200825405 九、發明說明： 【發明所屬之技術領域】 • 、本發明係關於一種瑕疵檢查裴置，針對具有既定之被檢查區 域^檢查對象，將由區域感測器所拍攝之影像資訊加以處理，並 、根據從該區域内之亮度變化所計算之特徵量識別瑕疵。 【先前技術】200825405 IX. Description of the invention: [Technical field to which the invention pertains] The present invention relates to a sputum inspection apparatus for processing image information captured by an area sensor for a predetermined inspection area Identifying 瑕疵 based on the amount of features calculated from changes in brightness within the area. [Prior Art]
—，有:種瑕窥檢查裝置，使用由區域感測器所拍攝之影像資 \針對液晶顯示元件等板狀檢查對象檢查亮度瑕/疵（如黑顯示 之p示不應發光之晝素發光之瑕疵）。 圖9係顯示由區域感測器所運作之瑕疵檢查裝置之基本系統 ί體圖。板狀之檢查對象100之被檢查區域A，透過透鏡 一由具有區域感測器之攝影機3〇〇所拍攝，所拍攝之影像資訊 父予影像處理裝置4〇〇。 Λ個^電腦等所構成之影像處理裝置4⑻，根據透過影像輸 ^取传之影像資訊，藉由將檢查對象100之既定區域之亮度 一貝訊鈿以數値方式處理以進行瑕疵檢查。 R。系顯示影像處理裝置400之具體構成例之功能方塊 二旦;禮次，機300所具有之CCD等構成之區域感測器301拍攝 =貝汛，經由影像輸入板4〇1由影像⑻交 予刚處理機構4〇2。 測哭處^機構402，由攝影機瑕疵修正機構402a進行區域感 ^1 ί之亮度遺漏處之修正處理後，由陰影修正機構402b執 處=勻化區域感測器之周邊部之感度與中心部之感度差之 处！ ’父予偵測處理機構403。 濾波機，為使亮點瑕酬顯而由微分系之加強 閨値純準之贼部分為“Γ，低位準之部分為“0”，以 200825405 抽出瑕疵之候選對象，並 予編號後交予判定處理機=4邱構4〇3c對各抽出候選對象賦 • 於判疋處理機構404，由转料旦古管4办丄* 對象逐-執行特徵量計算 ^:二、，_針對瑕疲 -値，例如平均亮度、最大亮^㈣定暇疲之數 對比等。 儿度冗度體1貝、平均對比、最大平均 構，被輪出至識別機 選對_値之候 ，機構撕出40目6丄貞測處理機構403交予判定處理 識別機構404b。 ^取I田之s値，作為閾値1^設定於 於步裝置⑽ 正，再於步驟S3進行陰影修正。 進仃攝影機瑕疵之修 ^ ^ #ίίί ss i 步，進行識別處理，於步驟S、/輸i判定、= 寺徵量之計算，於 學系二透鏡與區域感測器之配置關係之光 之光軸成像於區域感測器301之中心附近1B0方200 ，邊附近之圖像之主獅、B2，二對; 而入射於透鏡2⑽，成像於區域感^3〇心: 者，的透鏡時，即使檢查對象100之亮度全面皆係均勻 下，亦“發二胃=測器301之周邊圖像明亮度亦較為低 6 200825405 近高區ΐ感測器301所拍攝之影像，將其中央附 由中心部到相反侧周邊部為止之書素數。 為降低如此陰影之影響，於 數 =取藉由陰影修正機構402 影；而：，- 正處理，並設置加強遽波器的方式以^值/慮波"進订陰影修 之4亮一強度之瑕痖，區域感測器 成水平狀）者，但視透應，固定（在曲線圖上應繪 瑕疵，區域感测器之 尺寸同一強度之 在内之瑕財身之特徵量。X彳―“μ修正包含平均亮度 評價Ξ/Ϊ二:特徵量’與以目視進行之主觀 感測器之平均對比輸出之^性圖。尺寸、同—強度之瑕疯，區域 對比一般可由下列式計算。 ，亮度—周邊亮度）X100/周邊意产 圖其原因在於透鏡之解析度。 析度亦會之特性圖。關於透鏡之解 而言解析度在光軸上最高光軸^頃斜角θ變化。-般 低。，由於透鏡之像差愈“ ㈣大）愈為降 圖17中，橫軸顯千金而馬曰加之故。 示拍攝黑白紋路時之對光軸之間之相對距離，縱軸顯 度（空間頻率）之差显。 、'、圖上之3組線顯示黑白紋路之細 200825405 40[lp^m]eflt^^ 1〇[lp/mm] ' 2〇[lp/mm] ' 而，:値閣値1之位準。然 如籍由ίί:2ί中：：當之閾値， 2’::目;:以==^”置一機 2。1個—域中最適當之閾値作為閾値Li，設定】識二二 【專利文獻1】日本特開雇—⑽96號公報 【發明内容】 決之 1¾ 8 200825405 使用習知裝置時’如上述必須為各一旦/你_ι/ν 士 r t丄、* =之閨値，若有多數檢查項叫ς須^檢g ‘ ^查區域準備瑕㈣測之最適當·，_之設定操作極= 狀之崎度隨各個透鏡不同。並且亦隨透鏡之隹距調敕 狀悲、大幅受化：伴隨於此，瑕轉徵量之計算値亦不同：、、、。正 以往，為對應處理如此問題，必須進 更換透鏡，及於每次焦距調整狀態變化於母次 位置確認不同之多數閾値調整後 二„項目及視 〜料間或縣之增加，触轉隨;裝置之維 資訊之瑕症彼此間之強度比較。丁故判疋，非常難以有效的識別 本發明係為解決上述問題 瑕症檢查裝置，其不需進行 其^的在於實現-種 穩定的檢查，且可提昇瑕_度之ί 重f設定，可進行 此之強度。 、/、精名度，能比較使瑕疵彼 „如此之課題，本發明構成 由區_測器所拍攝之檢查區域之檢 该£域内之亮度變化所計算之旦，貝汛加以處理，並根據從 其特徵在於： 具有-特量來識別瑕苑； 感測器之區域中心部與構，其可將起因於該區域 正。 1之冗度輪出差之該特徵量差加以修 ，’預先測定該區域感測器内，中具有—修正資料產生機 取ίίΐ-ΐϊ ’該特徵量面内修布特性並保存所產 取侍L正，料。 枝構错由雜正資料產生機構 9 200825405 (3 )如（2 )之瑕症檢杳 X方向及Y方向成格子狀分^ 2 ^該修正資料產生機構將 性瑕症圖以該區域感測器加以拍職所形成之假 x方向或γ方向之多次多項式近鼻之特徵量分布以 (4)如（2)或（3)之3之形恶保存。 修正機構參照所取得之該修正m裝置，其中該特徵量面内 之區域特徵量換算為該區域之卜修正演算，將所拍攝 發明之效果 伟被1。 依本發明可期待如以下之效果。 (υ可不受影像位置所拘束二曾 密度進行瑕疵判定得以實現。藉此斤％疋之特徵量，可使以高精 ((ba))2i1=f變更‘判定之_。 時間或成本之增加及伴隨此之因此檢查裝置維修 減輕。 座線作動率之降低等問題將大幅 ⑶即使不進行藉由影像處理 =同或更為穩定之精密度進行4;^ Γ免攝與假性瑕簡等之影像叹生修正資料，能實 =資料自_定二値化 實現檢查時間之縮短。、&、象之日守點去除雜訊之閾値。藉此可 【實施方式】 實施發明之 10 200825405 查裝1顯示適財發明之瑕疵檢 與習知裝置相將賦予同—符號以省略説明。 定處理機構404之構成，第1特徵部，在於判 之間新_特徵量面内修正機機構4〇4a與識別機構 影像處理裝置400之第2特料却/。 生機構407，其產生用以執行面t正有一修正資料產 該特徵量面内修正機構40如。仏正之貝料D，再交予新設之 且影像處理裝置4〇〇之第3丄 内修正機構404c，可省略習知壯卩，在於藉由設置特徵量面 影修正機構4〇2b。 衣中设置於前處理機構402之陰 序之.找號處独 ，疲之敍，不進行於即在麵S2進行攝影 丽進至步驟S4。 又置之步知幻執行之陰影修正即 S7之特徵討算|#^2 ^ ’與f知裝置者_。 =構407之修正D執又S10 ’根據來自修正> ^ 者位於區域_器之中央^、日士Α=Α4之瑕苑 2定之特_比即知中央與邊緣:比= 1 口改善為I·2倍而趨於 200825405 瑕藏圖同一尺寸、同一強度之 16之特性相比，t*!均對輸出之4寸性圖。與f知裝置中之圖 *於穩定。 卩知中央與邊緣之比從2.5倍改善為U倍而趨 - 囷係”’、頁示由修正資料產生機構407所運作之修正眘粗吝* ^於步驟.輸人假性瑕疲影像，該假性《影像 彳 =職讀性贼目者，在㈣U計算所象 witA於步驟S3 ’產生將所計算之假性瑕鋪徵量於X方 v步i S4作為次多項式近似之近似式，健此近似式於 象二係===之m生瑕疵圖500係模擬檢查對 *内呈格子二;:能同一強度之假性，x方向及γ方向 上，圖500設置於標準光源 裝置400。 I體構成將所拍攝之影像輸入資訊影像處理 以此影像貧訊為基礎，藉由特徵量 ,3之=徵量，例如平均亮度、最^構亮4以异： 1 “二大平均對比等，將所計算之特徵量P交予修正資料產i 以機構407對於所交予過來之每一特徵量，計管 • ^域中央部為丨時之時，此比率係對於每—^ .之夕土 Λ之/行（1〜N) ’利用相對於X方向之座標（Pix!〇 夕t夕項式，分別求出近似式FI、F2...FN—1、。 最後產生整合有各行之γ方向之座標（ 之各行近似式之各項係數之修正用資料並^子之)/、母i徵量 12 200825405 驟sr，程㈣賴。於步 徵量面内修正機構404c。 斤產生之修正貧料D輸入特 標之至欲修正之瑕疲之座 部之比率。下灯比革之線性内插，計算瑕苑位置相對於中央 之相對於部^匕^對算除以在步驟S2所計算 置時之特徵量（面内修正特徵量又疵=，央部位 於中央部之比料算無椒=如额執行相對—, There is a kind of peek-inspecting device that uses the image sensor photographed by the area sensor to check the brightness 瑕/疵 for a plate-shaped inspection object such as a liquid crystal display element (if the black display p indicates that the illuminance should not be illuminated) After that)). Figure 9 is a diagram showing the basic system of the inspection device operated by the area sensor. The inspected area A of the plate-shaped inspection object 100 is imaged by a camera 3 having an area sensor through a lens, and the captured image information is fed to the image processing apparatus. The image processing device 4 (8) composed of a computer or the like performs processing on the basis of the image information transmitted and transmitted through the image, and the brightness of the predetermined area of the inspection object 100 is processed in a number of ways to perform the defect inspection. R. The function block of the specific configuration example of the image processing apparatus 400 is displayed; the ceremony, the area sensor 301 of the CCD or the like of the machine 300 is photographed, and the image is input from the image input board 4〇1 by the image (8). Just processed the organization 4〇2. After the camera 瑕疵 correction mechanism 402a performs the correction processing of the brightness defect of the area sensation, the shadow correction mechanism 402b performs the sensitivity and the center part of the peripheral portion of the homogenization area sensor. The difference in feeling! 'Father detection processing mechanism 403. The filter machine, in order to make the highlights pay, is enhanced by the differential system. The thief part is “Γ, the low-level part is “0”, and the candidates of the 瑕疵 2008 are extracted by 200825405, and are numbered and then submitted to the judgment. Processor=4Qiu4〇3c is assigned to each candidate for extraction. 疋Processing mechanism 404, by the transfer of the ancients 4 丄* object-by-execution feature quantity calculation ^: Second, _ for fatigue -値, for example, the average brightness, the maximum brightness ^ (four) 暇 之 之 对比 对比 。 。 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿 儿The target processing unit 403 is given to the determination processing identification unit 404b. ^I 之 値 値 値 値 値 値 I I I I 设定 设定 设定 设定 设定 设定 设定 设定 设定 设定 设定 设定 设定 设定 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影 阴影^ ^ #ίίί ss i step, identification processing, calculation in step S, / input i judgment, = temple levy, optical axis of light in the configuration relationship between the two lenses and the area sensor 1B0 side 200 near the center of the detector 301, the main lion, B2, two pairs of images near the side; When incident on the lens 2 (10) and imaged in the lens of the region, even if the brightness of the inspection object 100 is uniformly uniform, the brightness of the surrounding image of the second stomach = detector 301 is also low. 6 200825405 The image captured by the near-high-area sensor 301 is attached to the center of the image from the center to the peripheral side of the opposite side. In order to reduce the influence of such shadows, the number = is taken by the shadow correction mechanism 402; and:, - is processing, and the way of strengthening the chopper is set to ^ value / care wave "瑕痖, the area sensor is horizontal), but the visualization is fixed, fixed (the graph should be painted on the graph, the size of the area sensor is the same strength and the characteristics of the wealth. X彳― The "μ correction includes the average brightness evaluation Ξ / Ϊ 2: feature quantity" and the average contrast output of the subjective sensor by visual observation. The size, the same - intensity of the madness, the area contrast can generally be calculated by the following formula. , brightness - peripheral brightness) X100 / peripheral intention map is due to the resolution of the lens. The resolution is also a characteristic map. The resolution of the lens is the highest optical axis on the optical axis. - Generally low. Because the aberration of the lens is "(4) large), the more it falls, the lower the axis, the more the horse is added. It shows the relative distance between the optical axes when shooting black and white lines, and the difference between the vertical axis (spatial frequency) is obvious. , ', the three groups of lines on the map show the black and white texture of the fine 200825405 40 [lp ^ m] eflt ^ ^ 1 〇 [lp / mm] ' 2 〇 [lp / mm] ' And,: 値 値 之 1 position . However, if ίί:2ί中::When the threshold is 値, 2'::目;:==^^" set a machine 2. 1 - the most appropriate threshold 域中 in the domain as the threshold 値Li, set] 知二二 [ Patent Document 1] Japanese Special Employment - (10) 96 [Invention] Contents 13⁄4 8 200825405 When using a conventional device, 'if the above must be for each / you _ι/ν 士 丄 丄, * =, if There are a lot of inspection items called ς 检 检 ' ' ' ' ' ' ' ' ' ' ' ' 查 查 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 四 设定 设定Greatly affected: With this, the calculation of the enthalpy of enthalpy is different: , , ,. In the past, in order to deal with such problems, it is necessary to change the lens, and each time the focus adjustment state changes to the parent position to confirm different After the adjustment of the majority of the thresholds, the increase in the number of items and the items or the counties in the county will be compared with the intensity of the information on the devices. It is very difficult to effectively identify the present invention as a septic check device for solving the above problems, which does not need to perform a stable check, and can improve the 瑕 度 重 f f Do this intensity. , /, the fine name, can compare the subject of such a problem, the present invention constitutes the inspection of the inspection area taken by the zone detector, the brightness of the change in the field is calculated, and the treatment is based on It is characterized in that: it has a special quantity to identify the center of the sensor; the central part and structure of the sensor, which can be repaired by the characteristic difference of the redundancy of the area. In the area sensor, there is a correction data generating machine that takes ίίΐ-ΐϊ 'the characteristic amount of in-plane repairing characteristics and saves the produced L-positive material. The misalignment is generated by the miscellaneous data generating mechanism 9 200825405 ( 3) If (2), the sputum examination 杳 X direction and Y direction are in a grid shape ^ 2 ^ The correction data generation mechanism will use the area sensor to take the false x direction or γ formed by the area sensor The multi-polynomial near-nose characteristic quantity distribution of the direction is preserved by (4) the shape of (2) or (3) 3. The correction mechanism refers to the corrected m device obtained, wherein the regional features of the feature quantity surface The amount is converted into the correction calculation of the area, and the effect of the invention is taken. Wei is 1. According to the present invention, the following effects can be expected. (υ The 曾 υ υ 二 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾 曾2i1=fChanges the 'decision'. The increase in time or cost and the maintenance of the inspection device are alleviated. The problem of lowering the seat line actuation rate will be large (3) even if the image processing is not the same or more stable precision Degree 4; ^ Γ 摄 与 与 假 与 与 与 与 与 与 与 与 与 与 假 假 假 假 假 假 假 假 假 假 假 假 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像 影像The threshold is 借此. 实施 实施 实施 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 2008 The second component of the new in-plane correcting mechanism 4〇4a and the recognizing mechanism image processing device 400 is determined by the generating mechanism 407, which generates a feature for performing the correction of the surface t. In-plane correction mechanism 40 such as The bedding material D is then supplied to the third correction mechanism 404c of the newly formed image processing apparatus 4, and the conventional technique is omitted, by providing the feature amount mask correction mechanism 4〇2b. The pre-processing mechanism 402 is in the negative order. The search for the position is unique, and the fatigue is not performed, so that the photograph is performed on the surface S2 to the step S4. The shadow correction of the step-by-step execution is the characteristic of the S7 | #^2 ^ 'And the device that knows the device _. = The correction of the structure 407 D and S10 'based on the correction from the ^ ^ is located in the center of the area _ device ^, Japanese Α Α = Α 4 瑕 2 2 2 That is to say, the center and the edge: compared to the = 1 port is improved to I 2 times and tends to 200825405. Compared with the characteristics of the same size and 16 of the same intensity, t*! is a 4-inch map of the output. And the figure in the device is stable. It is known that the ratio of the center to the edge is improved from 2.5 times to U times - the system is "", and the correction is performed by the correction data generating institution 407. * In the step of inputting the false fatigue image, The false "image 彳 = job-reading thief, in the (four) U calculation of the witA in step S3 'generate the calculated pseudo 瑕 征 于 于 X X X X X X , , , , , , , , , , , , , , , , , This approximation is in the image of the second system === m 瑕疵 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 500 The I body constitutes the image input information image processing based on the image poor information, by the feature quantity, 3 = levy, such as the average brightness, the best structure 4 different: 1 "two major average comparison Etc., the calculated feature quantity P is given to the correction data production i. For each feature quantity that the institution 407 gives, when the central part of the domain is 丨, the ratio is for each ^^. The eve of the earth / / / line (1 ~ N) 'Using the coordinates relative to the X direction (Pix! 〇 t 夕 夕 夕 , , , , Formula FI, F2...FN-1, Finally, the coordinates of the gamma direction of each row are integrated (the correction data of each coefficient of each line approximation is added) and the mother i eigenvalue 12 200825405 sr , Cheng (4) Lai. Correction mechanism 404c in the step of the amount of correction. The correction of the poor material D is the ratio of the input to the seat of the seat to be corrected. The linear interpolation of the lower lamp is calculated. The relative position of the position relative to the center is divided by the feature quantity calculated by the step S2 (the in-plane correction feature quantity is 疵=, and the central part is located at the central part, and the ratio is calculated as no relatively
=實施形態中雖已使用光電轉換 I 盗301以説明本發明，但作 ^配置之£域感測 電轉換元件以！維（直線上)配—'，亦可_於使用光 置。 _置之、、泉型感測器之瑕疵檢查裝 吾人雖已例示液晶顯示元件作為 可適用於任何形成有與瑕疵同程度 1⑽，但本發明亦 查對象物。亦可適用於例如形成有Τρ；:ΐ= 見則性模式之檢 有彩色濾'光片之液晶用玻璃、PDp ^二面^用基板、形成 晶胞(cell)、形成有IC模式之石夕晶圓等冤水如益面板）用玻璃或 吾人雖已例示就假性瑕疵而言， 之假性瑕疯圖500，但若能針對多數^巧―尺寸假性瑕痖 則可應檢查縣之尺寸域修正:# =^樣之修正資料’ 行檢查。 、十猎此可以更高之精密度進 吾人雖已例示保存以多次多項式 、 修正用資料，但亦可藉由將相對於中之近似式之係數作為 器或線型感測器之座標加以映射，、而承、^^比率預先以區域感測 叩文马向逮地進行修正。 【圖式簡單說明】 200825405 方』。1係顯示適用本發明之瑕疫檢查裝置之一實施形態'之功能 綱圖。2 蝴咖峨料理程序之 不意圖 =3係如説懈本發騎 圖' 々丁之知试里面内修正之概念之 特性圖。手如4寸如面内修正後區mm、之平均亮度輸出之 之 特性I5 m示特徵量面内修正後區域感測器之平均對比輸出 ^係顯示修正資 圖7係假性瑕簡之平面圖序之4圖。 之舰量面_球叙流程圖。 之立體圖。1 丁叫域感測器運作之瑕錄查裝置基本1統構成 圖ι〇係頒示影像處理裝 =示影像處二 學系:糸顯示檢查對象、透鏡與區域感測器Γ酉Γ置關係之光 圖。圖13 _以說明以區域感測器所拍攝之影像之陰影之特性 圖14係針對同+ 均亮度輪出特性圖。寸、同一強度之瑕疵，區域感測器之平 圖π係顯示針:c。 之平，t比輸出之特^ 、同-強度之瑕疵，區域感測器 圖17係顯示一般透鏡之解析度之特性圖。 14 200825405 圖18係顯示以圖16之平均對比輸出特性之區域感測器進行 亮點瑕疵^拍攝時之特徵量偵測模式之特性圖。 ' 【主要元件符號說明】 - 1、2〜閾値 1⑻〜檢查對象 200〜透鏡 300〜攝影機 301〜區域感測器 400〜影像處理裝置 ^ 401〜影像輸入板 402〜前處理機構 402a〜攝影機瑕蘇修正機構 402b〜陰影修正機構 403〜彳貞測處理機構 403 a〜加強滤波機構 403b〜二値化機構 403c〜標示機構 404〜判定處理機構 ( 404a〜特徵量計算機構 404b〜識別機構 404c〜特徵量面内修正機構 405〜判定結果輸出機構 - 406〜閾値設定機構 - 407〜修正資料產生機構 500〜假性瑕疵圖 A〜被檢查區域In the embodiment, although the photoelectric conversion I thief 301 has been used to explain the present invention, it is configured as a domain sensing electrical conversion element! Dimensions (on the line) with -', can also be used for light. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ It can also be applied, for example, to the formation of Τρ;:ΐ= see the pattern of the liquid crystal for the color filter, the PDp^ two-sided substrate, the cell, and the IC pattern.夕 晶圆 冤 冤 ) ) ) ) ) ) ) ) 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆 晶圆Size domain correction: # =^ Sample correction data 'line check. The ten-hunt can be higher precision. Although we have exemplified the preservation of multiple polynomial and correction data, we can also map the coordinates of the approximate approximation to the coordinates of the linear or linear sensor. , and the ratio of the bearing and the ^^ is corrected in advance by the area sensing 叩文马. [Simple description of the diagram] 200825405 side. 1 shows a functional outline of an embodiment of a plague inspecting apparatus to which the present invention is applied. 2 蝴 峨 峨 峨 峨 = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = The hand such as 4 inch, such as the in-plane correction of the back zone mm, the characteristic of the average brightness output I5 m shows the average contrast output of the area sensor after the correction of the feature amount in-plane correction correction system map 7 is a pseudo-simplified plan 4 of the sequence. The ship's measuring surface _ ball theory flow chart. Stereo view. 1 Ding called the domain sensor operation 瑕 瑕 瑕 瑕 基本 基本 〇 〇 〇 颁 颁 颁 颁 颁 颁 颁 颁 颁 颁 颁 颁 颁 = = = = = = = = = = = = = = = = = = = = = = Light map. Fig. 13 _ to illustrate the characteristics of the shadow of the image taken by the area sensor. Fig. 14 is a graph showing the same + uniform brightness rotation. Inch, the same intensity, the flatness of the area sensor π shows the needle: c. Level, t is more than the output of the ^, the same - intensity of the area, the area sensor Figure 17 is a characteristic diagram showing the resolution of the general lens. 14 200825405 Fig. 18 is a characteristic diagram showing the feature amount detection mode when the area sensor of the average contrast output characteristic of Fig. 16 performs bright spot detection. ' [Main component symbol description] - 1, 2 to threshold 値 1 (8) - inspection target 200 to lens 300 - camera 301 - area sensor 400 - image processing device ^ 401 to video input pad 402 - pre-processing mechanism 402a - camera 瑕Correction mechanism 402b to shading correction mechanism 403 to detection processing unit 403a to enhancement filter unit 403b to demultiplexing mechanism 403c to labeling unit 404 to determination processing unit (404a to feature amount calculation unit 404b to identification unit 404c to feature amount) In-plane correction mechanism 405 to determination result output means - 406 - threshold threshold setting means - 407 - correction data generation means 500 - false map A - inspected area
Al、A2、A3、A4〜區域感測器周邊部位置 B〜區域感測器中央部位置 15 200825405 Β0···Β2〜主光線 D〜修正資料 FI、F2".FN—1、FN〜相對於X方向之座標（pixel)之以 多次多項式表達之近似式 Li〜閾値 LI、L2."Ln〜言曼定値 P〜特徵量 S1-"S10〜步驟 0〜傾斜角Al, A2, A3, A4~area sensor peripheral position B~area sensor central position 15 200825405 Β0···Β2~main ray D~correction data FI, F2".FN-1, FN~ relative The approximation of the coordinate in the X direction (pixel) with multiple polynomial expression Li~threshold 値LI, L2."Ln~ 言曼定値P~feature amount S1-"S10~step 0~tilt angle
CC
1616