201209393 六、發明說明： 【發明所屬之技術領域】 本發明關於一種偏光膜卷材的品質判斷系統及其方 法’更詳細來說，是關於一種使用自動光學檢測機的檢測 結果資料來判斷偏光膜卷材的品質的系統及其方法。 【先前技術】 生產一種作為LCD光學材料來使用的偏光膜的製造 商’為了即時檢測高速生產的產品，使用内聯N - LINE ) 自動光學檢測系統（Automated Optical Inspection system，後面稱作自動光學檢測機）。一般來說，内聯自 動光學檢測機，根據在缺陷產生位置上標記顏色或標記條 碼，來在後續步驟中廢棄被標記的部位、或是對此類部位 再進行追加檢測。 但是，1 )上述缺陷在特定時間點上連續集中地產生的 匱况下，1 1 )對細微的缺陷等在無法明確地判斷是否為實 質的缺陷的情況下’由於上述自動光學檢測機的過負載 以及H/W性能問題，會發生不能標記全部所述缺陷位置的 問題。 沒有被標記缺陷位置的 有作為不合格產品被辨識出 性。由於這個理由，在後續 進行最終檢測的步驟，這種 產品，會有在後續步驟中也沒 來而被發送到客戶公司的可能 步驟中，有檢測員對產品再次 目視檢測作業會有需要报多的 201209393 費用和時間的問題。 另外，在現有技術中，因為只能得到僅使用在卷材內 所测出的缺陷邀_而1 ★數而成的缺陷產生率資訊，所以會有沒 慮到缺陷遺溫f B 气 (即’缺陷未被標記）的可能性的問題。 【發明内容】 [發明所欲解決的問題] I月疋#於上述問題點而完成@，本發明的目的在 I:供-種品質判斷系統及其方法，其可以提高對偏光膜 的品質判斷的正確性以及可靠性。 法，本發明的其他目的在於提供一種品質判斷系統及其方 生其藉由使用自動光學檢測機的檢測結果資料來獲取判 材品質的重要資訊以及將這些重要資訊傳達到後續步 能：由此來履行作為對於不合格卷材的早期預警系統的機 本發明的另—其他目的在於提供—種品質判斷系統及 ,法’其料預想為會出現較高缺陷遺漏率的卷材進行 自動篩選。 [解決問題的技術手段] 根據本發明的一個實施形態’提供—種偏光膜卷材的 ::判斷系統，該系統包括：儲存部，其儲存從對偏光膜 材進行缺陷檢測的自動光學檢測機所得到的檢測結果資 ，缺陷資料分析部，其根據包含於上述檢測結果資料中 201209393 的缺陷位置資訊，算出檢測對象卷材的各單位區域的缺陷 產=密度，並根據上述缺陷產生密度變為規定值以上的缺 二度異㊉區域數，算出上述檢測對象卷材的缺陷密度異 …及卷材質判斷部’其在上述檢測對象卷材的 缺陷密度異常指數戀為招a# △ 對象卷心丨^ 以上時，將上述檢測 野象卷材判斷為不合格卷材。 姓果U施例中，還包括將上述自動光學檢測機的檢測 =貝料轉換為通用格式的資料轉換部，上述缺陷資料分 «轉換為上料詩式後的㈣結 仃資料分析。 #施例中，上述缺陷資料分析部，根 迷檢測結果資料中的亮點缺陷資訊，進 剛對象卷材内的；BA 4出將上述檢 /』缺&數除以上述檢測對象卷材的長度 于| 、母個單位長度的亮點缺陷數， :述卷材品質判斷部’根據上述算出的 的免點缺陷數來判斷卷材是否為不合格， 早… 上述母個皁位長度的亮點缺陷數的算出，可 4自動光學拾.、丨 猎由上 的-點亮點缺陷之中區分被判斷為⑽ 缺和被判斷為NG的亮點缺陷來分別進行。 在-實施例中’上述檢測結果資料令 ，學檢測機所檢出的缺陷圖案資訊，上述圖：t 的寬度、長度、形態、圓形性、線形性資訊， 上述缺陷資料分析部，根據上 1，進一牛曾“ 固茱資矾之令的至少 〜出檢_象卷材的外觀缺陷數或者每個單 201209393 位長度的外觀缺陷數， 此處，上述外觀缺陷，包含：TAC (三醋酸纖維素酯片 基）膜的皺褶、PVA (聚乙缔醇）膜的皺褶、下部Μ膜的 不均勻、塗層的條紋、黏結層的條紋以及擠壓性的條紋之 中的至少一個， 上述卷材品質判斷部，可以在上述外觀缺陷數或者上 述每個單位長度的外觀缺陷數在規定的允許值以上時，將 上述檢測對象卷材判斷為不合格卷材。 在一實施例中，上述缺陷資料分析部，分別算出上述 檢測對象卷材的每個第一單位區域的缺陷產生密度以及每 個第二單位區域的缺陷產生密度， 此處，上述第一單位區域相當於將上述檢測對象卷材 劃分為NxM的行列形態的時候的各個區分區域’上述第二 單位區域相當於將上述檢測冑象卷#的長度分《l等份的 時候的各個區分區域， 上述缺陷密度異常區域數，可以相當於將上述每個第 一單位區域的缺陷產生密度為第三允許值以上的第一單位 區域的數、和上述每個第二單位區域的缺陷產生 四允許值以上的第二單位區域的數，進行相加而得2值。 在一實施例中，上述缺陷資料分析部，對於未實施塗 層狀態的偏光膜卷材以及實施塗層狀態的偏光膜卷材，分 別進行上述每個單位區域的缺陷產生密度的算出， 上述卷材品質判斷部，可以在下述的其t任意一個情 況下將偏光膜卷材判斷為不合格卷材，即：對於上述未實 201209393 施塗層狀態的偏光膜卷材’其未實施塗層的缺陷密度異常 指數在規定的第五允許值以上的情況；以及對於上述實施 塗層狀態的偏光膜卷材，其實施塗層的缺陷密度異常指數 在規定的第六允許值以上的情況。 在一實施例中，上述缺陷資料分析部，對上述未實施 塗層的缺陷密度異常指數賦予了第一加權值的值、和對上 述實施塗層的缺陷密度異常指數賦予了第二加權值的值， 進行相加而算出密度異常合計指數， 上述卷材品質判斷部，可以在上述算出的密度異常合 計指數在規定的第七允許值以上時，將上述偏光膜卷材判 斷為不合格卷材。 此處，與上述密度異常合計指數進行大小比較的上述 第七允許值，可以根據藉由上述自動光學檢測機判斷為 的缺陷未被標記的機率也就是缺陷遺漏率來計算。 種偏光膜卷材的品 根據本發明的其他實施例，提供一 質判斷方A，該I法是使用對偏光膜卷材進行缺陷檢測的 自動光學檢測機的檢測結果資料來判斷檢測對象卷材的品 質的方法’該方法包括：根據包含於上述檢測結果資料中 的缺陷位置資訊，#出檢測對象卷材的各單位區域的缺陷 產生密度的階段；根據上述缺陷產生密度變為規定值以上 的缺陷密度異常區域數，算出上述檢測對象卷材的缺陷密 度異常指數的階段’·以及在上述檢測對象卷材的缺陷密度 異常指數變為規定的允許值以上時，將上述檢測對象卷材 判斷為不合格卷材的階段。 201209393 [功效] 根據本發明的實施例，可以提高偏光膜卷材的品 斷的正確性以及可靠性。 句 另外，根據本發明的實施例，藉由使用自動光學檢测 機的檢測結果資料來獲取判斷卷材品質的重要資訊以1 這些重要資訊傳達到後續步驟，由此來履行作為對 格卷材的早期預警系統的機能，可以在後續步驟中按照卷 材品質的不同來構築不同的流程、以及可以降低追加檢測 所需要的費用以及時間。 >還有’根據本發明的實施例，可以對於預想為會 較尚缺陷遺漏率的卷材進行自動篩選。 【實施方式】 本發明可以給予多種的 例，特定的實施例如附圖所 這並非將本發明限定於特定 本發明的構思以及技術範圍 以及代替物均屬本發明的保 在進行本發明的說明時 當被判斷為反倒使本發明的 詳細的說明。另外，本說明 (例如第一、第二等）是為 構成要素的辨識標號而已。 變更，具有各種各樣的實施 示’進行詳細地說明。但是， 的實施形態，而應該理解為在 中包含的所有變化，等同替代 護範圍。 ’對相關公知技術的具體說明 要點不清楚的時候，則省略該 書在說明過程中所使用的數字 了將一個構成要素區別於其他 201209393 下面，為了幫助理解本發明首先對自動光學檢測機進 行簡短地說明。第1圖是對偏光膜卷材進行缺陷檢測的自 動光學檢測機的一個例子的圖面。 製作出來的偏光膜，是藉由卷材的形態來保存的，這 樣的偏光膜卷材，在缺陷檢測時從開卷部（未圖示）被開 卷，藉由輥12向規定的方向搬運❶此時，偏光膜1〇的搬 運速度，藉由編碼器Π等那樣的速度探測裝置讀出，用於 之後的缺陷位置資訊的生成。 此處，上述自動光學檢測機，包括具備至少一個照明 裝置21和與此對應的至少一個的攝像裝置22的光學裝置 20，由此可以在上述搬運中得到偏光膜的光學影像。 光學裝置20的構成以及設計方式，對應於需要用上述 自動光學檢測機檢出的缺陷的項目（即，種類以及内容） 的不同而不同，通常的自動光學檢測機對偏光膜進行透過 檢測、反射檢測、偏光遮斷檢測（也稱作交叉（cr〇ss )檢 測）等。 因此’光學裝置20被設計為：對應於需要檢測的缺陷 的項目’藉由選擇性地或者並列性地驅動反射照明（以及 與此對應的攝像裝置）和透過照明（以及與此對應的攝像 裝置），從而可以得到用於檢出各種各樣的缺陷類型的光 學影像。 這樣’將從光學裝置20得到的光學影像，傳達至影像 分析/檢測裝置4〇中，影像分析/檢測裝置4〇藉由分析被 傳達的光學影像可以檢出偏光膜卷材中存在的各種各樣的 201209393 缺1^。另外，影像分析/檢測裝置40生成被檢出 M. ( LV 八I妇貝 下’稱作檢測結果資料），並將其儲存。 時基於上述檢測結果資料的缺陷產生位置，被傳 達到標記系統30 ’上述標記系、统3〇的標記控制部32，藉 由控制標記單元31在對應於上述缺陷產生位置的上述偏曰 光膜上標記為缺陷顯示。 另外，上述檢測結果資料，可以被傳送到輸出裝置5〇、 顯不裝置60等輸出該結果’並可以傳送到管理者的飼服器 70進行儲存以及管理。 过的缺1¾檢測結束之後，偏光膜可以由捲繞部（未 圖不）再次進行捲繞，以原來那樣的卷材的形態進行保管。 以下，根據第2圖以及第3圖，並參照第4圖〜第1〇 圖對本發明實施例的偏光膜卷材的品質判斷方法以及其 系統進行說明。 第2圖以及第3圖是用於說明本發明實施例的偏光膜 卷材的°π質判斷系統以及使用該系統的品質判斷方法的圖 面。 本發明的偏光膜卷材的品質判斷系統以及其方法，如 第2圖所示，大體分為三個階段來進行。具體來說，分為： 自動光學檢測機的檢測結果資料的轉換階段（S1 00)、缺 陷貝料的为析階段（S200 )以及卷材品質判斷階段（S300 )。 "即，本發明提供一種新的方法，該方法將從對偏光膜 卷材進行光學性缺陷檢測的自動光學檢測機所得到的檢測 結果資料作為其基礎，判斷需要檢測的偏光膜卷材（以下， 201209393 稱作檢測對象卷材）的品質。藉由本發明的偏光膜卷材的 品質判斷方法以及系統，可以進行更高可靠性的卷材品質 判斷’同時也可以得到構成不同卷材的品質水準的標準的 各種各樣的資訊，因此可以構築與每個卷材的品質相符的 後續步驟流程。 這可以通過之後的說明來更明確地理解。以下，參照 第3圖對上述的各階段（S100、S200、S300 )依次進行說 明。同時，以下說明的進行各階段的主體，既可以從上述 的自動光學檢測機分離而獨立地實現，也可以與自動光學 檢測機以一體結合的形態來實現。 另外’後述的卷材品質判斷方法的各階段，也可包含 在記錄了用於執行這些階段的命令語言的程式的電腦可讀 取的記錄媒體（例如，硬碟、光碟等）中而被提供，這對 於本領域普通技術人員而言是顯而易見的。 [檢測結果資料的轉換階段（s丨〇 〇 )] 自動光學檢測機的檢測結果資料，通常是以資料庫 (database)或者excei表格這樣的檔案格式來儲存。自 動光學檢測機，可以生成並儲存作為對每個卷材批次（L〇T ) 的檢測結果的缺陷的亮度、寬度、長度、大小（面積）、 形態（圓形性/線形性）、檢出光學群（投射/反射/偏光遮 斷）等各種各樣的數據。 此處，在上述檢測結果資料的資料傳送時，為了確保 大多數使用者或者管理者的密切的關係可以使用FTp( f i i e transfer protocol)。這樣被傳送的檢測結果資料可以儲 201209393 存到本發明的卷材品質判斷系統内的儲存部⑴l $ 了 將分析格式統-可以藉由資料轉換部12〇將其轉換為規定 的通用格式。此處，上述通用格式，是指數據語言的標記 眶等。因此’以後的資料分析’能夠以轉換了的皿等 進行。 但是，像上述那樣的檢測結果資料的轉換階段 (S100)，由於不同行的檢測機儲存資料的形式會得到不〆 樣，所以為了分析從不同機種的檢測機所得到的資料，需 要將其轉換為統—了的通用格式。與此不Θ，檢測機的儲 存資料的形式全部相同的時候，可以省略本階段。進而， 各個儲存資料的形式不同的時候也可以按照各個資料的形 式進行資科分析，因此在這樣的情況下當然也可以省略。 [缺陷資料的分析階段（S200 )] 缺陷資料的分析階段（S200 )，在各卷材批次（L〇T) 中，如第8圖所示，可以進行算出總共七個品質判斷標準 的工作。參照第8圖，上述的七個品質判斷標準包括三個 缺陷密度異常指數（總指數/未實施塗層指數/實施塗層指 數）、兩個亮點DPM (亮點DPM ( 0K )、亮點DPm ( NG ))、 重要缺陷數、總標記DPM。 但是’第8圖所示的品質判斷標準不是全部都是必需 的’报明顯既可以只使用這些的一部分，或者也可以進一 步使用包括這些在内的其他的品質判斷標準。 即，本實施例中’假定並說明了 ：先算出對全部上述201209393 VI. Description of the Invention: [Technical Field] The present invention relates to a quality judgment system for a polarizing film coil and a method thereof. More specifically, it relates to a method for determining a polarizing film using an inspection result data of an automatic optical inspection machine. The system of the quality of the coil and its method. [Prior Art] A manufacturer of a polarizing film used as an optical material for LCDs uses an inline N-LINE automatic optical inspection system (hereinafter referred to as automatic optical inspection for immediate detection of high-speed products). machine). Generally, an inline automatic optical inspection machine discards a marked portion in a subsequent step or additionally performs an additional detection on such a portion based on marking a color or marking a barcode at a defect generation position. However, 1) in the case where the above defects are continuously concentrated at a specific time point, 1 1) in the case where a fine defect or the like cannot be clearly judged whether or not it is a substantial defect, 'because of the above-mentioned automatic optical detector For load and H/W performance problems, problems can occur that cannot mark all of the defect locations. Those that are not marked with a defective position are identified as defective products. For this reason, in the subsequent step of final testing, such a product may be sent to the customer company in a possible step in the subsequent steps, and there is a need for the inspector to report the product again. 201209393 Fee and time issues. In addition, in the prior art, since only the defect generation rate information obtained by using only the number of defects detected in the coil material is obtained, there is no consideration of the defect temperature f B gas (ie, The question of the possibility of 'defects not being marked'. SUMMARY OF THE INVENTION [Problems to be Solved by the Invention] I 疋 于 于 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在 在Correctness and reliability. Another object of the present invention is to provide a quality judgment system and a method for obtaining quality information of a judgment quality by using an inspection result data of an automatic optical inspection machine and transmitting the important information to a subsequent step energy: To fulfil the machine as an early warning system for defective coils, another object of the present invention is to provide a quality judgment system and a method for automatically screening a coil which exhibits a high defect miss rate. [Technical Solution for Solving the Problem] According to an embodiment of the present invention, there is provided: a determination system for a polarizing film web, the system comprising: a storage portion that stores an automatic optical inspection machine for detecting defects of a polarizing film The defect data analysis unit obtains the defect yield=density of each unit area of the detection target coil based on the defect position information of 201209393 included in the detection result data, and changes the density according to the defect. The defect density of the above-mentioned detection target coil is calculated by the number of the second-order heterogeneous regions of the above-mentioned detection target, and the defect density abnormality index of the above-mentioned detection target coil is taken as a# △ object core 丨^ When the above is detected, the above-mentioned field image coil is judged as a defective coil. In the case of the surname U, the data conversion unit for converting the detection of the above-mentioned automatic optical inspection machine into a common format is also included, and the above-mentioned defect data is classified into (four) data analysis after conversion to the upper poem. In the example, the defect data analysis unit detects the bright spot defect information in the test result data into the original object coil; the BA 4 outputs the above-mentioned check/"missage number" by the above-mentioned test object roll. The number of bright spot defects of length and the length of the parent unit: The coil quality determining unit judges whether the coil is unqualified based on the calculated number of points free from defects, as early as... The bright spot defect of the above-mentioned mother soap length The calculation of the number can be performed separately by distinguishing between the upper-lighting point defects and the bright point defects judged to be (10) missing and judged as NG. In the embodiment, the above-mentioned detection result data is used to determine the defect pattern information detected by the inspection machine, the above-mentioned figure: width, length, shape, circularity, linearity information of t, and the above-mentioned defect data analysis section, according to the above 1, into a cow once "fixed at least the order of the 茱 出 出 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ At least one of a wrinkle of a film, a wrinkle of a PVA (polyethyl propylene glycol) film, a non-uniformity of a lower ruthenium film, a streaks of a coating layer, a streaks of a bonding layer, and a squeezing streak The coil quality determination unit may determine the detection target web as a defective coil when the number of appearance defects or the number of appearance defects per unit length is equal to or greater than a predetermined allowable value. The defect data analysis unit calculates a defect occurrence density of each of the first unit regions of the detection target coil and a defect generation density for each second unit region, where The unit area corresponds to each of the divided areas when the detection target roll is divided into the lining form of the NxM. The second unit area corresponds to each of the divided areas when the length of the detected image roll # is divided into 1 equal parts. The number of the defect density abnormal regions may be equivalent to the number of the first unit regions in which the defect occurrence density of each of the first unit regions is equal to or greater than the third allowable value, and the defect of each of the second unit regions is allowed to be four. The number of the second unit regions is equal to or greater than the value, and is added to obtain a value of 2. In the embodiment, the defect data analysis unit is a polarizing film roll in which a coating state is not applied and a polarizing film roll in which a coating state is applied. The material is subjected to calculation of the defect occurrence density in each of the unit regions, and the coil quality determination unit can determine the polarizing film web as a defective coil in any of the following t, that is, for the above-mentioned实201209393 The polarizing film coil of the coated state is characterized in that the defect density anomaly index of the uncoated coating is above the specified fifth allowable value. And the case where the defect density index of the coating layer is set to be equal to or greater than a predetermined sixth allowable value for the polarizing film coil in which the coating state is applied. In one embodiment, the defect data analysis unit does not implement the above-described defect data analysis unit. The defect density abnormality index of the coating layer gives a value of the first weighting value and a value of the second weighting value to the defect density abnormality index of the coating layer, and adds the density abnormal total index to calculate the quality of the coil. The determination unit may determine the polarizing film web as a defective coil when the density abnormal total index calculated above is equal to or greater than a predetermined seventh allowable value. Here, the first comparison with the density abnormal total index may be performed. The allowable value can be calculated according to the probability that the defect judged by the above automatic optical detector is not marked, that is, the defect missing rate. The product of the polarizing film coil provides a qualitative judgment according to other embodiments of the present invention. A, the I method is to use the detection result data of the automatic optical inspection machine for detecting the defect of the polarizing film coil to judge a method of measuring the quality of a target web, the method comprising: a step of generating a density of defects in each unit area of the detection target coil based on the defect position information included in the detection result data; and generating a density according to the defect When the number of the defect density abnormal regions equal to or greater than the predetermined value is calculated, the stage of the defect density abnormality index of the detection target coil is calculated, and when the defect density abnormality index of the detection target coil is equal to or greater than a predetermined allowable value, the detection target is The coil is judged to be a stage of failing the coil. 201209393 [Efficacy] According to the embodiment of the present invention, the correctness and reliability of the break of the polarizing film web can be improved. In addition, according to an embodiment of the present invention, important information for judging the quality of the web is obtained by using the detection result data of the automatic optical inspection machine. 1 This important information is transmitted to the subsequent steps, thereby fulfilling as a checker. The function of the early warning system can build different processes according to the quality of the coil in the subsequent steps, and can reduce the cost and time required for additional testing. > Also, according to an embodiment of the present invention, automatic screening can be performed for a web that is expected to have a defect defect rate. The present invention may be embodied in a variety of examples, and the specific embodiments are not limited to the specific concepts and technical scope of the present invention, and the alternatives are all in accordance with the present invention. When it is judged to be contrary to the detailed description of the present invention. In addition, the description (e.g., first, second, etc.) is an identification number for the constituent elements. The changes will be described in detail with various implementations. However, the embodiment is to be understood as all changes included in the equivalent of the alternative range. 'When the specific description of the related art is unclear, the numbers used in the description of the book are omitted. The one component is distinguished from the other 201209393. To help understand the present invention, the automatic optical inspection machine is first briefed. Description. Fig. 1 is a view showing an example of an automatic optical detector for detecting a defect of a polarizing film web. The polarizing film produced is stored in the form of a coil, and such a polarizing film coil is unwound from the unwinding portion (not shown) at the time of defect detection, and is conveyed by the roller 12 in a predetermined direction. At this time, the conveyance speed of the polarizing film 1 is read by a speed detecting device such as an encoder or the like, and is used for generation of subsequent defect position information. Here, the automatic optical detector includes an optical device 20 including at least one illumination device 21 and at least one imaging device 22 corresponding thereto, whereby an optical image of the polarizing film can be obtained during the conveyance. The configuration and design of the optical device 20 differ depending on the items (i.e., type and content) of the defect to be detected by the automatic optical detector, and the normal automatic optical detector transmits and reflects the polarizing film. Detection, polarized occlusion detection (also called cross-sectional (cr〇ss) detection), and the like. Therefore, the 'optical device 20 is designed to correspond to the item of the defect to be detected' by selectively or in parallel driving the reflected illumination (and the corresponding imaging device) and the transmitted illumination (and the corresponding imaging device) ), thereby obtaining an optical image for detecting various types of defects. Thus, the optical image obtained from the optical device 20 is transmitted to the image analysis/detection device 4, and the image analysis/detection device 4 can detect various types of the polarizing film coil by analyzing the transmitted optical image. Sample 201209393 is missing 1^. Further, the image analysis/detection device 40 generates a detected M. (hereinafter referred to as "detection result data") and stores it. The defect generation position based on the detection result data is transmitted to the mark control unit 32 of the mark system 30', and the above-mentioned defect light film corresponding to the defect generation position is controlled by the control mark unit 31. Marked as a defect display. Further, the above-mentioned detection result data can be transmitted to the output device 5, the display device 60, etc., and the result can be transmitted to the manager's feeder 70 for storage and management. After the completion of the detection, the polarizing film can be wound again by the winding portion (not shown) and stored in the form of the original coil. Hereinafter, a method for judging the quality of a polarizing film web according to an embodiment of the present invention and a system thereof will be described with reference to Figs. 2 and 3 and Figs. 4 to 1 . Fig. 2 and Fig. 3 are views for explaining the ?π quality judgment system of the polarizing film roll of the embodiment of the present invention and the quality judgment method using the same. The quality judging system of the polarizing film web of the present invention and the method thereof are roughly divided into three stages as shown in Fig. 2 . Specifically, it is divided into: a conversion phase (S1 00) of the detection result data of the automatic optical inspection machine, a precipitation stage (S200) of the defective bedding material, and a coil quality determination stage (S300). " That is, the present invention provides a new method for judging a polarizing film web to be detected from the detection result data obtained by an automatic optical inspection machine for optical defect detection of a polarizing film web as a basis ( Hereinafter, 201209393 is referred to as the quality of the detection target coil. According to the method and system for determining the quality of the polarizing film coil of the present invention, it is possible to determine the quality of the coil material with higher reliability, and at the same time, it is possible to obtain various kinds of information constituting the standard of the quality level of the different coil materials, and thus it is possible to construct A subsequent step process that matches the quality of each coil. This can be understood more clearly by the explanations that follow. Hereinafter, each of the above-described stages (S100, S200, and S300) will be sequentially described with reference to Fig. 3. Meanwhile, the main body for performing each stage described below may be realized separately from the above-described automatic optical detector, or may be realized integrally with the automatic optical detector. Further, each stage of the method for judging the web quality described later may be provided in a computer-readable recording medium (for example, a hard disk, a compact disc, or the like) in which a program for executing the command language of these stages is recorded. This will be apparent to those of ordinary skill in the art. [Conversion phase of test result data (s丨〇 〇 )] The test result data of the automatic optical inspection machine is usually stored in a file format such as a database or an excei form. The automatic optical inspection machine can generate and store the brightness, width, length, size (area), shape (circularity/linearity) of defects as a result of detection of each coil batch (L〇T), and inspection Various data such as optical group (projection/reflection/polarization interception) are available. Here, in the data transmission of the above-mentioned detection result data, an FTp (f i i e transfer protocol) can be used in order to ensure a close relationship between most users or managers. The test result data thus transmitted can be stored in the storage unit (1) 1 $ in the coil quality judging system of the present invention. The analysis format can be converted into a prescribed general format by the data conversion unit 12 . Here, the above-mentioned general format refers to a mark of the data language, and the like. Therefore, the 'subsequent analysis of data' can be carried out with a converted dish or the like. However, in the conversion phase (S100) of the detection result data as described above, since the form of the data stored by the detectors of different lines is not obtained, in order to analyze the data obtained from the detectors of different types of machines, it is necessary to convert them. The general format for the system. In this case, when the form of the stored data of the detector is all the same, this stage can be omitted. Further, when the form of each stored material is different, the analysis of the subject can be performed in accordance with the form of each data. Therefore, in such a case, it is of course possible to omit it. [Analysis stage of defect data (S200)] In the analysis stage (S200) of the defect data, in each of the coil lots (L〇T), as shown in Fig. 8, the work of calculating a total of seven quality determination standards can be performed. . Referring to Figure 8, the above seven quality judgment criteria include three defect density anomaly indices (total index/unimplemented coating index/implementation coating index), two bright spots DPM (bright spot DPM (0K), bright spot DPm (NG) )), the number of important defects, the total mark DPM. However, the quality judgment criteria shown in Fig. 8 are not all necessary. It is obvious that only a part of these may be used, or other quality judgment standards including these may be further used. That is, in the present embodiment, 'assumed and explained: first calculate all of the above
S 12 201209393 七個品質判斷標準的分析結 中的任意一個顯示為規定的 斷為不合格卷材。但是，很 中的至少一個進行分析，從 果（數值），上述七個標準之 允許值以上的值的卷材，被判 明顯也可以對上述七個標準之 而判斷為是不合格卷材。 以下，對上述七個品質判斷標準按順序進行說明。本 又所進仃的缺陷資科的分析，冑由缺陷資料分析部 進行。 [算出缺陷密度異常指數] 如第4圖所示，自動光學檢測機對各個卷材批次，記 錄相對於卷材的寬度/長度的標記位置。本說明書中的卷材 的長度，定義為藉由自動光學檢測機將卷材搬運的前進方 向所對應的長度；卷材的寬度，定義為與該前進方向垂直 相交的方向所對應的長度。 本發明的卷材品質判斷系統中的缺陷資料分析部 130 ’從自動光學檢測機的檢測結果資料中，得到如第4圖 所示的缺陷標記圖（缺陷位置資訊），，（S21 〇 )，對其 加以分析並算出卷材的各個區間的缺陷產生密度。 按照卷材的各個區間的缺陷產生密度的算出，是將卷 ㈣J分成規定的各個單位區域，#出被這樣劃分了的每個 單位區域中的缺陷產生密度（S212)。參照第5圖對其進 行說明的話，如下所示。 “各個區域的缺陷產生密度”，是如第5圖所示，算 出Μ/D方向（卷材的長度方向）和T/D方向（卷材的寬度 方向）設定的各個單位區域的該數值，該密度算出方法， 13 201209393 可以利用下述的式（l)算出。 MDi=MCi/(WmxHm). 丁 Di = TCi / (WfxHf)---( 1 ) 這晨，MCi是MDi區域的總缺陷數，tc.是 區域的總缺陷數。 即’Μ/D方向中的各個單位區域的缺陷產生密度（即， MDi)，利用將檢測對象卷材劃分為ΝχΜ的行列形1而被 分割的各個單位區域中存在的總缺陷数（即，Μ 〇 )S 12 201209393 One of the analysis levels of the seven quality judgment standards is displayed as a specified defective material. However, at least one of the analysis is carried out, and it is judged that the coil of the value of the above-mentioned seven standards or more is judged to be a defective coil. Hereinafter, the above seven quality determination criteria will be described in order. The analysis of the defects of this subject is carried out by the Defective Data Analysis Department. [Calculation of defect density abnormality index] As shown in Fig. 4, the automatic optical inspection machine records the mark position with respect to the width/length of the web for each of the coil batches. The length of the web in the present specification is defined as the length corresponding to the advancing direction of the web conveyance by the automatic optical inspection machine; the width of the web is defined as the length corresponding to the direction perpendicular to the advancing direction. The defect data analysis unit 130' in the coil quality judging system of the present invention obtains the defect flag map (defect position information) as shown in Fig. 4 from the detection result data of the automatic optical detector, (S21 〇), It was analyzed and the density of defects generated in each section of the web was calculated. In the calculation of the defect occurrence density in each section of the web, the volume (four) J is divided into predetermined unit areas, and the defect generation density in each unit area thus divided is obtained (S212). When it is explained with reference to Fig. 5, it is as follows. "The density of defects in each area" is the value of each unit area set in the Μ/D direction (longitudinal direction of the coil) and the T/D direction (width direction of the coil) as shown in Fig. 5, This density calculation method, 13 201209393, can be calculated by the following formula (1). MDi=MCi/(WmxHm). Di Di TCi / (WfxHf)---(1) This morning, MCi is the total number of defects in the MDi region, and tc. is the total number of defects in the region. That is, the defect generation density (i.e., MDi) of each unit region in the 'Μ/D direction, the total number of defects existing in each unit region divided by dividing the detection target coil into the matrix shape 1 of ΝχΜ (ie, Μ 〇 )
來算出。並且，T/DTo calculate. And, T/D
以該單位區域的面積（即，WmXH 方向中的各個單位區域的缺陷產生密度（即，TD 用將檢測對象卷材的長度分成L等份而被分㈣各個單位 區域中存在的總缺陷数（即，TCi)，除以該單位區域的 面積（即，W f xH f )來算出。 此處，定義的卷材中的m/d方向的單位區域、T/D方 向的單位區域的Wm ' Hm、H f作為設定參數，在設計上 也可以可變地構成為具有適#的值。並且，Wf作為卷材的 寬度而被設定。 之後，缺陷資料分析部130，如上述的那樣，根據算 出的各個單位區域的缺陷產生密度來算出缺陷密度異常區 域數（S214)。 缺陷密度異常區域數”，是對於算出的缺陷產生密 度作為超過由上述卷材品質判斷系統預先設定的基準值 的區域的總數而被算出。第6圖是算出相對各個單位區域 的Μ/D方向的缺陷產生密度以及T/D方向的缺陷產生密度 201209393 的例子的圓面，The area of the unit area (i.e., the defect density of each unit area in the WmXH direction is generated (i.e., the TD is divided into L equal parts by dividing the length of the detection target web into four equal parts) (4) the total number of defects existing in each unit area ( That is, TCi) is calculated by dividing the area of the unit area (that is, W f xH f ). Here, the unit area of the m/d direction and the Wm ' of the unit area of the T/D direction in the defined coil. Hm and Hf are set as parameters, and can be variably designed to have a value of #. Further, Wf is set as the width of the coil. Then, the defect data analysis unit 130 is as described above. The calculated defect density occurrence area of each unit area is calculated (S214). The defect density abnormal area number is an area where the calculated defect occurrence density is a reference value that is set in advance by the above-described coil quality determination system. The sixth figure is a circular surface for calculating an example of the defect occurrence density in the Μ/D direction and the defect generation density 201209393 in the T/D direction with respect to each unit region.
在Τ/D方向中為1 1個，合計是5個。 各個單位區域内， 陷密集度的值。抒 即，上述算出的缺陷密度異常區域數，是在卷材内 ，缺陷密集產生的程度，即數值化表現缺 所謂缺陷密集度較高，意味著缺陷在特定 時間點（即，卷材内的特定區間）集中且連續地產生。 因此，缺陷在特定區間集中且連續地產生的時候，因 為自動光學檢測機的缺陷檢出步驟產生過負載，所以由於 系統的性能問題導致不能對所有的缺陷都進行標記的發生 概率變高。缺陷密度異常區域數的算出，在判斷上述那樣 的缺陷遺漏的可能性（即，存在NG缺陷，而未被標記的可 月t*性）時被有效利用。即，將缺陷密集度數值化了的密度 值，由於與缺陷遺漏率有密切的關係，所以構成為卷材品 質判斷的主要項目。 如上所述，算出缺陷密度異常區域數之後，缺陷資料 分析部130，將算出的缺陷密度異常區域數按照預先定義 的方式’算出指數化表現了的缺陷密度異常指數（S216)。 “缺陷密度異常指數”如第8圖所示，分別算出“總 指數”、“未實施塗層的指數”、“實施塗層的指數”， 這是為了將“缺陷密度異常區域數”區分為未實施塗層檢 測機、實施塗層檢測機的各個以及將它們综合了的總指 數’之後在使用了 EW ( Ear ly Warning )指數的不合格卷 15 201209393 材判斷階段來利用。 ‘‘為了區分並算出這樣的“未實施塗層的指數，’以及 . 貫施塗層的指數”，本發明的卷材品質判斷系統中的缺 陷資料分析部13°，可以從上述的“缺陷產生密度”的算 出階段來區分未實施塗層狀態的偏光膜卷材以及實施^ 狀態的偏光膜卷材的情況，並進行上述密度算出工作/ 此處，上述未實施塗層的指數”以及“實施塗層的 指數，例如可以利用下述的式（2)來算出。 未實施塗層的指數=(A/c) xl〇〇〇 實施塗層的指數=(B/C) xlOOO ——（2) 此處，A是相對于未實施塗層狀態的偏光膜卷材的缺 陷密度異常區域數，B是相對于實施塗層狀態的偏光膜卷 材的缺陷密度##區域數，CS表示檢測對象的偏光膜卷 材的長度。另外，在上述的式（2)中，在算出各個指數時 所乘的1 000，是為了算出每個1000m的缺陷產生密度。 另外，上述的“總指數”是例如如下述的式（3)所示 的，可以在分別從未實施塗層檢測機和實施塗層檢測機算 出的資料（此處，是指上述“未實施塗層的指數，，以及“實 施塗層的指數”）中，藉由分別賦予-規定的加權值來算出。 ' 總指數=未實施塗層的指數xai +實施塗層的指數xa2 • =[(Axai+Bxa2)/C] xl〇〇〇 ——（3) 此處’ ai意味著對未實施塗層的指數賦予的加權值， aa意味著對實施塗層的指數賦予的加權值，ai以及a2可以 設定為大於0小於等於1。 16 201209393 如上所述’為了算出總指數將未實施塗層的指數以及 實施塗層的指數進行區分’並分別賦予規定的加權值的原 因’疋未實施塗層狀態的偏光膜卷材中存在的缺陷密度、 和實施塗層狀態的偏光膜卷材中存在的缺陷密度，在之後 的不合格卷材的判斷時’可以互相作為不同的重要度來利 用。因此，可以使上述的各個加權值對應於偏光膜的製作 步驟的特性而設定為適當的值。 [亮點DPM的算出] 冗點缺陷疋缺陷的大小為較小的細微缺陷，由於難以 通過目視檢測來辨識，是不合格檢出時極為困難的缺陷類 型。因此，本發明的卷材品質判斷系統，按照不同缺陷種 類的DPM將亮點DPM另行進行分類，將其使用於不合格卷 材判斷中。另外’上述亮點DPM，向不合格卷材判斷之後 的後續步驟流程傳達，在後續步驟作為有用的資訊使用。 上述DPM (Defect per meter)是將卷材内的缺陷產 生頻率數值化的值，定義為將全部缺陷的產生數除以卷材 長度的值。 上述焭點DPM的算出中必要的卷材内的總亮點缺陷 數，可以從自動光學檢測機的檢測結果資料中包含的亮點 缺陷（bright defect)資訊中得到（S220 )。自動光學檢 測機的檢測結果資料，也可以包含藉由偏光遮斷檢測（交 叉檢測）檢出的亮點缺陷資訊。根據偏光遮斷檢測，所有 的光（包括偏光）都變得不透過偏光膜，但是由於偏光膜 的缺陷，一部分的偏光透過偏光膜的一部分的區域（位置） 17 201209393 產生亮點缺陷，這樣的亮點缺陷資訊就被記錄在上述檢測 結果資料中。In the Τ/D direction, it is 1 1 and the total is 5. The value of the concentration in each unit area. That is, the number of defect density abnormal regions calculated as described above is the extent to which the defects are densely generated in the coil, that is, the numerical representation lacks the so-called defect density, which means that the defects are at a specific time point (ie, within the coil). Specific intervals are generated collectively and continuously. Therefore, when defects are concentrated and continuously generated in a specific section, since the defect detection step of the automatic optical detector generates an overload, the probability that the failure of all the defects cannot be marked due to the performance problem of the system becomes high. The calculation of the number of defect density abnormal regions is effectively utilized when judging the possibility of missing defects as described above (i.e., there is an NG defect and the flag is not marked). In other words, the density value that quantifies the defect density is closely related to the defect miss rate, and is therefore the main item for determining the quality of the coil. After the number of the defect density abnormal regions is calculated as described above, the defect data analysis unit 130 calculates the defect density abnormality index which is exponentially expressed in accordance with the predetermined number of defective density abnormal regions (S216). As shown in Fig. 8, the "defect density anomaly index" is calculated as "total index", "index of uncoated coating", and "index of applied coating", respectively, in order to distinguish "number of defect density regions" into The coating index detector, the respective coating inspection machines, and the total index of the combination of the coating detectors were used in the judgment stage of the unqualified volume 15 201209393 using the EW (early warning) index. ''In order to distinguish and calculate such an index of the unimplemented coating, 'and the index of the applied coating layer', the defect data analysis unit 13° in the coil quality judging system of the present invention may be from the above-mentioned "defect In the calculation stage of the "density", the polarizing film coil in which the coating state is not applied and the polarizing film coil in the state of the coating are separated, and the above-described density calculation operation/here, the index of the unimplemented coating is described, and " The index of the coating layer can be calculated, for example, by the following formula (2). Index of unimplemented coating = (A / c) xl 指数 index of applied coating = (B / C) x lOOO - (2) Here, A is a polarizing film coil relative to the uncoated state The number of defective density abnormal regions, B is the number of defect density ## regions of the polarizing film web in which the coating state is applied, and CS indicates the length of the polarizing film web to be detected. Further, in the above formula (2), 1 000 multiplied in the calculation of each index is for calculating the density of defects per 1000 m. In addition, the "total index" described above is, for example, as shown in the following formula (3), and data which can be calculated from the coating tester and the coating inspection machine, respectively (here, the above-mentioned "not implemented" The index of the coating, and the "index of the applied coating") are calculated by assigning a prescribed weighting value respectively. 'Total index = index of the unimplemented coating xai + index of the applied coating xa2 • =[ (Axai+Bxa2)/C] xl〇〇〇——(3) where 'ai means the weight given to the index of the uncoated coating, aa means the weight given to the index of the applied coating, ai And a2 can be set to be greater than 0 and less than or equal to 1. 16 201209393 As described above, the reason for distinguishing the index of the uncoated coating and the index of the applied coating to calculate the total index and assigning the predetermined weighting value respectively is not implemented. The defect density existing in the polarizing film roll in the coated state and the defect density existing in the polarizing film roll in the state of the coating state can be utilized as different importance levels in the subsequent judgment of the defective coil material. Therefore, Each of the weighting values described above is set to an appropriate value in accordance with the characteristics of the manufacturing process of the polarizing film. [Determination of the bright spot DPM] The size of the redundant defect 为 defect is a small fine defect, and it is difficult to recognize by visual inspection. It is a type of defect that is extremely difficult to detect when it is unqualified. Therefore, the coil quality judging system of the present invention separately classifies the bright spot DPM according to the DPM of different defect types, and uses it in the judgment of the defective coil. The highlight DPM is conveyed to the subsequent step flow after judging the defective coil, and is used as useful information in the subsequent step. The above DPM (Defect per meter) is a value that quantifies the frequency of defect generation in the coil, and is defined as all The number of defects generated is divided by the value of the length of the coil. The total number of bright spot defects in the coil required for the calculation of the defect DPM can be from the defect information contained in the test result data of the automatic optical detector. Obtained (S220). The detection result data of the automatic optical inspection machine may also include detection by polarization occlusion (cross-checking) The bright spot defect information detected. According to the polarization blocking detection, all the light (including the polarized light) becomes non-transmissive to the polarizing film, but a part of the polarized light passes through a part (position) of the polarizing film due to the defect of the polarizing film. 201209393 A highlight defect is generated, and such highlight defect information is recorded in the above test result data.
因此，本發明的卷材品質判斷系統中的缺陷資料分析 部130,可以基於上述檢測結果資料中包含的亮點缺陷資 訊來算出卷材内的亮點缺陷數’這樣，就可以算出亮點DM (S222 )。在本發明的實施例中，上述亮點DpM可以利用 下述的式（4)算出。 檢測長度 亮點 DPM(OK) 亮麵妒 ----(4 ) 此處，上述Si是藉由自動光學檢測機被檢出（被監 視），但是由於缺陷程度較低而沒有被判斷為進行標記的 亮點缺陷’即，意味著是被判斷為οκ的亮點缺陷，是被 判斷了進行標記的亮點缺陷’即，意味著是被判斷為阢的 免點缺陷β 上述的被判斷為0Κ的亮點缺陷（Si)，是在各個單一 圖70的情況下難以判斷為缺陷的細微的缺陷，但是這些細 微的缺陷如果個數變多，或者在一處集中分佈的話，就可 以作為不合格被判斷的細微的缺陷，所以本實施例中對被 判斷為NG的亮點缺陷的亮點_ (即，參照上述亮點_ (NG))與亮點DPM(〇K) 一起也被另行算出。因此，本 發明的卷材品質判斷系、統，考慮到亮點缺陷的重㈣㈣ 201209393 同時考慮亮點DPM ( NG )以及亮點dpm ( OK )。 [重要缺陷數的算出] 本發明貫施例的卷材品質判斷系統，除了上述的“亮 缺陷 還自b夠以選擇其他各種各樣的重要缺陷類型並 算出的方式來構成。 因此，上述重要缺陷數，意味著本發明的系統中被定 義的特定缺陷類型的產生數，在重要缺陷中，有例如“ TAC 膜，的皺褶”、“㈣膜的敵褶’’、“下部道膜的不均 勻” AS塗層的條紋” 、“NCFT/D的條紋（黏結層的條 紋）”、“擠壓性的條紋”等。 述的PVA膜’通常作為顯示偏光特性的偏光元件來 被利用，構成偏光膜的核心層，上述TAC膜用於支撐以及 保護上述PVA膜’被疊層在上述m膜核心層的兩面，黏 層用於在之後將偏光膜附著在液晶面板上並疊層在上述 TAC膜上。另夕卜’上述AS塗層’意味著偏光膜的靜電防止 用塗層’擠壓性的條紋，意味著在自動光學檢測時由輥擠 壓產生的條紋。 、上述的這樣六種缺陷類型，是辨識率較低，在特定區 域的產生頻率較高，藉由自動光學檢測機不能將全部的產 生缺陷都進行標記的危險性較高的缺陷群的代表性類型。 :此’本發明的卷材品f判斷純，自動識別上述這樣的 重要缺陷的產生件數並同時使用於卷材品質判斷。但是， 陷只不過是—個例子’與上述示例進行不同的 選擇明顯也是可以的。 19 201209393 上述這樣的重要缺陷數’可以根據自動光學檢測機的 檢測結果資料中包含的缺陷圖案資訊來算出（s23q)。上 述檢測結果資料中’關於檢出的缺陷的缺陷產生形態，可 以包含上述缺陷的大小、寬度、長度、面積、圓形性/線形 性等的圖案資訊。因此，本發明的卷材品質判斷系統的缺 陷資料分析部13〇，藉由使用上述缺陷圖案資料將外觀上 的缺陷的特徵轉換為數值化了的特徵，就可以辨識上述示 例這樣的重要缺陷（S232 )。 例如，TAC膜的皺褶”的時候，如第7A圖所示，通 常來說該缺陷形態是圓形性的可能性較低，缺陷的長度/ 寬度的比率較高，，與缺陷的寬度相比具有長度較大的 :觀上的特徵。關於這-點，參照第7B圖以及第7C圖， 當缺陷的長度除以缺陷的寬度所得的比率為2以上、缺陷 面積除以將缺陷的最長的長度作為直徑的假定的圓的面積 所得的比率為A以下、缺陷面積為B以上的缺陷的時候， 將其判斷為上述的“ TAC膜的皺褶”，便是那種例子。 如上所述，本發明的卷材品質判斷系統中的缺陷資料 刀析13 0，根據母個缺陷所顯現的外觀上的特徵，藉由 對缺的特徵性資料的組合將各個重要缺陷形態數值化並 、行辨識此時，每個重要缺陷中顯現的特徵性的圖案資 訊（缺陷類型的辨識要素），可以藉由分析自動光學檢測 機的過去檢測累積資料來得到。除此以外’作為搜索每個 缺陷類型的辨識要素的方法當然也可以使用各種各樣的統 计方法。 20 201209393 有關的示例那樣，“重要 陷類型設定各種各樣的辨 陷作為重要缺陷來識別。 如與所述TAC膜的鈹;|習，， 缺陷數”對於預先特定的特定缺 識要素’將符合該設定條件的缺 的時候一樣，重要缺陷辨識時 缺陷和被判斷為NG的缺陷。這 此時，與所述“亮點缺陷” 也同時考察被判斷為〇Κ的 是因為，可以將從標記判斷中落選的、辨識率較低的缺陷 產生資訊提供給後續步驟’作為非常有用的資訊進行利 用此時’上述重要缺陷數，可以利用卿來表現。 [總標記D Ρ Μ的算出] “總標記DPM”從自動光學檢測機作為被缺陷標記了 的DPM被算出。總標記DpM，，的算出（似2 )可以利用 下述的式（5 )求出。 檢測長度Therefore, the defect data analysis unit 130 in the coil quality determination system of the present invention can calculate the number of bright spot defects in the coil based on the bright spot defect information included in the detection result data, and can calculate the bright spot DM (S222). . In the embodiment of the present invention, the above-described bright spot DpM can be calculated by the following formula (4). Detection length bright spot DPM (OK) Bright surface 妒----(4) Here, the above Si is detected (monitored) by an automatic optical detector, but is not judged to be marked because of a low degree of defect The bright spot defect 'is, that is, the bright spot defect judged as οκ, is the bright spot defect judged to be marked', that is, the point-free defect β which is judged to be 阢, the above-mentioned bright spot defect judged to be 0 Κ (Si) is a subtle defect that is difficult to judge as a defect in the case of each single graph 70. However, if these subtle defects are increased in number or concentrated in one place, they can be judged as subtle defects. In the present embodiment, the bright spot _ (that is, referring to the above-mentioned bright spot _ (NG)) and the bright spot DPM (〇K) of the bright spot defect judged as NG are also separately calculated. Therefore, the quality judgment system of the coil of the present invention takes into account the weight of the bright spot defect (4) (4) 201209393 Considering both the bright spot DPM (NG) and the bright spot dpm (OK). [Calculation of the number of important defects] The coil quality determination system according to the present embodiment of the present invention is configured such that the above-mentioned "bright defects are sufficiently calculated by selecting various other important defect types. The number of defects means the number of generations of specific defect types defined in the system of the present invention, among which are, for example, "TAC film, wrinkles", "(4) film entanglement", "lower film Non-uniform "streak of AS coating", "stripes of NCFT/D (stripes of bonding layer)", "squeezing stripes", etc. The PVA film 'described is generally used as a polarizing element that exhibits polarization characteristics. a core layer constituting the polarizing film, the TAC film is used for supporting and protecting the PVA film 'on both sides of the m film core layer, and the adhesive layer is used for attaching the polarizing film to the liquid crystal panel and laminating the above On the TAC film, the 'AS coating as described above' means the stripe of the static electricity prevention coating of the polarizing film, which means the stripe produced by the roll extrusion during the automatic optical inspection. Species The type of defect is a representative type of defect group with a low recognition rate, a high frequency of occurrence in a specific area, and a high risk of not being able to mark all the defects generated by the automatic optical detector. The coiled material f of the invention is judged to be pure, and the number of the above-mentioned important defects is automatically recognized and used for the quality judgment of the coil. However, the trapping is merely an example. It is also possible to make a different selection from the above example. 19 201209393 The above-mentioned number of important defects ' can be calculated based on the defect pattern information included in the detection result data of the automatic optical detector (s23q). In the above-mentioned detection result data, the defect generation form of the detected defect may include The pattern information of the size, the width, the length, the area, the circularity, the linearity, and the like of the above-described defects. Therefore, the defect data analysis unit 13 of the coil quality judging system of the present invention has an appearance by using the above-described defect pattern data. The characteristics of the defect are converted into numerical features, and the important defects such as the above examples can be identified ( S232). For example, when the TAC film is wrinkled, as shown in Fig. 7A, generally, the defect form is less likely to be circular, the length/width ratio of the defect is higher, and the defect The width has a larger length than the view: the feature. With regard to this point, referring to FIG. 7B and FIG. 7C, the ratio obtained by dividing the length of the defect by the width of the defect is 2 or more, and the area of the defect is divided by the area of the assumed circle having the longest length of the defect as the diameter. When the ratio is A or less and the defect area is B or more, it is judged as the above-mentioned "wrinkle of the TAC film". As described above, the defect data analysis method in the coil quality judging system of the present invention numerically digitizes each important defect form by combining the characteristic data of the defect based on the appearance characteristics exhibited by the parent defect. At this time, the characteristic pattern information (identification element of the defect type) appearing in each important defect can be obtained by analyzing the accumulated detection data of the automatic optical inspection machine in the past. In addition to this, as a method of searching for the identification elements of each defect type, it is of course possible to use various statistical methods. 20 201209393 As with the related examples, "the important trap type sets various discriminants as important defects to identify. As with the TAC film, the number of defects" is for the specific specific defect element that is specified in advance. The same is true for the absence of the set condition, the defect is identified when the important defect is identified, and the defect is judged to be NG. At this time, the "bright spot defect" is also examined at the same time as being judged as 〇Κ because the defect generation information which is selected from the mark judgment and has a low recognition rate can be provided to the subsequent step as a very useful information. By using this time, the number of important defects mentioned above can be expressed by Qing. [Calculation of total mark D Ρ ]] "Total mark DPM" is calculated from the automatic optical detector as DPM marked with a defect. The calculation of the total mark DpM, (2) can be obtained by the following formula (5). Detection length
總標記DPM -----(5) 此處，缺陷標記（Mi)數，可以根據自動光學檢測機 的檢測結果資料中包含的缺陷標記資訊算出（S240 )。上 述總標記DPM是表示卷材内的缺陷產生頻率的值，作為之 後的品質判斷的重要項目來使用。 [卷材品質判斷階段（S300 )] 在本階段’使用上述的缺陷資料的分析階段（S2〇〇 ) 中算出的“缺陷密度異常指數，，（總指數/未實施塗層指數 /實施塗層指數）、“亮點DPMC0K/NG)，， ' 重要缺陷 數以及總標記DPM”的共計七個資訊來辨別不合格卷 21 201209393 材。這可以由本發明的系統中的卷材品質判斷部140來進 行。 如第8圖所示’即使當用於卷材品質判斷的七個標準 • 之中的任意一個項目超過規定的允許值（EW基準，參照u '〜k7 )時，卷材品質判斷部140也會判斷相應的卷材為不 合格卷材（S310、S320、S33〇、S34〇)。上述的Ew基準， 可以設定為在量產過程中不合格卷材篩選的正確性是被驗 證了的適當的值。 第9圖示出了各個卷材L批次的缺陷遺漏率（即， 缺陷從標記中落選的比率）以及將EW指數中的“總指數” 標記了的例子。卷材品質判斷部14〇，例如為了確保缺陷 遺漏率在0.3%以下的品質水準，根據上述的第9圖設定 “EW基準（總指數）” ^ki (此處，ki是正數），從而將 超過上述允許值的卷材判斷為不合格卷材。 因此，為了不合格卷材判斷的正確性，藉由對不同卷 材的缺陷遺漏率和各個別指數的標記關係進行統計性相 關關係的累積統計工作’有必要設定適當# “ Ew基準”。 藉由上述這樣的方法，—旦適當的“ EW基準，，被設 定，則藉由使用被設定的“EW基準’’（參照第9圖以及第 -1〇圖），可以篩選超過該基準（允許值）的不合格卷材和 •沒超過該基準的卷材。# ’根據本發明的卷材品質判斷系 統以及方法，藉由對不合格卷材進行早期預警（hHy Warni ng )，對預想為缺陷遺漏率較高的卷材批次在後續步 驟中預先確認，可以設置能夠防止該卷材的缺陷遺漏的別 22 201209393 的步驟。另外，對於沒有作為不合格卷材而被篩選的卷材 批次’在後續步驟也可以不使用另外的目視檢測，或者可 以簡化檢測流程，因此具有可以降低追加檢測的時間和費 用的優點。在第9圖中示出了，在總共八個卷材批次中， 對於EW指數L以内的六個卷材，即使在後續步驟中不進 订另外的檢測也可以確保遺漏率在〇. 3%以下的品質水準。 以上’雖然已經參照本發明的實施例對本發明進行了 說明，但對於本領域技術人員應該可以理解，在不脫本發 明所記載的構思和範圍的情況下，可以對本發明進行多樣 的修改以及變更。 【圖式簡單說明】 第1圖是對偏光膜卷材進行缺陷檢測的自動光學檢測 機的一個例子的圖面。 第2圖是用於說明本發明實施例的偏光膜卷材的品質 判斷系統以及使用該系統的品質判斷方法的圖面。 第3圖是用於說明本發明實施例的偏光膜卷材的品質 判斷系統以及使用該系統的品質判斷方法的圖面。 第4圖是對偏光膜卷材進行缺陷標記的圖面。 第5圖是用於說明由缺陷資料分析部對缺陷密度異常 區域數的計算方法的圖面。 第6圖是用於說明由缺陷資料分析部對缺陷密度異常 區域數的計算方法的圖面。 23 201209393 第7A圖是舉例表示TAC膜的褶皺的圖面。 第7B圖是用於說明TAC膜的褶皺中的缺陷的特徵的圖 面。 第7C圖是用於說明TAC膜的褶皺中的缺陷的特徵的圖 面。 第8圖是用於本發明的一個實施例的卷材品質判斷的 七個標準的圖面。 第9圖是用於說明缺陷密度里當扣 „ 也又〇 τ扎數和缺陷遺漏率之 間的相互關係的圖面。 第10圖是根據缺陷密度里當沪 特判斷方㈣圖面。 ”^數用於說明不合格卷 【主要元件符號說明】 110儲存部 120資料轉換部 130缺陷資料分析部 14 0卷材品質判斷部 24Total mark DPM -----(5) Here, the number of defect marks (Mi) can be calculated based on the defect mark information included in the test result data of the automatic optical detector (S240). The above-mentioned total mark DPM is a value indicating the frequency of occurrence of defects in the coil, and is used as an important item for quality judgment thereafter. [Coil quality judgment stage (S300)] In this stage 'defect density abnormality index calculated using the above-mentioned defect data analysis stage (S2〇〇), (total index/unimplemented coating index/implementation coating) A total of seven pieces of information on the index, "bright spot DPMC0K/NG", 'the number of important defects and the total mark DPM" are used to identify the defective roll 21 201209393. This can be performed by the web quality judging section 140 in the system of the present invention. As shown in Fig. 8, when any one of the seven standards for the quality judgment of the coil exceeds the predetermined allowable value (EW reference, refer to u ' to k7 ), the coil quality determining unit 140 It will also judge that the corresponding coil is a defective coil (S310, S320, S33〇, S34〇). The above Ew benchmark can be set to verify that the correctness of the defective coil screening during mass production is verified. Appropriate values. Figure 9 shows the defect miss rate for each of the L-batch batches (ie, the ratio of defects missing from the mark) and the "total index" in the EW index. Department 14〇, for example To ensure the quality level of the defect omission rate below 0.3%, set the "EW standard (total index)" ^ki (here, ki is a positive number) according to the above-mentioned figure 9, so that the coil exceeding the above allowable value is judged as not Qualified coils. Therefore, in order to judge the correctness of unqualified coils, it is necessary to set the appropriate statistical work by statistically correlating the defect omission rate of different coils and the marking relationship of individual indices. "Based". By the above method, if the appropriate "EW standard" is set, it can be filtered by using the set "EW reference" (refer to Figure 9 and Figure -1). The reference (allowed value) of the defective coil and the coil that does not exceed the reference. # 'The coil quality judgment system and method according to the present invention, by early warning of the defective coil (hHy Warni ng) For the coil batch that is expected to have a high defect omission rate, it is pre-confirmed in the subsequent step, and the step of 2012 09393, which can prevent the defect of the coil from being defective, can be set. The coil batch that is screened for the defective coil may not use another visual inspection in the subsequent steps, or may simplify the inspection process, and thus has the advantage of reducing the time and cost of additional detection. In the total of eight coil batches, for the six coils within the EW index L, even if no additional inspections are made in the subsequent steps, the quality level of the missing rate below 3% can be ensured. The present invention has been described with reference to the embodiments of the present invention, but it should be understood by those skilled in the art that various modifications and changes can be made without departing from the spirit and scope of the invention. . BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a view showing an example of an automatic optical detector for detecting a defect of a polarizing film web. Fig. 2 is a view for explaining the quality judging system of the polarizing film web and the quality judging method using the same according to the embodiment of the present invention. Fig. 3 is a view for explaining the quality judging system of the polarizing film web and the quality judging method using the same according to the embodiment of the present invention. Fig. 4 is a view showing a defect mark on a polarizing film web. Fig. 5 is a view for explaining a method of calculating the number of defect density abnormal regions by the defect data analysis unit. Fig. 6 is a view for explaining a method of calculating the number of defect density abnormal regions by the defect data analysis unit. 23 201209393 Fig. 7A is a view showing a pleat of a TAC film. Fig. 7B is a view for explaining characteristics of defects in wrinkles of the TAC film. Fig. 7C is a view for explaining characteristics of defects in wrinkles of the TAC film. Fig. 8 is a view showing seven standards for the quality judgment of the coil used in one embodiment of the present invention. Figure 9 is a diagram for explaining the relationship between the defect density and the defect omission rate in the defect density. Figure 10 is based on the defect density in the Huate judgment side (four). The number is used to describe the unqualified volume. [Main component symbol description] 110 storage unit 120 data conversion unit 130 defect data analysis unit 14 0 coil quality determination unit 24