塊 31與區塊32係具有位於垂直於螢光體之條紋方向之方向上 之位置關係，即如圖所示之上下之關係。此外，第4圖之各 邛與第3圖同樣者，附上同一標號。41係表示差異影像，42 則係顯示2值化影像。於第4圖中，實行區塊31與區塊32之 2〇差異影像檢測33時，區塊31之各色之螢光體3之位置與區塊 32之各色之螢光體3之位置沒有正破地一致時，即各色之螢 光體3之位置有偏差時，於差異影像41中，除了缺陷弘以 外還顯現有區塊31與區塊32之各色之勞光體3之差分之信 號之條紋43，並誤測有條紋狀之缺陷。因此，輸出至2值化 12 WU408269 衫像42中，亦有條紋狀之2值化差分值44，為誤測。 因此’如第4圖，印刷圖形之條紋方向與區塊31及區塊 32之排列方向不同時，必需使兩區塊領域内之r、〇、b之 條紋位置正確地一致，用以達成前述需要之處理，係必需 心像相關次异等魔大之計算處理，因此相當不易實現可設 置於衣k線之圖形缺陷檢查裝置。然而，如第3圖所示，印 刷圖形之條紋方向與區塊31及區塊32之排列方向相同時， 如利用第2圖所作之說明，攝影部6之移動方向與條紋之長 向致，因此區塊31與區塊32之對位並不需要進行條紋組 之對位，只需使區塊與區塊32—致就可對位，因此可極 簡易地對位。 換5之’令攝影部6之排列方向為X軸，且攝影部6之移 動方向為Y軸時，區塊31與區塊32於X軸方向上之位置，係 以移動機構7經常保持為固定，不必進行定位，而只要使區 15塊31與區塊32於丫方向上之位置一致，就可簡易地使區塊31 與區塊32内之各色之條紋之位置一致，因此可簡易地僅檢 測出缺陷34。 接著，採用如第3圖所示之圖形缺陷檢查方法時，必需 測出玻璃基板上之條紋之方向。針對此，以第5圖及第6圖 20作說明。第5圖係顯示於玻璃基板2上縱向地塗布有r、g、 B之條紋狀螢光體之狀態，以第i圖所示之圖形缺陷檢查裝 置攝影其，並藉處理攝影之影像之亮度信號測出條紋之方 向。即，51係顯不橫向像素之亮度信號值(加算投影波形）， 52則係顯示縱向像素之亮度信號值(加算投影波形）。53、54 13 200408269 係表示亮度信號之0值，55、56則係表示用以檢測螢光體之 判定值（閾值）。攝影之影像之亮度信號，係於影像輸入部12 處理，並利用測出週期地超過判定值之方向，界定圖形方 向。即，測出週期地超過判定值之亮度信號值52之方向係 5 條紋狀螢光體之長向。 第6圖係顯示於玻璃基板2上橫向地塗布有r、g、Β之 條紋狀螢光體之狀態。因此，與第5圖同樣地，測出週期地 超過判定值之亮度信號值5 2之方向係條紋狀螢光體之長 向。此外，第6圖之各部之標號係對應第5圖之各部之標號。 10 針對本發明之另一實施例，以第7圖作說明。第7圖所 示之實施例係顯示於玻璃基板2上橫向地塗布有條紋狀之 螢光體3之情況。此外，第7圖之各部之標號與第3圖相同者 係附上同一標號。於第7圖中，用以檢測差異影像之區塊領 域71及72之位置關係’係各色之條紋之長向，即，於第7圖 15中，區塊71、72係配置成橫向之位置關係。形成如此之位 置關係時，區塊領域71内之各色之條紋之位置關係，與區 塊領域72内之各色之條紋之位置關係相同，因此實行區塊 領域71、72之差異影像檢測33時，於差異影像塊得到缺 陷36之信號。因此，於2值化影像37得到2值化之缺陷觀2 20 值化信號。 接著，針對本發明之圖形缺陷檢查裝置之動作之一 例’使用第8圖作說明。首先’第1步驟繼中，將塗布有條 紋狀營紐(例如⑻螢纽)之玻璃騎冰人並固定於載 置台！時，攝影部6藉移動機構7由 14 200408269 第2步驟l〇2，係進行檢查領域之檢測。此係攝影部6 使最先藉紫外線光激發之螢光體3，例如紅色(R)螢光體受 光之時點，全部之處理步驟將以此為基準而開始。 第3步驟1〇3中，首先，進行已印刷之條紋狀榮光體之 5方向之判定。即，如以第3圖及第4圖所作之說明，本發明 中判定條紋狀螢光體之方向係對針孔等缺陷檢查極為重 要。該方向性判定，係攝影部6使藉紫外線光激發之螢光體 3，例如紅色（R)螢光體受光後直接測出之諸如兇像素X 像素之影像信號，並以第5圖及第6圖說明之方法判定螢光 1〇體之條紋方向。此外，32像素可涵蓋多條螢光體之條紋， 因此係可充分地判定方向性之像素數。 第4步驟1〇4係決定比較區塊之方向。即，依照如前述 以第3步驟判定出之條紋狀螢光體之方向，決定用以比較之 2個區塊之位置關係。例如，當條紋狀螢光體之方向如第3 15圖般為縱向時，選擇使2個區塊之位置關係為上下方向之關 係，而條紋狀螢光體之方向如第7圖般為橫向時，選擇使2 個區塊之位置關係為橫向之關係。 以上說明之各實施例中，2個區塊之位置關係為沒有間 隙且互相密接之位置關係，然而亦可依照處理方法，使一 20部份區塊重疊，或使區塊之間具有空隙而攝取影像。 第5步驟105係維持第4步驟所決定之2個區塊之位置關 係，並含括檢查對象之電漿顯示面板等之玻璃基板全體， 求出2個區塊之差值影像。 第6步驟106係比較由第5步驟所求出之差值影像得到 200408269 之党度h號值’與缺陷判定值（閾值），若有較缺陷判定值還 鬲之信號差值時，判定其為圖形缺陷。此外，缺陷判定值（閾 值)係設定為由影像信號得到之最高值之約5〇%左右，然而 可由貫驗’或於檢查之過程中依據需要地適當調節，並可 5 改變設定。 如河述之步驟係依序對已印刷或塗布之條紋狀螢光體 之諸如R、G、B之各色螢光體進行。當然，如前述於R螢光 體檢查中測出有缺陷時，中止下一顏色之螢光體之塗布或 印刷之步驟，且該玻璃基板進入除去螢光體之步驟，並再 10 生。 以上’係針對本發明詳述，本發明之圖形缺陷檢查方 法係採用以2個區塊之差異影像檢測缺陷之方法，因此有不 月匕由差異衫像;^測結果判定2個區塊中哪一個有缺陷之缺 點。針對用以解決該缺點之方法，使用第9圖作說明。 15 帛9圖係顯示用以由多數差異影像判定哪個區塊領域 有缺之方法。第9圖係顯示以、G、B之榮光體3之一部份， 且針對於6之錢狀螢讀上有針孔缺陷社情況作說 明。此外，省略玻璃基板。首先，區塊幻係於領域卜且區 塊83於領域2日守，2個區塊82與Μ之差異影像84顯現出缺陷 2〇 87。於"亥^又中’然法判定缺陷81係位於區塊82或區塊83 之區塊領域。接下來使區塊幻触移動i區塊。即，使區塊 82和動至錢2，且使區塊幻移動至領域3，並求出區塊 與幻之差異影像85，而測出有缺陷88時，可判定領域2具有 缺fe81且，使區塊82與區塊幻再移動1區塊。即，於使區 16 200408269 塊82移動至領域3，且使區塊83移動至領域4，得到區塊幻 與區塊83之差異影像86時，若差異影像秘沒有缺陷，可得 知領域3、4沒有缺陷。因此，依據該方法，可得知領域2具 有缺陷’而領域1、3、4沒有缺陷。 5 #著，針對本發明之另-實施例進行說明。前述之實 施例中，於如第3圖所示般在玻璃基板2上週期地塗布有r、 G、B各色之螢光體3成條紋狀之情況下，可極高精度地測 出螢光體之塗布缺陷，然而塗布於玻璃基板2上之螢光體， 不是如第3圖所示之均_之條紋構造之螢光體塗布膜時，會 1〇有热法適用該方法之問題。 第10圖係可說明用以說明該問題之圖形缺陷檢查裝置 之針孔等圖形缺陷檢測之原理之圖。於第10圖中，對於與 第3圖同-者附上同_標號。6〇係塗布於玻璃基板2上之勞 光體塗布膜，然而係構成為格子狀。即，朝橫向週期地反 覆k布有紅色(R)、綠色(G)、藍色⑼之勞光體。縱向係由 1隙區卩岡且刀離成島狀。以下將如前述之螢光體塗布 膜稱作格子狀螢光體塗布膜。 乂下針對以使用差異影像之螢光體之塗布缺陷之檢 查方法，檢查如前述之構造之格子狀之榮光體塗布膜之針 20孔等圖形缺陷之情況作說明。與第3圖同樣地，於差異影像 檢測33比較區塊31與區塊32之像素之亮度信號值。針孔等 之缺陷34於格子狀螢光體塗布膜60(於第1〇圖中，顯示出R 之格子狀螢光體塗布膜6〇具有缺陷34之情況)上時，於差異 影像35測出亮度信號值之差之缺陷36。 17 於差，、々像35測出成為差異影像62之間隙6i。即， 比較區塊31與區塊32時之像素時可清楚地明白，間隙部6! 於區塊31與區塊32中位於不同位置，因此差異影像檢卿 輪出之差異影像35中顯現出缺陷％與間隙Μ之差異影像， 5因而無法區別缺陷36與間隙61之差異影像62。因此，2值化 之影像37亦得到2值化之缺陷％與間隙之〕值化影⑽之信 號，故，可自動地檢測缺陷38。 第11圖係用以說明本發明之圖形缺陷檢查裝置檢測針 孔等圖形缺陷之原理。於第u圖中，對於與第關同一者 10附上同-標號。於第u圖中，於玻璃基板2上塗布勞光體塗 布膜60，並構成為格子狀。即，朝横向週期地塗布有紅色 ⑻、綠色(G)、藍色⑼之螢光體。於縱向上，係由間隙61 分別區隔紅色⑻、綠色(G)、藍色(B)之螢光體且分離成 島狀。此外，以下之說明係針對塗布有R、G、Bg色之螢 15光體塗布膜60之玻璃基板作說明，然而如同前述，實際之 製造線於每次依序塗布各色螢光體塗布膜6〇時進行檢查。 且，攝影部6攝影之影像資料，係送至影像處理部8， 再輸入影像輸入部12。於影像輸入部12中，算出袼子狀螢 光體塗布膜之圖形之袼子間距，並將影像分割成多數區 20塊。接著，切成區塊231及區塊232，並輸出至差異影像檢 測部13。有關區塊231及232之大小與格子狀螢光體塗布膜 60之關係後述。 於差異影像檢測部13，比較區塊231與區塊232。比較 方法，係例如藉比較區塊231與區塊232之分別之像素之亮 18 200408269 度信號值’進行區塊231與區塊232之差異影像檢測33。針 孔等之缺陷34於螢光體塗布膜60(於第11圖中，顯示出R之 發光體塗布膜具有缺陷34之情況)上時，於差異影像35測出 梵度彳§號值之差之缺陷36。差異影像檢測部13之輸出，係 5於缺陷檢測部W比較差異影像與事先設定好之判定值（閾 值）’超過判定值時，測出有缺陷。該差異影像35係直接顯 不於顯示部9，或得到2值化之影像37及2值化之缺陷38之信 號’因此可自動地測出缺陷。此外，由第11圖可清楚明白， 已除去於第10圖說明之間隙61之差異影像62。以下，針對 10該原理以第U圖進行說明。 第丄2圖係說明本發明之原理之圖，且顯示格子狀螢光 體塗布膜60與影像切割之區塊241之關係。此外，區塊241 係用以為了測定格子狀榮光體之塗布膜60之格子間距進行 〜像切副之區塊’且第11圖之區塊231與232並不必相同， 為Π樣大小。首先，為了測定格子狀勞光體塗布膜 =之袼子間距’先攝影塗布有格子狀螢光體之基板，並於 影像輸入部由攝影之影像切出區塊241。接著，求出格子狀 營光體塗布_之圖形之縱向像素及橫向像素之間距。 求出間距之方法’例如，於第12圖中求出崎光體之亮Wait for the panel manufacturing line, and it is real; see high-speed defect inspection, and cheap graphics defect inspection equipment and _ defect inspection methods. Picture of the Ministry and the public ① The missing 1 ^ inspection device includes a camera # 彡 部, a moving mechanism '"image processing section, a display section, and a control section. This photography section is used for the camera section 2 ::: 上 之 萤For a stripe pattern of a light body, the aforementioned moving mechanism is == a person who moves along the aforementioned pattern, and the image processing unit is input; the image from the other photography unit displays the input of the image processing unit. It is used to describe the moving mechanism section and the front, and the control section is used to control the front including the image processing section, the difference: the 'a' the aforementioned image processing section, the image round-in section system section, and the defect detection section. The direction of the stripe pattern of the nine-body firefly clip is 408269. 4 Shaoxing Image Detection Department is used to compare at least two image data with the aforementioned figure. "According to the comparison 5 10 15, the pattern defect inspection device of the present invention includes a mechanism portion, an image processing portion, a display portion, and a grid-shaped phosphor formed on a substrate that controls a photographing portion and a moving shadow. ; Cloth = Department of Photography. The Hai moving mechanism department is used to make the aforementioned photographing unit along the aforementioned figure, and the image processing unit inputs the image signal from the aforementioned photographing unit. The person who is not in front of the department displays the aforementioned image ^, The display is used to control the aforementioned axonal μ, and the control unit is a private movement mechanism unit and the aforementioned image processing unit. The image processing unit includes an image input unit and a 'thumb shadow unit'. Two of the detected patterns = ;! Calculate the image of at least 2 areas that is an integer multiple of the lattice spacing of the aforementioned grid-like phosphor coating film, and compare with the above-mentioned ϋ 光 二 发 B: The pattern defect inspection method includes photographic formation In the step of the stripe pattern of the substrate, the direction of the stripe pattern of the phosphor is detected by the image material obtained by the foregoing photography. The comparison of the image data according to the foregoing "steps of two less image data, And second ratio = the step of the defect of the positional relationship. And the aforementioned pattern is detected based on the result of the rigid speed ratio. The pattern defect inspection method of the present invention includes photographing the pattern of the grid-like glare body coating film formed on the substrate The step of calculating the grid spacing of the pattern of the grid-shaped glazed body coating film from the image data obtained by the foregoing photography, and comparing the image data of at least two domains with integer multiples of the grid spacing of the large slave image data from the aforementioned image data. The steps and the steps of detecting the defects of other graphics according to the aforementioned comparison ~ fruit. C square package method j Detailed description of the preferred embodiment Fig. 1 is a diagram showing an embodiment of a pattern defect inspection device of the present invention. In the first figure, a glass substrate mounting platform such as a 1 series display, a glass substrate such as a plasma display, and a phosphor (phosphor coating film) 3 series R, G, and 6 are used. The light source for ultraviolet illumination that makes the phosphor 3 emit light, 5 is an optical system equipped with a lens and R, G, and B color filters, 6 is a photography section such as a line-sensing camera for photography, and 7 is used for photography The part 6 and the light source 4 are moving mechanism parts that move along the broken substrate 2 for scanning on the glass substrate 2. $ is an image processing part for detecting defects such as pinholes, 9 is a color screen that displays or prints inspection results And printers 10 is used to drive the movement: the driving part of the mechanical part 7, n is the control part used to control the image processing part 8 and the driving part, and 15 is the operation part, and is used to perform the optional part of the inspection device As described later, the image processing unit 8 is composed of the image input unit 12, the difference image detection unit 13, and the defect detection unit 14. The light source 4 is not limited as long as it is a light source that emits light from a phosphor coating film. It is a source of ultraviolet rays, and it can also be other electromagnetic waves such as gamma rays or X-ray particles. Figure 2 shows the placement of the pattern defect inspection device a shown in Figure 1, a glass substrate 2 and a photography section 6 The enlarged view is the same as that in the first figure = the same reference number is attached. The mounting table 1 mounts the glass substrate 2 during inspection, and the glass substrate of the V, plasma head display panel is coated with a phosphor such as red (R) In order to check the coating state widely, the glass substrate coated with the red (R) phosphor coating film is transported in the direction indicated by the arrow and fixed at a predetermined position shown in FIG. 2 to check for defects. The same applies to the lines coated with green (G) phosphor coating films and the lines coated with blue (B) phosphors. In addition, the size of the 'glass substrate in this embodiment is 1460 mm x 1030 mm, but it is not limited to this. The 21 is a part of the moving mechanism 7 and is a supporting member for supporting the photographing section 6 and the ultraviolet light source and light source 4. In order to inspect a glass substrate, the imaging unit 6 is configured as shown in the figure to arrange four line-sensing cameras in a row to shield a glass substrate having a width of 1030 mm. The imaging width of the x-line sensor cameras is about 260mm, and the field of view between the line-sensing cameras is partly heavy. The phosphor 3 is excited by the ultraviolet rays 22 from the ultraviolet light source 4 and, for example, an image of the phosphor 3 which is excited red) is taken by the photographing unit 6 through the optical system 5. The supporting member 21 moves from the right end to the left end at a constant speed in the γ direction of the red (R) phosphor 3 and scans the entire surface of the glass substrate. This operation will be described in detail below. The ultraviolet light source 4 is used to irradiate ultraviolet rays 22 on a glass substrate 2 such as a plasma display, so that the phosphor 3 that has been coated (including coating by printing) is illuminated. The photography section 6 is used to take pictures of the luminescent shirt. At this time, according to the type (R, G, B) of the phosphors to be detected, a color filter corresponding to each color of phosphors can be mounted on the shirt section 6. The image taken by the photographing section 6 is sent to the image processing section 8. Fig. 3 is a diagram illustrating the principle of detecting a pinhole in a pattern defect inspection device of the present invention. FIG. 3 shows the case where the phosphors 3 of each color of R, G, and B are stripedly coated on the glass substrate 2 cycle 200408269. The following description is for the glass substrate coated with the phosphors 3 of R, G, and B. However, the actual manufacturing line is of course as described above, and is checked every time the phosphors of each color are sequentially coated, and When a defect is detected, the next 5 phosphor coating or printing steps are stopped, and the defective glass substrate is washed, and the phosphor is recoated or printed again, which has the advantage of unnecessary coating or printing. In addition, since the inspection must be performed in accordance with the working sequence of the manufacturing line, a pattern defect inspection device with a fast inspection speed must be used. The image data captured by the photography unit 6 is sent to the image processing unit 8, and the image 10 image input unit 12 is input, and the difference image is stored in the memory unit (not shown) in order to obtain the difference image. The image turn-in section 12 measures the direction of the striped pattern of the phosphor, and divides the stored image data into a plurality of blocks. For example, a 4 × 4 block (hereinafter referred to as 4 × 4 block, etc.), 8 × 8 block, or 32 × 32 block 31 and 32, which are generally known, are cut out and output to the difference image detection section 13. Blocks 31 and 32 are the most appropriate unit blocks for detecting defects, and the size is appropriately set experimentally by inspection speed, processing speed, and defect detection accuracy. The difference image detection unit 13 compares the image data of at least two places by comparing the block 31 and the block 32. The comparison method is to perform a difference image detection 33 between the block 31 and the block 32 by comparing the block ^ and the block 302 with the degree signal values of other pixels. When there is a defect 34 such as a pinhole in the illuminant 3 (Fig. 3 shows that the phosphor of the display clock has a defect 34), the difference 36 in the difference image 35 is measured as the difference in the luminance signal value. The output of the difference image detection unit 13 is compared with the predetermined value (threshold value) of the difference image and the defect detection unit 14. When the determination value is exceeded, it is detected that 11 200408269 is defective. The difference image 35 is directly displayed on the display portion 9 or a signal of the binarized image 37 and the binarized defect 38 are obtained, so that the defect can be automatically detected. Therefore, the blocks 31 and 32 are sequentially moved to detect the difference image of the entire glass 5 substrate, thereby inspecting all the defects of the stripe-shaped glory body. In addition, memorizing these inspection data in the memory section (not shown) and analyzing the inspection data can also contribute to manufacturing quality management. Also, the aforementioned comparison method is to explain the comparison of the luminance signal value, but it is not limited to this. Of course, it is also possible to detect a 10 difference image by comparing the luminance statistics of the image signal. Moreover, in the pattern defect inspection device of the present invention, the positional relationship between the block 31 and the block 32 must be in the long direction of the phosphor (the positional relationship between the top and bottom in FIG. 3). This relationship must also be maintained when the block 32 moves. This reason will be described using FIG. 4. 15 In Figure 4, the phosphor 3 is coated laterally on the glass substrate 2, and the blocks 31 and 32 have a positional relationship in a direction perpendicular to the stripe direction of the phosphor, as shown in the figure. The relationship above and below. In addition, each 邛 in FIG. 4 is the same as that in FIG. 3 and is given the same reference numeral. 41 is a difference image, and 42 is a binary image. In Figure 4, when the difference image detection 33 between block 31 and block 32 is performed, the positions of the phosphors 3 of each color in block 31 and the positions of the phosphors 3 of each color in block 32 are not positive. When the ground is the same, that is, when the positions of the phosphors 3 of each color are deviated, in the difference image 41, in addition to the defect, the difference signal of the current color of the laborer 3 of the existing block 31 and block 32 is displayed. Streak 43, and streak-like defects were erroneously measured. Therefore, the output to the binarized 12 WU408269 shirt image 42 also has a striped binarized difference value 44, which is a false measurement. Therefore, as shown in FIG. 4, when the direction of the stripes of the printed pattern is different from the arrangement direction of the blocks 31 and 32, the positions of the stripes of r, 0, and b in the two block areas must be correctly consistent to achieve the foregoing. The required processing is the calculation processing of the second and other magical figures related to the mental image, so it is quite difficult to implement a pattern defect inspection device that can be installed on the clothing line. However, as shown in FIG. 3, when the direction of the stripes of the printed pattern is the same as the arrangement direction of the blocks 31 and 32, as explained using FIG. 2, the moving direction of the photographing section 6 is the same as the length of the stripes. Therefore, the alignment of the blocks 31 and 32 does not need to be aligned with the stripe group. The alignment can be performed only by matching the blocks with the block 32, so the alignment can be extremely simple. For 5 ', if the arrangement direction of the photographing section 6 is the X axis and the moving direction of the photographing section 6 is the Y axis, the positions of the blocks 31 and 32 in the X axis direction are such that the moving mechanism 7 is always maintained as Fixed, no positioning is required, and as long as the positions of the blocks 15 and 31 in the 15 direction are the same, the positions of the stripes of each color in the block 31 and the block 32 can be easily consistent, so it can be easily Only defects 34 were detected. Next, when using the pattern defect inspection method shown in Fig. 3, it is necessary to measure the direction of the stripes on the glass substrate. This will be described with reference to Figs. 5 and 6. FIG. 5 shows a state where stripe phosphors r, g, and B are longitudinally coated on the glass substrate 2 and photographed with the pattern defect inspection device shown in FIG. I, and the brightness of the photographed image is processed by processing The signal measures the direction of the stripes. That is, 51 indicates the luminance signal value of the horizontal pixels (adding the projection waveform), and 52 indicates the luminance signal value of the vertical pixels (adding the projection waveform). 53, 54 13 200408269 is the zero value of the brightness signal, and 55 and 56 are the judgment values (threshold values) used to detect the phosphor. The brightness signal of the photographed image is processed by the image input section 12 and the direction of the figure is defined by using the direction in which the measured value exceeds the judgment value periodically. That is, the direction in which the luminance signal value 52 which exceeds the determination value periodically is measured is the lengthwise direction of the 5-striped phosphor. FIG. 6 shows a state where stripe phosphors of r, g, and B are coated on the glass substrate 2 in the lateral direction. Therefore, as in Fig. 5, the direction in which the luminance signal value 52 which periodically exceeds the determination value is measured is the length direction of the striped phosphor. In addition, the reference numerals of the parts in FIG. 6 correspond to the reference numerals of the parts in FIG. 5. 10 According to another embodiment of the present invention, FIG. 7 is used for description. The embodiment shown in Fig. 7 shows a case where the stripe-shaped phosphors 3 are coated on the glass substrate 2 in the lateral direction. In addition, the same reference numerals are assigned to the parts in FIG. 7 that are the same as those in FIG. 3. In FIG. 7, the positional relationship of the block areas 71 and 72 used to detect the difference image is the length direction of the stripes of various colors, that is, in FIG. 7, the blocks 71 and 72 are arranged in a horizontal position. relationship. When such a positional relationship is formed, the positional relationship of the stripes of each color in the block area 71 is the same as the positional relationship of the stripes of each color in the block area 72. Therefore, when the difference image detection of the block areas 71 and 72 is performed, 33 A signal of defect 36 is obtained from the difference image block. Therefore, a binarized defect view 2 20 binarized signal is obtained in the binarized image 37. Next, an example of the operation of the pattern defect inspection apparatus according to the present invention 'will be described using FIG. 8. First, following the first step, a glass rider coated with a stripe-shaped camping button (such as ⑻fluor button) is fixed on the mounting table! At that time, the photographing department 6 uses the moving mechanism 7 to perform the inspection in the inspection area from step 20042 to step 02. This is the point at which the photographing unit 6 makes the phosphor 3 excited first by ultraviolet light, such as a red (R) phosphor, receive light. All the processing steps will be based on this. In the third step 103, first, the five directions of the printed stripe-shaped glare body are determined. That is, as described with reference to Figs. 3 and 4, the determination of the direction of the striped phosphor in the present invention is extremely important for defect inspection such as pinholes. This directivity determination is based on the image signal of a fierce pixel X pixel directly detected by the photographing unit 6 such as the phosphor 3 excited by ultraviolet light, such as a red (R) phosphor. The method illustrated in Fig. 6 determines the direction of the fringe of the fluorescent body 10. In addition, 32 pixels can cover the stripes of a plurality of phosphors, so it is sufficient to determine the number of pixels of directivity. The fourth step 104 is to decide the direction of the comparison block. That is, the positional relationship between the two blocks to be compared is determined according to the direction of the striped phosphors determined in the third step as described above. For example, when the direction of the stripe-shaped phosphor is vertical as shown in FIG. 3-15, the positional relationship between the two blocks is selected to be a vertical relationship, and the direction of the stripe-shaped phosphor is horizontal as shown in FIG. 7. At this time, choose to make the positional relationship between the two blocks a horizontal relationship. In the embodiments described above, the positional relationship between the two blocks is a positional relationship without gaps and close contact with each other. However, according to the processing method, a part of 20 blocks may be overlapped, or a gap may be formed between the blocks. Capture images. The fifth step 105 maintains the positional relationship between the two blocks determined in the fourth step, and includes the entire glass substrate including the plasma display panel to be inspected, and obtains a difference image between the two blocks. The sixth step 106 is to compare the difference image obtained in the fifth step to obtain the party number h of 200408269 and the defect judgment value (threshold value). If there is a signal difference that is more than the defect judgment value, determine the difference. For graphic defects. In addition, the defect judgment value (threshold value) is set to about 50% of the highest value obtained from the image signal. However, it can be appropriately adjusted by inspection 'or during the inspection as needed, and the setting can be changed. For example, the steps described above are sequentially performed on the phosphors of various colors, such as R, G, and B, which have been printed or coated with a stripe phosphor. Of course, when a defect is detected in the R phosphor inspection as described above, the step of coating or printing the phosphor of the next color is suspended, and the glass substrate enters the step of removing the phosphor, and regenerates. The above is a detailed description of the present invention. The graphic defect inspection method of the present invention adopts a method of detecting defects with a difference image of 2 blocks, so there is a difference between the image of the shirt and the test result. Which one has flaws. A method for solving this disadvantage will be described using FIG. 9. Figures 15 to 9 show the methods used to determine which block areas are deficient from the majority of difference images. Figure 9 shows a part of the glorious body 3 with G, B, and explains the situation of the pinhole defect society on the money-like fluorescent reading of 6. The glass substrate is omitted. First, the block illusion is in the field, and the block 83 is guarded in the field on the 2nd. The difference image 84 between the two blocks 82 and M shows a defect 087. In "quotation", it is determined that defect 81 is located in the block area of block 82 or block 83. Next, make the block phantom move the i block. That is, block 82 is moved to money 2, and the block is moved to area 3, and the difference image 85 between the block and the area is obtained. When a defect 88 is detected, it can be determined that area 2 has a missing fe81 and , Make block 82 and block magic move another block. That is, when the area 16 200408269 block 82 is moved to the area 3, and the block 83 is moved to the area 4 to obtain the difference image 86 of the block fantasy and the block 83, if the difference image has no defects, it can be known that the area 3 No defects. Therefore, according to this method, it can be known that domain 2 has defects' and domains 1, 3, and 4 have no defects. 5 # ， Describes another embodiment of the present invention. In the foregoing embodiment, when the phosphors 3 of r, G, and B are periodically coated on the glass substrate 2 as shown in FIG. 3 in a stripe shape, the fluorescence can be measured with high accuracy. Coating defects of the body, however, when the phosphor coated on the glass substrate 2 is not a phosphor coating film having a uniform stripe structure as shown in FIG. 3, there is a problem that the method is applicable to the thermal method. Fig. 10 is a diagram explaining the principle of detecting a pattern defect such as a pinhole of a pattern defect inspection device used to explain the problem. In FIG. 10, the same symbols as those in FIG. 3 are assigned the same _ symbols. 60 is a glazed body coating film applied to the glass substrate 2, but it is formed in a grid shape. In other words, red (R), green (G), and blue laborers are periodically repeated in the horizontal direction. Longitudinal system consists of a gap area and a knife-like island. Hereinafter, the above-mentioned phosphor coating film is referred to as a grid-like phosphor coating film. The following is a description of the inspection method for the coating defects of the phosphor using the difference image, and the description of the pattern defects such as the 20-hole pin of the grid-like glory coating film with the structure described above. As in FIG. 3, the luminance signal values of the pixels of the blocks 31 and 32 are compared in the difference image detection 33. When a defect 34 such as a pinhole is on the grid-like phosphor coating film 60 (in FIG. 10, it is shown that the grid-like phosphor coating film 60 of R has a defect 34), the difference image 35 is measured Defect 36 of the difference in luminance signal value. 17 In the case of the difference, the gap 35 of the difference image 62 is measured as the artifact 35. That is, when comparing the pixels of block 31 and block 32, it can be clearly understood that the gap portion 6! Is located at a different position in block 31 and block 32, and thus the difference image 35 in the rotation of the difference image inspector appears The difference image between the defect% and the gap M, 5 so the difference image 62 between the defect 36 and the gap 61 cannot be distinguished. Therefore, the binarized image 37 also obtains the binarized defect% and the gap] signal, so that the defect 38 can be automatically detected. Fig. 11 is a diagram for explaining the principle of detecting pattern defects such as pinholes by the pattern defect inspection device of the present invention. In the figure u, the same reference numerals are assigned to the same person 10 as that of the level 10. In Fig. U, the glazed body coating film 60 is coated on the glass substrate 2 and is configured in a grid pattern. That is, the phosphors of red gadolinium, green (G), and blue gadolinium are periodically applied in the lateral direction. In the longitudinal direction, the phosphors of red, green (G), and blue (B) are separated by gaps 61 and separated into island shapes. In addition, the following description is directed to the glass substrate coated with the fluorescent 15-coated fluorescent coating film 60 of R, G, and Bg colors. However, as mentioned above, the actual manufacturing line sequentially coats each color of the fluorescent coated coating 6 Check at 0 o'clock. The image data captured by the photography unit 6 is sent to the image processing unit 8 and then input to the image input unit 12. In the image inputting section 12, the distance between the rafters of the pattern of the rafter-shaped phosphor coating film is calculated, and the image is divided into a plurality of regions of 20 pieces. Then, it is cut into blocks 231 and 232 and output to the difference image detection section 13. The relationship between the sizes of the blocks 231 and 232 and the lattice-shaped phosphor coating film 60 will be described later. In the difference image detection section 13, block 231 and block 232 are compared. The comparison method is to perform a difference image detection 33 between block 231 and block 232 by comparing the pixel value 18 200408269 degrees of the respective pixels of block 231 and block 232. When the defect 34 such as pinholes is on the phosphor coating film 60 (in Fig. 11, it is shown that the phosphor coating film of R has the defect 34), the value of the Fandu 梵 § number is measured on the difference image 35 Poor defect 36. The output of the difference image detection unit 13 is that when the difference detection unit W compares the difference image with a predetermined determination value (threshold value) 'exceeding the determination value, a defect is detected. The difference image 35 is directly displayed on the display portion 9 or a signal ′ obtained by binarizing the image 37 and the binarizing defect 38 is obtained, so that the defect can be automatically detected. In addition, it is clear from FIG. 11 that the difference image 62 in the gap 61 described in FIG. 10 has been removed. In the following, this principle will be described with reference to Fig. U. Fig. 2 is a diagram illustrating the principle of the present invention, and shows the relationship between the lattice-shaped phosphor coating film 60 and the image-cut block 241. In addition, the block 241 is used to measure the grid pitch of the coating film 60 of the grid-like glare body ~ to the block of the image-cutting pair ', and the blocks 231 and 232 in FIG. 11 are not necessarily the same, and have a Π-like size. First, in order to measure the lattice-shaped gloss coating film = 袼 子 间距 ', a substrate coated with a lattice-shaped phosphor is photographed first, and a block 241 is cut out from the photographed image at an image input section. Next, the distance between the vertical pixels and the horizontal pixels of the pattern of the grid-shaped light-coated body is obtained. How to find the distance ’, for example, in Figure 12
2〇度值之縱向與橫向之像素值之加算值。此外，於此係以R 榮光體作說明，然而G、B之榮光體亦為同樣間距，因此省 略說明。 又，求出像素之亮度值之加算值，係因為1像素之亮度 值係小之党度值，而某程度之大值可測定正確之間距。 19 於第12圖中，242係縱向之像素之亮度值之加算值，243 係表示預定之閾值。朗值係事先藉實驗設定，例如為亮 度值之242之70%，以求出正確之間距。因此針對超過間 值243之亮度值檢測像素間之間距(pxi)，然後求出各個平均 值Px。即，Addition of 20-degree vertical and horizontal pixel values. In addition, here, the R glory body is used for explanation, but the G and B glory bodies are also at the same pitch, so the explanation is omitted. In addition, the added value of the brightness value of the pixel is obtained because the brightness value of 1 pixel is a small degree value, and a large value to a certain degree can be used to determine the correct distance. 19 In Figure 12, 242 is the added value of the luminance value of the vertical pixels, and 243 is the predetermined threshold. The brightness value is set by experiment in advance, for example, it is 70% of the brightness value of 242 to obtain the correct distance. Therefore, the pixel-to-pixel distance (pxi) is detected for the luminance value exceeding the interval 243, and then each average value Px is obtained. which is,
PxPx
於此，i=1、2.......n〇 同樣地，244係橫向之像素之亮度值之加算值，245係 表不預定之閾值。因此，針㈣過閾值245之亮度值檢測像 素間之間距(Pyi)，然後求出各個平均值py。即，Here, i = 1, 2 .... n0 Similarly, 244 is the added value of the luminance value of the horizontal pixels, and 245 is the predetermined threshold. Therefore, the inter-pixel distance (Pyi) is detected by the brightness value exceeding the threshold 245, and then each average value py is obtained. which is,
於此，i=l、2.......m。 根據前述求出縱向及橫向之平均間距Ρχ、Py。根據該 等平均間距Px、Py決定區塊231及232之大小。即，區塊231 及232之大小中，至少將區塊231與區塊232之排列方向之大 小設定成該方向之平均間距之整數倍。第η圖之例中，區 塊231、232係配置成縱向之位置關係，區塊之排列方向係 縱向，因此將區塊之縱向（Υ方向）之大小設定成Py之整數 倍。 且，藉差異影像檢測部13，為了檢測差異影像將區塊 231及232設定成同樣大小，並使該等區塊231與區塊232依 序移動，而針對玻璃基板全體，藉進行差異影像之檢測可 200408269 檢查格子狀螢光體塗布膜60全部之缺陷。此外，藉使該等 檢查資料記憶於記憶部（圖未示），且分析檢查資料，可有助 &製造上之品質管理。又’前述比較方法係說明藉亮度信 ?虎值所作之比較’但並不限定於此，藉影像信號之亮度統 5 計圖所作之比較等亦當然可檢測差異影像。 再對本發明之另一實施例，使用第13圖說明。此外， 對於與第11圖相同者附上同一標號。第13圖所示之實施例 係顯示橫向地塗布格子狀螢光體塗布膜60之長向於玻璃基 板2上之情況。第13圖之例中，用以檢測差異影像之區塊領 10域71及72之位置關係，係配置成以橫向為排列方向。如前 述之位置關係中，於區塊領域之大小中至少將與排列方向 同為橫向（X方向）之大小設定為前述平均間距ρχ之整數 倍。如此一來，區塊領域1内之各色之格子狀螢光體塗布膜 60之位置關係，與區塊領域72内之各色之格子狀螢光體塗 15布膜60之位置關係相同，因此實行區塊領域71、72之差異 影像檢測33日^ ’於差異影像35中僅得到缺陷％之信號。因 此’於2值化影像37中，得到2值化之缺陷38之2值化信號。 此外，於第13圖所示之實施例中，區塊之縱向之大小並不 一定要設定為Py之整數倍。有關第11圖所示之實施例中區 20 塊之橫向之大小亦相同。理由後述。 此外，第11圖所示之例中，顯示出區塊231與區塊232 之位置關係係於Y方向（第11圖中為縱之位置關係）上準確 地對位之情況，但於使區塊231與區塊232移動時，亦以保 持该關係為佳。其理由係如以第;[〇圖所作之說明，為了要 21 200408269 檢測2個區塊231與232之差異影像，所以兩區塊並不必為同 一圖开^。此外’為同一圖形時，並不特別要準確地對位， 但至少將區塊231與區塊232或區塊71與區塊72之排列方向 上之區塊大小設定成該方向之平均間距之整數倍時，區塊 5 231與區塊232或區塊71與區塊72之對位會相當容易。 換言之’不管攝影部6之移動方向係朝X軸方向或γ軸 方向，於利用移動機構7使區塊231與區塊232於X軸上之位 置經常維持固定而不需對位之情況下，只要使區塊231與區 塊232於Y方向上之位置一致時，就可簡易地使區塊231與區 10塊232内之各色之條紋之位置一致，因此可簡易地僅檢測出 缺34。又，前述說明中，係針對求出2個區塊(2領域)之差 異影像進行說明，但是為了提昇檢查效率時，當然可輕易 地同時檢查多於2個區塊之區塊。例如，同時檢查於橫向及 縱向分別配置2個區塊之合計4個之區塊時，區塊之大小 I5中仏向可5又疋為PX之整數倍，且縱向可設定為Py之整數 倍。此外，攝影部6之排列方向（第2圖所示之攝影部6之例 中，攝影部6之排列方向為χ方向，攝影部6之移動方向為γ 方向），與區塊231及區塊232或區塊71及區塊72之排列方向 同一方向時，容易受到照相機之透鏡等光學系統之偏歪影 20響。所以，使攝影部6之移動方向與前述區塊之排列方向一 致時’攝影部6之排列方向與前述區塊之排列方向為不同方 向’因而可更加提昇檢查精度。 接著，針對本發明之圖形檢查裝置之動作之一例，使 用第14圖作說明。首先，第χ步驟2〇1，將塗布有格子狀螢 22 200408269 光體塗布膜60(例如（R)螢光體）之玻璃基板2搬入並固定於 載置台1時’攝影部6藉移動機構7由Y軸之原點〇開始攝影。 第2步驟202，係進行檢查領域之檢測。此係攝影部6 使最先藉紫外線光激發之螢光體3，例如紅色(11)營光體受 5 光之時點’且全部之處理步驟將以此為基準而開始。 第3步驟203中，如以第12圖所作之說明般算出格子狀 螢光體塗布膜60於X方向與Y方向上之間距。 第4步驟204中，決定比較區塊之位置關係及大小。即， 決定用以比較之2個區塊之位置關係，且決定2個如前述以 10第3步驟算出之X方肖、γ方向上之間距之整數倍之區塊。 此時，如前述，至少使區塊於排列方向上之大小為該方向 之平均間距之整數倍之大小。 以上說明之各實施例中，2個區塊之位置關係為沒有間 隙之互相密著之位置關係，然而依處理方法不同，區塊亦 15可重疊，或區塊間亦可具有空隙而攝取影像。 第5步驟205中，係維持第4步驟所決定之2個區塊之位 置關係，並含括檢查對象之電製顯示面板等之玻璃基板全 體，求出2個區塊之差異影像。 ’ tbk由第5步驟所求出之差異影像得到 2〇之亮度信號值，與缺陷判定值（閾值），若有較缺陷判定值還 高之信號差值時，判定其為圖形缺陷。此外，缺陷判定值（閣 值)係設定為由影像信號得到之最高值之約鄕左右，然而 亦可由貝驗《於才欢查之過程中依據需要地適當調節，並 可改變設定。 200408269 如前述之步驟係依序對已印刷或塗布之格子狀螢光體 塗布膜60之諸如R、G、B之各色螢光體進行。當然，如前 述於R螢光體檢查中測出有缺陷時，中止下一顏色之螢光體 之塗布或印刷之步驟，且該玻璃基板進入除去螢光體之步 5 驟，並再生。 以上，針對本發明詳細地說明，但本發明並不限定於 前述記載之電漿顯示面板等之玻璃基板之圖形缺陷檢查裝 置及圖形缺陷檢查方法，當然可廣泛應用於前述以外之圖 形缺陷檢查裝置及圖形缺陷檢查方法。 10 發明效果 如以上說明，本發明可自動地檢測出於電漿顯示面板 等之玻璃基板上之已塗布或印刷成條紋狀或格子狀之R、 G、B之各色螢光體，又，可於不影響條紋狀、格子狀等之 螢光體塗布圖形之情況下，高感度地檢查圖形缺陷。又， 15 由於可自動地進行電漿顯示器等之細微之圖形之缺陷檢 查，因此可輕易地設置於電漿顯示器等之顯示面板之製造 線，而可實現可進行高速之缺陷檢查且價格便宜之圖形缺 陷檢查裝置及圖形缺陷檢查方法。 I：圖式簡單說明3 20 第1圖係顯示本發明之圖形缺陷檢查裝置之一實施例 之方塊圖。 第2圖係顯示本發明之一實施例之一部份之放大圖之 圖。 第3圖係說明本發明之動作原理之圖。 24 200408269 第4圖係說明本發明之動作原理之圖。 第5圖係用以檢測本發明之螢光體之條紋方向之原理 說明圖。 第6圖係用以檢測本發明之螢光體之條紋方向之原理 5 說明圖。 第7圖係說明本發明之另一實施例之動作原理之圖。 第8圖係說明本發明之缺陷檢查方法之一實施例之圖。 第9圖係說明本發明之其他之實施例之圖。 第10圖係應用第3圖之動作原理於格子狀螢光體塗布 10 膜之情況之說明圖。 第11圖係說明本發明之動作原理之圖。 第12圖係說明本發明之動作原理之圖。 第13圖係說明本發明之其他之實施例之動作原理之 圖。 15 第14圖係說明本發明之缺陷檢查方法之一實施例之 圖。 第15圖係顯示習知圖形缺陷檢查裝置之一例之動作之 圖。 【圖式之主要元件代表符號表】 1.. .載置台 2.. .玻璃基板 3.. .螢光體 4.. .紫外線照明用光源 5.. .光學系統 25 200408269 6.. .攝影部 7.. .移動機構 8…影像處理部 9.. .顯示部 10.. .驅動部 11.. .控制部 12.. .影像輸入部 13.. .差異影像檢測部 14.. .缺陷檢測部 15.. .操作部 21.. .支持構件 22.. .紫外線 31，32,71，72,82,83,231，232,241 …區塊 33.. .差異影像檢測 34,36,81，87,88...缺陷 35,41，62,84,85狐..差異影像 37,42···2值化之影像 38.. .2.化之缺陷 43.. .條紋 44.. .2.化差分值 51··.橫向像素之亮度信號值 52.. .超過判定值之亮度信號值 53,54,93,94·.·亮度信號之零值 55,56,243,245···閾值 200408269 60.. .螢光體塗布膜 61.. .間隙 63…間隙之2值化影像 91.. .玻璃基板之一部份 92.. .缺陷部 95·.·紅色螢光體之亮度信號值 96…沒有塗布螢光體之間隙部分之亮度信號值 97、99…亮度信號值之閾值 98.. .以單點虛線C表示之部分之亮度信號值 242…縱向之像素之亮度值之加算值 244…橫向之像素之亮度值之加算值 A，B，C···單點虛線 R、G、B.··螢光體Here, i = 1, 2 ....... m. The average pitches Px and Py in the vertical and horizontal directions are obtained from the foregoing. The sizes of the blocks 231 and 232 are determined based on the average pitches Px and Py. That is, among the sizes of the blocks 231 and 232, at least the size of the arrangement direction of the blocks 231 and 232 is set to an integer multiple of the average pitch in that direction. In the example in FIG. N, the blocks 231 and 232 are arranged in a vertical positional relationship, and the arrangement direction of the blocks is in a vertical direction. Therefore, the size of the block in the vertical direction (Υ direction) is set to an integer multiple of Py. In addition, the difference image detection unit 13 sets the blocks 231 and 232 to the same size in order to detect the difference images, and moves the blocks 231 and 232 in sequence, and performs the difference image for the entire glass substrate. Inspection 200408269 can inspect all the defects of the lattice-shaped phosphor coating film 60. In addition, by storing such inspection data in the memory (not shown) and analyzing the inspection data, it can help & manufacturing quality management. Also, the aforementioned comparison method is for explaining the comparison made by the luminance signal and the tiger value, but it is not limited to this. Of course, the difference image can also be detected by the comparison made by the luminance signal of the image signal. Another embodiment of the present invention will be described with reference to FIG. 13. In addition, the same reference numerals are assigned to those that are the same as those in FIG. 11. The embodiment shown in Fig. 13 shows a case where the lengthwise direction of the grid-like phosphor coating film 60 is applied to the glass substrate 2 in the lateral direction. In the example in FIG. 13, the positional relationship between the block domains 71 and 72 used to detect the difference image is arranged in the horizontal direction. In the positional relationship described above, at least the size in the horizontal direction (X direction) that is the same as the arrangement direction in the size of the block area is set to an integer multiple of the aforementioned average distance ρχ. In this way, the positional relationship of the grid-like phosphor coating films 60 of each color in the block area 1 is the same as the positional relationship of the grid-like phosphor coating 15 cloth film 60 of each color in the block area 72. The difference image detection in the block areas 71 and 72 was performed on the 33rd ^ 'Only the% defect signal was obtained in the difference image 35. Therefore, in the binarized image 37, a binarized signal of the binarized defect 38 is obtained. In addition, in the embodiment shown in Fig. 13, the vertical size of the block is not necessarily set to an integer multiple of Py. The horizontal size of the block 20 in the embodiment shown in FIG. 11 is also the same. The reason will be described later. In addition, in the example shown in FIG. 11, it is shown that the positional relationship between the block 231 and the block 232 is accurately aligned in the Y direction (vertical positional relationship in FIG. 11). When the blocks 231 and 232 move, it is better to maintain the relationship. The reason for this is as explained in the figure; [0], in order to detect the difference image between the two blocks 231 and 232 in 21 200408269, so the two blocks need not be opened for the same picture ^. In addition, when it is the same figure, it is not particularly necessary to accurately align, but at least the block size in the arrangement direction of block 231 and block 232 or block 71 and block 72 is set to the average distance in that direction. For multiples, the alignment of block 5 231 and block 232 or block 71 and block 72 will be quite easy. In other words, regardless of whether the moving direction of the photographing section 6 is toward the X-axis direction or the γ-axis direction, when the position of the block 231 and the block 232 on the X-axis is constantly maintained without using alignment by the moving mechanism 7, As long as the positions of the blocks 231 and 232 in the Y direction are consistent, the positions of the stripes of each color in the block 231 and the block 10 232 in the region 10 can be easily matched, so that only the defect 34 can be easily detected. In the foregoing description, the difference image of two blocks (two fields) was obtained. However, in order to improve the inspection efficiency, it is of course possible to easily check the blocks of more than two blocks at the same time. For example, when checking a total of 4 blocks with 2 blocks arranged horizontally and vertically at the same time, the size of the block I5 can be 5 and can be an integer multiple of PX, and the vertical can be set to an integer multiple of Py. In addition, the arrangement direction of the photographing section 6 (in the example of the photographing section 6 shown in FIG. 2, the arrangement direction of the photographing section 6 is the χ direction and the moving direction of the photographing section 6 is the γ direction), and the block 231 and the block When the arrangement directions of 232 or block 71 and block 72 are the same, they are easily affected by the distortion 20 of the optical system such as the lens of the camera. Therefore, when the moving direction of the photographing section 6 is consistent with the arrangement direction of the aforementioned blocks, 'the arrangement direction of the photographing section 6 is different from the arrangement direction of the aforementioned blocks', and thus the inspection accuracy can be further improved. Next, an example of the operation of the pattern inspection apparatus of the present invention will be described with reference to FIG. 14. First, in step χ2, the glass substrate 2 coated with the grid-shaped fluorescent film 22 200408269 light-body coating film 60 (for example, (R) fluorescent body) is moved into and fixed to the mounting table 1. 7 Start shooting from the origin 0 of the Y axis. The second step 202 is to perform inspection in the inspection area. This is the photographing unit 6 that makes the phosphor 3 excited first by ultraviolet light, for example, the point at which the red (11) camping light receives 5 light ', and all the processing steps will be started based on this. In the third step 203, the distance between the lattice-shaped phosphor coating film 60 in the X direction and the Y direction is calculated as described with reference to Fig. 12. In the fourth step 204, the positional relationship and size of the comparison block are determined. That is, determine the positional relationship of the two blocks for comparison, and determine two blocks that are integer multiples of the distance between the X-squared and the γ direction calculated as described in step 3 of 10. At this time, as described above, at least the size of the blocks in the arrangement direction is an integer multiple of the average pitch in that direction. In the embodiments described above, the positional relationship between the two blocks is a close positional relationship without a gap. However, depending on the processing method, the blocks may also overlap, or there may be gaps between the blocks to capture images. . In the fifth step 205, the positional relationship between the two blocks determined in the fourth step is maintained, and the entire glass substrate including the electric display panel and the like to be inspected is included to obtain a difference image between the two blocks. ′ Tbk obtains a luminance signal value of 20 from the difference image obtained in step 5 and a defect determination value (threshold value). If there is a signal difference higher than the defect determination value, it is determined to be a graphic defect. In addition, the defect judgment value (cabinet value) is set to about 最高 of the highest value obtained from the image signal. However, it can also be adjusted appropriately according to needs during the inspection of Yucai Yucai, and the setting can be changed. 200408269 The foregoing steps are performed sequentially on the phosphors of various colors such as R, G, and B of the grid-like phosphor coating film 60 that has been printed or coated. Of course, when a defect is detected in the R phosphor inspection as described above, the coating or printing step of the phosphor of the next color is suspended, and the glass substrate enters the step of removing the phosphor, and is regenerated. The present invention has been described in detail above, but the present invention is not limited to the pattern defect inspection device and pattern defect inspection method for glass substrates such as plasma display panels described above, and of course, it can be widely applied to pattern defect inspection devices other than the foregoing. And graphics defect inspection methods. 10 Effects of the Invention As described above, the present invention can automatically detect R, G, and B phosphors of various colors coated or printed on a glass substrate such as a plasma display panel or the like. The pattern defect is checked with high sensitivity without affecting the phosphor-coated patterns such as stripes and lattices. Also, 15 can automatically inspect the defects of fine graphics such as plasma displays, so it can be easily installed on the display panel manufacturing line of plasma displays, etc., and high-speed defect inspection can be realized at a low price. Graphic defect inspection device and method. I: Brief Description of Drawings 3 20 FIG. 1 is a block diagram showing an embodiment of the pattern defect inspection device of the present invention. Fig. 2 is an enlarged view showing a part of an embodiment of the present invention. Fig. 3 is a diagram illustrating the operation principle of the present invention. 24 200408269 Figure 4 is a diagram illustrating the operation principle of the present invention. Fig. 5 is an explanatory diagram of the principle for detecting the fringe direction of the phosphor of the present invention. Fig. 6 is a diagram for explaining the principle 5 of the fringe direction of the phosphor of the present invention. FIG. 7 is a diagram illustrating the operation principle of another embodiment of the present invention. FIG. 8 is a diagram illustrating an embodiment of the defect inspection method of the present invention. Fig. 9 is a diagram illustrating another embodiment of the present invention. Fig. 10 is an explanatory diagram of the case where the operation principle of Fig. 3 is applied to a lattice-shaped phosphor coated with 10 films. Fig. 11 is a diagram illustrating the operation principle of the present invention. Fig. 12 is a diagram illustrating the operation principle of the present invention. Fig. 13 is a diagram for explaining the operation principle of another embodiment of the present invention. 15 FIG. 14 is a diagram illustrating an embodiment of the defect inspection method of the present invention. Fig. 15 is a diagram showing an example of the operation of a conventional pattern defect inspection device. [Representative symbols for main components of the drawing] 1.............. The mounting base 2.... Glass substrate 3........... Section 7.... Movement mechanism 8... Image processing section 9... Display section 10... Drive section 11.... Control section 12.... Image input section 13... Difference image detection section 14... Detection section 15. Operation section 21. Support member 22. Ultraviolet 31, 32, 71, 72, 82, 83, 231, 232, 241 ... Block 33 ... Difference image detection 34, 36, 81, 87, 88 ... Defects 35, 41, 62, 84, 85 Fox: Difference images 37, 42 ... Binarized images 38 .. 2. 2 Defects 43 .. Stripes 44 .. 2. The difference value 51 ········································ 0 ·············· Threshold 200408269 60 .. Fluorescent coating film 61 .. Gap 63. Binary image of gap 91 .. Part of glass substrate 92 .. Defective part 95 .. Brightness signal value of red phosphor 96. None The brightness signal value of the gap portion of the coated phosphor is 97, 99 ... The threshold value of the brightness signal value is 98 .. The value of the luminance signal 242 in the portion indicated by the dotted line C is the addition value 244 of the luminance value of the vertical pixels ... the addition value A, B, C of the luminance value of the horizontal pixels. · Fluorescent body
2727