200842342 (1) 九、發明說明 【發明所屬之技術領域】 本發明係關於用以從木材的圓木等切出之單板用料等 木質材料檢測變色所導致之缺陷部份之木材的檢查方法及 ' 裝置。例如，製造合板時，以利器切削圓木而得到連續之 • 厚度數毫米的單板，使該單板具有特定的大小且進行乾燥 後’利用接者劑接者複數片單板使其一體化。於該等製造 步驟，必須針對對於單板的品質會造成影響之木材表面的 ® 變色所導致之缺陷、變形、單板之木節瘤脫落所形成之孔 之部位、龜裂等之缺陷位置、數量、及面積等之程度，篩 選構成成爲合板時之表層，亦即，美觀缺點較少之物，以 及用以構成合板之內層之物，亦即，缺點較多但不會形成 問題之物。該篩選係例如分成5〜7個階段。 【先前技術】 0 傳統上，用以構成成爲合板時之表層之物、及用以構 成合板內層之物之篩選，係由作業者以肉眼判定利用輸送 帶搬運之單板。 此外，傳統之木材缺陷部的檢查方法’係以彩色 " C C D攝影機實施木材之攝影，針對將映像信號’利用色 影像析出裝置將松脂及變色之基準色進行對比來實施二値 化，對與檢測對象區域一致之二値化影像實施標示處理並 與判定値進行對比，來檢測松脂等之附著樹脂、腐鈾、變 色之缺陷部位（參照專利文獻1 )。 -4- 200842342 (2) [專利文獻1]日本特開平9-2 1 0785號公報 【發明內容】 前述之傳統技術殘留如下所示之課題。 利用肉眼之判定時，會有人所造成之判定誤差而不正 '確’而且，無法加快輸送帶的速度，而有生產性較差等問 題。 此外，傳統之與基準色對比來實施二値化之木材缺陷 ® 部的檢查方法時，並非實施使用色分佈之正確缺陷檢查。 本發明之目的在於解決如上所示之傳統課題，藉由攝 影手段實施單板等木材之攝影，從該攝取之影像，利用色 分佈，正確地檢測對木材品質會造成影響之木材表面的變 色所導致之缺陷部份，連常異常部之色彩、亮度等之變化 也可正確地檢測。 第1圖係單板篩選裝置的說明圖。第1圖中，1係影 ^ 像處理裝置（影像處理手段），2係分選機控制裝置，3 係操作盤，4係帶式輸送機，5係透射光用照明，6係反 射光用照明，7係等級別分配裝置，8係生產線感測攝影 機（攝影手段），9係單板（木材）。 ' 本發明爲了解決上述課題，具有以下之構成。 (1)利用攝影手段8實施木材9的彩色攝影，利用 影像處理手段1求取藉由前述攝影手段8所攝取之彩色影 像的色分佈，將該所求取之色分佈與預先設定之正常木材 的色分佈進行比較，將該所求取之色分佈從前述正常木材 -5- 200842342 (3) 的色分佈偏離特定値以上時視爲異常色分佈，該異常色分 佈於藉由前述攝影手段所攝取之木材面上之區域大於特定 値時，視爲檢測出木材缺陷。因此，可利用色分佈正確地 檢測對木材品質會產生影響之木材表面的變色所造成之缺 ^ 陷部份。 k ( 2 )如前述（1 )的木材的檢查方法或裝置，以缺陷 面積相對於檢查對象區域爲相對較小之木材爲檢查對象時 ，作爲前述預設之正常木材的色分佈，置換並使用每次針 ® 對各檢查對象所取得之影像分佈置換前述預設之正常木材 的色分佈。因此，容易取得預設之正常木材的色分佈。 (3)如前述（1)或（2)的木材的檢查方法或裝置 ，求取前述攝取之彩色影像的色分佈之亮度直方圖，檢測 亮度異常部。因此，焦黑等亮度異常部之檢測十分容易。 (4 )如前述（3 )的木材的檢查方法或裝置，前述預 設之正常木材的色分佈的亮度直方圖係整體作爲遵從正規 分佈者，而由部份區域的累積頻率來推算整體的正規分佈 。因此，即使未預定正常松木材之色分佈，亦可從檢查之 木材推算正常木材的色分佈。 * 依據本發明，具有以下之效果。 ' (1 )利用影像處理手段，求取藉由攝影手段所攝取 之彩色影像的色分佈’將該所求取之色分佈與預先設定之 正常木材的色分佈進行比較’將該所求取之色分佈從前述 正常木材的色分佈偏離特定値以上時視爲異常色分佈，該 異常色分佈於藉由前述攝影手段所攝取之木材面上之區域 -6 - 200842342 (4) 大於特定値時，視爲檢測出木材缺陷，可利用色分佈正確 地檢測對木材品質會產生影響之木材表面@變色所造成之 缺陷部份。 (2 )以缺陷面積相對於檢查對象區域爲相對較小之 - 木材爲檢查對象時’作爲前述預設之正常色分佈’置換並 、 使用每次針對各檢查對象所取得之影像分佈’容易取得預 設之正常木材的色分佈。 (3 )求取攝取之彩色影像的色分佈之亮度直方圖， ^ 來檢測亮度異常部，很容易檢測焦黑等之亮度異常部。 (4)預設之正常木材的色分佈的亮度直方圖係整體 作爲遵從正規分佈者，而由部份區域的累積頻率來推算整 體的正規分佈，故即使未先設定正常木材的色分佈，亦可 從檢查之木材推算正常木材的色分佈。 【實施方式】 (1 )單板篩選裝置的說明 第1圖係單板篩選裝置的說明圖。第1圖係單板篩選 裝置的整體構成。單板篩選裝置配設著影像處理裝置1、 分選機控制裝置2、操作盤3、帶式輸送機4、透射光用 照明5、反射光用照明6、等級別分配裝置7、以及生產 線感測攝影機8。 影像處理裝置1執行來自生產線感測攝影機8之影像 資料的處理，係將單板品質等級等的處理結果輸出至分選 機控制裝置2之影像處理手段。分選機控制裝置2依據影 200842342 (5) 像處理裝置1的輸出，驅動輸送帶、輸出停止等之分選機 輸送帶控制信號、以及輸出等級別分配裝置7的控制信號 。操作盤3係用以執行影像處理裝置1之設定値的變更、 分選機控制裝置2之控制等之操作的操作盤。帶式輸送機 * 4係用以搬運單板9之搬運手段。透射光用照明5係以檢 ' 測單板9之孔、龜裂等爲目的之LED等的照明手段（光 源），使用與反射光用照明6爲不同色之照明（例如綠色 的照明）。其目的係在與來自反射光用照明6之反射光進 W 行區別（利用色及強度進行區別），用以檢測單板之孔（ 木節瘤孔）、龜裂等。反射光用照明6係以檢測單板9之 反射光爲目的之LED等的照明手段（光源），通常係使 用白色的照明。生產線感測攝影機8係實施單板9之生產 線影像之攝影的攝影手段。 該單板篩選裝置的動作係以生產線感測攝影機8對帶 式輸送機4所運送之單板9進行攝影，並將影像資料輸出 ^ 至影像處理裝置1。於影像處理裝置1執行該影像資料的 處理，並將單板品質等級等處理結果輸出至分選機控制裝 置2。分選機控制裝置2對等級別分配裝置7輸出控制信 號，實施單板9之等級別筛選。該篩選係依據蟲孔數、孔 * '脫落木節瘤數、活木節瘤數、死木節瘤數、缺口數、龜 裂數、樹脂·樹穴數、青變數、變形（無條理値）等及其 大小（面積）等之程度來實施。 (2 )影像處理裝置的說明 -8- 200842342 (6) 第2圖係影像處理裝置的說明圖。第2圖中，影像處 理裝置配設著3台的生產線感測攝影機8a、8b、8c、攝 影機影像取得用基盤11a、11b、11c、雷射標不器12a、 12b、雷射驅動器13a、13b、主電腦14。 生產線感測攝影機8a、8b、8c係利用3台攝影機， 將單板9於垂直於搬運方向之方向分割成3份，以彩色進 行攝影之攝影手段。攝影機影像取得用基盤1 1 a、1 1 b、 g 1 1 c係每次從生產線感測攝影機分別取得1生產線的影像 度，即實施數位化處理，並將影像資料傳送給主電腦14 。雷射標示器12a、12b係用以照射於單板之搬運方向的 光線作爲以合成（結合）來自生產線感測攝影機8a、8b 、8c之各影像爲目的之標識。可以該照射之光線於以後 之處理可容易除去之方式，照射與單板（木材）之色爲不 同色之微細光線（例如，紅色雷射單色光線）。雷射驅動 器13a、13b係連結於AC電源，用以驅動雷射標示器12a 、：12b。主電腦14焉備處理手段、儲存手段、輸出手段等 ，用以執行單板9之影像的處理（影像的合成、木節瘤探 查、缺陷探查處理等）。此處，攝影機影像取得用基盤 1 1 a、1 1 b、1 1 c及主電腦1 4係影像處理手段。 影像處理裝置的動作係對搬運而來之單板9照射來自 透射光用光源5及反射光用光源6之光線，每次於攝影機 影像取得用基盤11a、11b、11c從生產線感測攝影機8a 、8b、8c取得1生產線影像時，對主電腦14傳送該資料 。主電腦1 4執行接收到之影像的補正並檢測無條理値， -9 - 200842342 ⑺ 依序結合影像。最後，於攝影機影像取得用基盤1 1 a、 1 1 b、1 1 c結束影像取得之時點，主電腦1 4也大致完成彩 色影像的合成及黑白濃淡影像變換。其次，被分割成3部 份之單板的影像之來自攝影機影像取得用基盤1 1 a、1 1 b 、1 1 c的影像，利用主電腦1 4進行結合。 此處，係對單板9照射來自雷射標示器12a、12b之 雷射標識來分割成3部份，生產線感測攝影機8a、8b、 8c則分別組合至雷射標識爲止之生產線影像，很簡單即 ® 可結合影像。此外，爲了提高影像的處理速度，木節瘤的 探查處理亦可以畫素數較多之黑白濃淡影像來實施，死木 節瘤的探查等之彩色影像則以縮小（畫素數較少）之影像 來實施。 以下，將影像處理裝置的動作分成攝影中的處理及攝 影後的處理來進行說明。 <攝影中的處理之說明> 利用生產線感測攝影機8a、8b、8c所攝影之影像資 ® 料，針對每1生產線傳送給主電腦14，並合成爲1張整 體影像。 * •攝影機影像取得用基盤1 1 a、1 1 b、1 1 c的處理 * 從生產線感測攝影機8a、8b、8c取得1生產線彩色 影像，檢測雷射標識的位置（接合位置），將該資訊與1 生產線彩色影像一起傳送給主電腦1 4。 •主電腦1 4的處理 執行接收到之1生產線彩色影像的補正且檢測無條理 -10- 200842342 (8) 値’依據上述位置資訊（雷射標識）進行合成。此處，檢 查對象若存在無條理，因爲前述雷射標識的位置檢測軌跡 會變形成非直線狀，利用該變形量，可檢測無條理値。 於攝影機影像取得用基盤1 1 a、1 1 b、1 1 c之攝影結束 ，接收到最後之1生產線彩色影像之階段，主電腦1 4已 完成整體彩色影像的合成。此外，爲了有效利用攝影中的 時間’同時執行黑白變換及縮小處理等可針對1生產線執 _ 行之處理。 <攝影後的影像解析中的處理之說明> •攝影機影像取得用基盤1 1 a、1 1 b、1 1 c的處理 直到檢測到下一板（單板）之到達爲止，進行待機。 •主電腦1 4的處理 依據對象之板的大小及種類等之既定資訊，利用計算 之區域及設定値，執行木節瘤探查處理、利用透射光之缺 陷檢測處理等，最後，執行含有無條理値之等級分類處理 JI 。將結果顯示於未圖示之顯示裝置，而且，將結果輸出至 分選機控制裝置2。 此外，於前面之說明中，係針對影像處理裝置內之攝 影機影像取得用基盤1 1 a、1 1 b、1 1 c、主電腦1 4等使用 電腦（PC )時進行說明，然而，其所使用之電腦數可以 依據影像資料量及電腦的處理速度等來進行變更。此外， 亦可以1台電腦來執行處理。 此外，係針對使用3台生產線感測攝影機進行說明， 然而，可依據對象之板的大小及種類、以及電腦的處理性 -11- 200842342 Ο) 能，使用1台、2台或4台以上。 (3 )單板表面的變色所造成之缺陷部的檢測之說明 對木材品質會產生影響之木材表面的變色所造成之缺 陷部份，利用以下的手段及方法來進行檢測。 變色所導致之缺陷部係指從外部進入木材之黴菌等所 造成之變色部，或者，使用於合板之單板等時，呈現於表 g 面之乾燥機所造成之焦黑、木材的樹皮跡、形成於木材內 部之樹脂等。 將該等視爲缺陷部來進行檢測。 (缺陷部的檢測手段之說明） 1 )利用彩色生產線感測器攝影機實施木材的表面之 攝影，並利用輸入裝置（輸入手段）將該影像傳送給電腦 〇 2 )將該攝影影像以由RGB (紅、綠、藍）所構成之 p 各畫素之彩色影像保存於電腦記憶體上（儲存手段） 3 )利用電腦影像處理程式（影像處理裝置）將RGB 影像變換成HSV (色相、彩度、亮度）影像。 4 )利用118¥影像以下述方法來檢測缺陷部。 (缺陷部的檢測方法之說明） 同一樹種的正常（健全）木材之表面色與其濃淡無關 ’而分佈成大致爲特定之彩度及色相的區域。然而，黴菌 等之缺陷部係導致於材質的不同，與健全色的分佈於彩度 '色相會出現偏離。此外，將焦黑等之缺陷部與健全色的 -12- 200842342 (10) 分佈進行比較，會分佈於較黑（低亮度）之部份。 因此，針對檢查對象木材表面的色分佈調查相對於健 全木材表面的色分佈之彩度、色相的偏差及亮度的偏差， 將較大偏差値的部份檢測成缺陷部係此方法之特徵。 ' (4 )取得做爲基準之健全木材表面的色分佈的方法之說 明 ^ 1 )針對檢查對象樹種，利用彩色生產線感測器攝影 機實施健全木材的表面之攝影。 2 )爲了得到充份之統計精度，上述應針對同一樹種 ，實施複數張（20張程度以上）之不同狀態的表面之攝 影。 3 )將上述之全部影像的各畫素色配置於電腦記憶體 上的3次元色空間，建立3次元色分佈。 3次元色空間可以爲採用RGB (紅、綠、藍色空間） ^ 者、Hsv (色相、彩度、亮度色空間）、或Lab色空間（ 「L」爲亮度，「a」係表示從綠朝紅，「b」係表示從藍 朝黃之色相要素）等。 4 )針對各3次元色分佈的等亮度面求取2次元分佈 ，得到表示最大頻率之點。 5 )階段地變化亮度，可以得到近似連結上述4 )之 最大頻率點的曲線。將該曲線稱爲3次元色分佈的基準中 心軸。 例如，HSV色分佈之亮度取0·0〜1.0之範圍値時， -13- 200842342 (11) 求取各亮度區分爲〇· ο 2之具有等亮度値之畫素的色相、 彩度的2次元分佈’得到由連結該最大頻率點而成之曲線 ，將其視爲3次元色分佈的基準中心軸。此外，也同時求 取色相、彩度的2次元分佈之標準偏差cjc(v)。 6 )預先得知缺陷部面積相對於檢查對象區域爲相對 較小時’亦可以針對各檢查對象所取得之影像分佈置換該 基準分佈。亦即，因爲只要知道健全部的分佈的平均及標 _ 準偏差的値即可。 (5 )缺陷檢查方法之說明 1 )利用彩色生產線感測器攝影機實施檢查對象的木 材表面之攝影。 2 )將該影像的各畫素配置於3次元色空間，建立3 次元色分佈。 3 )以下述方法求取相對於3次元色分佈之基準中心 _ 軸的色彩偏差値。 例如，若對象影像之X、y位置的畫素爲g[x，y]、該 HSV色分佈空間的色爲 色相値：h ( g [ X，y ]) 彩度値：S ( g[x，y]) 亮度値：V ( g[x，y])， 則先前所求取之3次元色分佈的基準中心軸之特定売度V 的基準中心軸座標爲 色相値：H ( v )、彩度値：S ( ν ) ’ -14- 200842342 (12) 以該等亮度平面的橫軸做爲X，縱軸做爲γ，則如第3圖 所示。第3圖係將影像g之各點的色變換成HS平面上之 說明圖。第3圖中，係將對象影像之木材畫素g[x，y]變換 成HS平面上之垂直相交座標X2、Y2者。此外，雖然色 的分佈（參照網狀）不是圓形而爲各種形狀分佈，然而， 標準偏差大致爲圓形。 此處，基準中心軸座標H(v) 、S(v)的垂直相交 座標X1、Y1如下所示。 X1=S ( v ) · cos ( 2π · Η ( ν ) /3 60 ) Υ1 = S ( ν ) · sin ( 2π · Η ( ν) /360 ) 畫素g[x，y]之h(v) 、s(v)的垂直相交座標Χ2、 Υ2如下所示。 X2 = s ( ν ) · cos ( 2π · h ( ν ) /3 60 ) Y2 = s ( ν ) · sin ( 2π · h ( ν) /3 60 ) 距離基準中心軸之平方空間距離d可以下式求取。 d2= ( X1-X2 ) 2 十（Y1-Y2 ) 2 因此，色彩偏差値Z c [ X，y ]如下所示。。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 And 'device. For example, when manufacturing a plywood, the cutting board is cut with a sharp tool to obtain a continuous veneer having a thickness of several millimeters, so that the veneer has a specific size and is dried, and then the plurality of veneers are integrated by the connector. . In these manufacturing steps, it is necessary to correct the defects caused by the discoloration of the surface of the wood which affects the quality of the veneer, the deformation, the location of the hole formed by the knot of the veneer of the veneer, the crack, and the like. The degree of the quantity, the area, and the like, the screening constitutes the surface layer when the board is folded, that is, the object with less aesthetic defects, and the object for constituting the inner layer of the ply, that is, the object having many disadvantages but not forming a problem . The screening is for example divided into 5 to 7 stages. [Prior Art] 0 Conventionally, the panel for conveying the surface layer when the panel is laminated and the panel for forming the inner layer of the panel are visually judged by the operator to be conveyed by the conveyor belt. In addition, the traditional method for inspecting wood defects is to perform wood photography with a color "CCD camera, and to perform binary conversion by comparing the reference color of the rosin and discoloration with the image signal using the color image deposition device. The dichroic image in which the detection target area is identical is subjected to labeling processing and compared with the judgment , to detect a defect portion such as rosin or the like which adheres to resin, humus or discoloration (see Patent Document 1). 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 。 When the judgment of the naked eye is used, there is a problem that the judgment error caused by the person is not correct, and the speed of the conveyor belt cannot be accelerated, and the productivity is poor. In addition, the traditional method of checking the defect of the wood defect ® in comparison with the reference color is not the correct defect inspection using the color distribution. An object of the present invention is to solve the conventional problem as described above, and to perform photographing of wood such as a veneer by means of photographing, and to accurately detect the discoloration of the surface of the wood which affects the quality of the wood by using the color distribution from the image taken. The defect caused by the change in the color, brightness, etc. of the abnormal part can also be detected correctly. Fig. 1 is an explanatory view of a veneer screening device. In the first figure, the 1 system image processing device (image processing means), the 2 series sorter control device, the 3 series operation panel, the 4 series belt conveyor, the 5 series transmission light illumination, and the 6 system reflection light. Lighting, 7-series and other level distribution devices, 8 series production line sensing cameras (photographic means), 9 series veneers (wood). In order to solve the above problems, the present invention has the following constitution. (1) Color photographing of the wood 9 is performed by the photographing means 8, and the color distribution of the color image taken by the photographing means 8 is obtained by the image processing means 1, and the obtained color distribution is set to a predetermined normal wood. The color distribution is compared, and the color distribution obtained is deviated from the color distribution of the normal wood-5-200842342 (3) by a specific color or more, and the abnormal color distribution is distributed by the aforementioned photographing means. When the area on the surface of the ingested wood is larger than the specific enthalpy, the wood defect is considered to be detected. Therefore, the color distribution can be used to accurately detect the defective portion caused by the discoloration of the surface of the wood which affects the quality of the wood. (2) The method or apparatus for inspecting wood according to the above (1), wherein the color distribution of the predetermined normal wood is replaced and used when the wood having a relatively small defect area with respect to the inspection target area is to be inspected. The image distribution obtained by each needle® for each inspection object replaces the color distribution of the aforementioned predetermined normal wood. Therefore, it is easy to obtain the color distribution of the preset normal wood. (3) The method or apparatus for inspecting wood according to the above (1) or (2), wherein a luminance histogram of the color distribution of the color image taken as described above is obtained, and a luminance abnormal portion is detected. Therefore, the detection of abnormal brightness portions such as burnt black is very easy. (4) The method or apparatus for inspecting wood according to the above (3), wherein the brightness histogram of the color distribution of the predetermined normal wood is generally used as a normal distributor, and the cumulative frequency of the partial region is used to estimate the overall regularity. distributed. Therefore, even if the color distribution of the normal loose wood is not predetermined, the color distribution of the normal wood can be estimated from the examined wood. * According to the present invention, the following effects are obtained. ' (1) Using image processing means to obtain a color distribution of a color image taken by a photographing means to compare the obtained color distribution with a color distribution of a predetermined normal wood 'to obtain the desired The color distribution is regarded as an abnormal color distribution when the color distribution of the normal wood is deviated from a specific enthalpy or more, and the abnormal color is distributed over the area on the wood surface taken by the aforementioned photographic means -6 - 200842342 (4) is larger than a specific 値, It is considered to detect wood defects, and the color distribution can be used to correctly detect the defects caused by the surface of the wood which is affected by the quality of the wood. (2) The area of the defect is relatively small with respect to the area to be inspected - when the wood is the object to be inspected, 'the normal color distribution as the preset is replaced', and the image distribution obtained for each inspection object is easily obtained. Preset the color distribution of normal wood. (3) The luminance histogram of the color distribution of the ingested color image is obtained, and the luminance abnormal portion is detected, and the luminance abnormal portion such as burnt black is easily detected. (4) The brightness histogram of the color distribution of the normal wood of the preset is the normal distribution, and the cumulative distribution of the partial area is used to estimate the overall normal distribution, so even if the color distribution of the normal wood is not set first, The color distribution of normal wood can be estimated from the wood being inspected. [Embodiment] (1) Description of single-plate screening device Fig. 1 is an explanatory view of a veneer screening device. Fig. 1 is an overall configuration of a veneer screening device. The veneer screening device is provided with an image processing device 1, a sorter control device 2, an operation panel 3, a belt conveyor 4, a transmitted light illumination 5, a reflected light illumination 6, a level distribution device 7, and a production line sense. Camera 8 is measured. The image processing device 1 executes the processing of the image data from the line sensing camera 8, and outputs the processing result of the board quality level or the like to the image processing means of the sorter control device 2. The sorter control device 2 drives the sorter conveyor control signal of the conveyor belt, the output stop, and the like, and the control signal of the level equalizing device 7 according to the output of the image processing device 1 according to the image 200842342 (5). The operation panel 3 is an operation panel for performing an operation of changing the setting of the image processing apparatus 1, the control of the sorter control device 2, and the like. Belt conveyor * 4 is a means of transport for transporting the veneer 9. The illumination light 5 is an illumination means (light source) such as an LED for detecting a hole or a crack of the veneer 9, and illumination (for example, green illumination) of a different color from the reflected light illumination 6 is used. The purpose is to distinguish between the reflected light from the reflected light illumination 6 (differentiated by color and intensity) for detecting holes (wood knot holes), cracks, and the like of the single plate. The illumination 6 for reflection light is an illumination means (light source) such as an LED for detecting the reflected light of the single-plate 9, and white illumination is usually used. The line sensing camera 8 is a means of photographing the production of the line image of the veneer 9. The operation of the veneer screening device is to take a picture of the veneer 9 conveyed by the belt conveyor 4 by the line sensing camera 8, and output the image data to the image processing device 1. The image processing device 1 executes the processing of the image data, and outputs the processing result such as the quality level of the board to the sorter control device 2. The sorter control device 2 outputs the control signal to the peer-to-peer distribution device 7, and performs level-level screening of the board 9. The screening is based on the number of wormholes, the number of holes*'s number of detached ganglions, the number of live ganglions, the number of dead wood knots, the number of gaps, the number of cracks, the number of resin and tree holes, the number of cyanosis, and the deformation (unregulated) It is implemented to the extent of its size (area) and so on. (2) Description of the image processing device -8- 200842342 (6) Fig. 2 is an explanatory diagram of the image processing device. In Fig. 2, the image processing apparatus is provided with three line sensing cameras 8a, 8b, 8c, camera image acquisition substrates 11a, 11b, 11c, laser markers 12a, 12b, and laser drivers 13a, 13b. , the main computer 14. The line sensing cameras 8a, 8b, and 8c are three-part cameras, and the board 9 is divided into three in a direction perpendicular to the conveyance direction, and photographing means for photographing in color. The camera image acquisition bases 1 1 a, 1 1 b, and g 1 1 c each acquire the image quality of one line from the line sensing camera, that is, perform digitization processing, and transmit the image data to the host computer 14. The laser marker 12a, 12b is used to illuminate the light in the direction in which the sheets are conveyed as a mark for synthesizing (combining) the respective images from the line sensing cameras 8a, 8b, 8c. The light to be irradiated can be irradiated with a fine color of a different color (for example, a red laser monochromatic light) in a manner similar to the color of the veneer (wood). The laser drivers 13a, 13b are coupled to an AC power source for driving the laser markers 12a, 12b. The main computer 14 is configured to process, store, and output the image of the single board 9 (image synthesis, wood knot exploration, defect detection processing, etc.). Here, the camera image acquisition bases 1 1 a, 1 1 b, 1 1 c and the host computer 14 are image processing means. In the operation of the image processing apparatus, the light from the transmitted light source 5 and the reflected light source 6 is applied to the transported single board 9, and the camera 8a, 11b, and 11c are photographed from the line sensing camera 8a every time. When 8b, 8c obtains a production line image, the data is transmitted to the host computer 14. The main computer 1 4 performs the correction of the received image and detects the unorganized 値, -9 - 200842342 (7) Combine the images sequentially. Finally, at the time when the camera image acquisition bases 1 1 a, 1 1 b, and 1 1 c complete the image acquisition, the host computer 14 also substantially completes the color image synthesis and the black and white image conversion. Next, the images from the camera image acquisition bases 1 1 a, 1 1 b , and 1 1 c which are divided into three parts of the image of the single board are combined by the host computer 14 . Here, the single board 9 is irradiated with the laser marks from the laser markers 12a, 12b to be divided into three parts, and the line sensing cameras 8a, 8b, 8c are respectively combined to the line image of the laser mark, which is very Simple is the combination of images. In addition, in order to improve the processing speed of the image, the detection processing of the ganglion tumor can also be performed by drawing a black-and-white image with a high number of primes, and the color image of the dead wood tumor is reduced (the number of pixels is small). Image to implement. Hereinafter, the operation of the image processing apparatus will be described by dividing the processing in shooting and the processing after shooting. <Description of Process in Photography> The image material captured by the line sensing cameras 8a, 8b, and 8c is transmitted to the host computer 14 for each production line and combined into one entire image. * Processing of the camera image acquisition bases 1 1 a, 1 1 b, and 1 1 c * 1 line color image is acquired from the line sensing cameras 8a, 8b, 8c, and the position (joining position) of the laser mark is detected. The information is transmitted to the host computer 1 4 together with the color image of the production line. • Processing of the main computer 1 4 Performing correction of the received color image of the production line and detecting the unorganized -10- 200842342 (8) 値 'Synthesis based on the above position information (laser mark). Here, if the inspection object is unorganized, since the position detection trajectory of the aforementioned laser marker is changed into a non-linear shape, the amount of deformation can be used to detect the unstructured flaw. At the end of the photography of the camera image acquisition bases 1 1 a, 1 1 b, and 1 1 c , the main computer 14 has completed the synthesis of the overall color image at the stage of receiving the final 1 line color image. Further, in order to effectively utilize the time in photography, the black and white conversion and the reduction processing are simultaneously performed, and the processing can be performed for one production line. <Description of Process in Image Analysis after Photography> • Processing of Camera Image Acquisition Bases 1 1 a, 1 1 b, and 1 1 c Until the arrival of the next board (single board) is detected, standby is performed. • The processing of the main computer 14 is based on the predetermined information such as the size and type of the target board, and the calculation of the area and settings, the execution of the knot detection process, the use of the transmitted light defect detection process, etc., and finally, the execution is unorganized.値 等级 classification processing JI. The result is displayed on a display device not shown, and the result is output to the sorter control device 2. In addition, in the foregoing description, the case where the computer (PC) is used for the camera image acquisition bases 1 1 a, 1 1 b, 1 1 c, and the host computer 1 4 in the image processing apparatus is described. The number of computers used can be changed depending on the amount of image data and the processing speed of the computer. In addition, one computer can also perform processing. In addition, the description will be given for the use of three production line sensing cameras. However, one, two or four or more can be used depending on the size and type of the target board and the handling of the computer -11- 200842342 Ο). (3) Description of the detection of the defective portion caused by the discoloration of the surface of the veneer The defect caused by the discoloration of the surface of the wood which affects the quality of the wood is detected by the following means and methods. The defective portion caused by the discoloration refers to a discolored portion caused by mold or the like entering the wood from the outside, or when used in a veneer or the like, the coke black caused by the dryer on the surface of the surface, the bark of the wood, A resin or the like formed inside the wood. These are regarded as defective parts for detection. (Description of Detection Method of Defective Part) 1) The surface of the wood is photographed by a color line sensor camera, and the image is transmitted to the computer by an input device (input means) 2) The photographic image is made of RGB ( Red, green, and blue) The color images of each pixel are stored in computer memory (storage means). 3) The RGB image is converted to HSV by using a computer image processing program (image processing device) (hue, chroma, Brightness) image. 4) The defect portion is detected by the following method using the 118¥ image. (Explanation of the detection method of the defective portion) The surface color of the normal (sound) wood of the same tree species is distributed irrespective of its shading and is distributed into a region of substantially specific chroma and hue. However, the defect of the mold or the like causes a difference in the material, and the distribution of the sound color in the chroma 'color phase will deviate. In addition, the defect portion such as burnt black is compared with the distribution of the healthy color -12-200842342 (10), and is distributed in the darker (lower brightness) portion. Therefore, the color distribution of the surface of the wood to be inspected is investigated for the chroma of the color distribution of the surface of the solid wood, the deviation of the hue, and the deviation of the brightness, and the portion having the large deviation 値 is detected as the defect portion. (4) Description of the method of obtaining the color distribution of the surface of the sound of the wood as a reference ^ 1 ) For the inspection target tree species, the surface of the sound of the wood is photographed using a color line sensor camera. 2) In order to obtain sufficient statistical accuracy, the above-mentioned photographs of the surface of different states (more than 20 sheets) should be performed for the same tree species. 3) The respective pixel colors of all the above images are arranged in the 3 dimensional color space on the computer memory to establish a 3 dimensional color distribution. The 3 dimensional color space can be RGB (red, green, blue space) ^, Hsv (hue, chroma, luminance color space), or Lab color space ("L" is brightness, "a" is expressed from green In the red, "b" means the color element from the blue to the yellow). 4) A 2-dimensional distribution is obtained for the equal luminance planes of the respective 3 dimensional color distributions, and a point indicating the maximum frequency is obtained. 5) The brightness is changed stepwise, and a curve which approximates the maximum frequency point of the above 4) can be obtained. This curve is called the reference center axis of the 3-dimensional color distribution. For example, when the brightness of the HSV color distribution is in the range of 0·0 to 1.0, -13- 200842342 (11) Find the hue and chroma of each pixel having the same brightness as 〇· ο 2 The dimensional distribution 'obtains a curve obtained by joining the maximum frequency points and regards it as the reference central axis of the 3-dimensional color distribution. In addition, the standard deviation cjc(v) of the 2-dimensional distribution of hue and chroma is also obtained. 6) When it is known in advance that the area of the defective portion is relatively small with respect to the inspection target area, the reference distribution may be replaced for the image distribution acquired for each inspection target. That is, it is only necessary to know the average of the distribution of the health and the 标 deviation of the standard deviation. (5) Description of the defect inspection method 1) The photograph of the surface of the wood to be inspected is carried out using a color line sensor camera. 2) The pixels of the image are arranged in the 3 dimensional color space to establish a 3 dimensional color distribution. 3) Determine the color deviation 値 relative to the reference center _ axis of the 3 dimensional color distribution by the following method. For example, if the X, y position of the object image is g[x, y], the color of the HSV color distribution space is hue h: h ( g [ X, y ]) Saturation 値: S ( g[x , y]) Brightness 値: V ( g[x, y]), then the reference center axis coordinate of the specific curvature V of the reference central axis of the previously obtained 3 dimensional color distribution is the hue 値: H ( v ), Saturation 値: S ( ν ) ' -14- 200842342 (12) The horizontal axis of these luminance planes is X, and the vertical axis is γ, as shown in Fig. 3. Fig. 3 is an explanatory diagram for converting the color of each point of the image g to the HS plane. In Fig. 3, the wood pixel g[x, y] of the object image is transformed into the vertical intersecting coordinates X2, Y2 on the HS plane. Further, although the distribution of colors (refer to the mesh shape) is not circular but distributed in various shapes, the standard deviation is substantially circular. Here, the vertical intersecting coordinates X1 and Y1 of the reference center axis coordinates H(v) and S(v) are as follows. X1=S ( v ) · cos ( 2π · Η ( ν ) /3 60 ) Υ1 = S ( ν ) · sin ( 2π · Η ( ν) /360 ) h(v) of the pixel g[x,y] The vertical intersecting coordinates Χ2 and Υ2 of s(v) are as follows. X2 = s ( ν ) · cos ( 2π · h ( ν ) /3 60 ) Y2 = s ( ν ) · sin ( 2π · h ( ν ) /3 60 ) The square space distance d from the reference central axis can be given by Seek. D2= ( X1-X2 ) 2 十(Y1-Y2 ) 2 Therefore, the color deviation 値Z c [ X, y ] is as follows.
Zc[x,y]= ( Vd ) /( σ c ( ν ) X /3 c ) 此處，σ c ( v )係基準中心軸之亮度v之色相、彩度 2次元分佈的標準偏差ac(v) 。/3c係決定將距離基準 中心軸爲σ c ( ν )之數倍之色視爲異常之係數，例如， 1 . 0〜2.0程度的値。 此外，使用Lab色空間等其他色分佈時，亦同樣可求 取空間距離的偏差。 -15- 200842342 (13) 4) 其次，爲了求取實際缺陷部份的區域，色分佈空 間之與標準相隔之色的畫素，必須只選擇由原來之木材畫 素g[x，y]所形成之局部塊的畫素。其可以使用考慮周邊之 畫素的色，除去不連續之孤立點，強調偏差較大之部份的 . 處理，亦即，可以使用影像處理技法之一般被稱爲弛緩法 - 之方法等。 其實例如下所不’亦即，利用將相對於基準中心軸之 色彩偏差値Zc[x，y]視爲初始標記之弛緩法，決定色彩異 ^ 常所造成之缺陷區域（參照第4圖之說明）。 (亮度異常所造成之缺陷部份的檢測之說明） 5) 求取3次元色分佈之基準中心軸（亮度軸）方向 的直方圖。 6) 上述健全部之亮度直方圖係平均値Vm、標準偏差 σν之正規分佈（高斯分佈）時，以下述方式求取亮度偏 差値 zv[x，y]。 • Zv[x?y] = |Vm-g[x9y]. V|/ ( 〇 νΧ β γ) 冷ν係決定將距離亮度平均値Vm爲σν的數倍之亮 度視爲異常之係數，例如，1 . 0〜4 · 0程度的値。 • 色彩及亮度的綜合偏差値Zt[x，y]如下所示。Zc[x,y]= ( Vd ) /( σ c ( ν ) X /3 c ) where σ c ( v ) is the standard deviation ac of the hue and chroma 2 dimensional distribution of the luminance v of the reference central axis. v). /3c determines that the color which is a multiple of σ c ( ν ) from the reference center axis is regarded as an abnormal coefficient, for example, a degree of 1.0 to 2.0. In addition, when other color distributions such as the Lab color space are used, the deviation of the spatial distance can also be obtained. -15- 200842342 (13) 4) Secondly, in order to obtain the area of the actual defect part, the color of the color distribution space separated from the standard must be selected only by the original wood pixel g[x, y]. The pixels of the local block formed. It is possible to use a color that considers the surrounding pixels to remove discontinuous isolated points, emphasizing a portion with a large deviation. Processing, that is, a method generally called a relaxation method using image processing techniques. The example is as follows. That is, the defect region caused by the color difference is determined by using the relaxation method of the color deviation 値Zc[x, y] with respect to the reference central axis as the initial mark (refer to FIG. 4). Description). (Description of the detection of the defective part caused by the abnormal brightness) 5) Find the histogram of the reference center axis (brightness axis) direction of the 3rd color distribution. 6) When the luminance histogram of the above-mentioned health is the normal distribution (Gaussian distribution) of the mean 値Vm and the standard deviation σν, the luminance deviation 値 zv[x, y] is obtained in the following manner. • Zv[x?y] = |Vm-g[x9y]. V|/ ( 〇νΧ β γ) The cold ν system determines the brightness from the brightness average 値Vm as several times σν as the coefficient of the abnormality, for example, 1 . 0~4 · 0 degree of 値. • The overall deviation of color and brightness 値Zt[x,y] is as follows.
Zt[x?y] = Zc[x?y]-fZv[x5y] 亮度直方圖因爲攝影機特性等而非正規分佈時（例如 ，亮度接近1·〇之明亮部份爲飽和等），無法正確求取平 均値Vm、標準偏差σν。此時，直方圖係標準正規分佈者 -16- 200842342 (14) ，規格化之累積機率分佈涵數F ( x )可表示如下。 [數1] 1 (x-/i)2 F(x)= J f(x)dx=--J exp{—-- }dx ηΐπσ1) 2σ2 (此處，x爲亮度、//爲亮度的平均値、σ爲標準偏差） 從亮度較低之一方開始積算亮度直方圖，利用將該積 算値除以全畫素數割之値（累積頻率），分別求取相當於 以下之pl、p2、P3、p4之亮度，並將其視爲V1、V2、 V3、Vm。 P1=F ( // -2.0 σ ) =0.0228 p2 = F ( μ -1.0α ) =0.1587 p3=F ( // -0.05 α ) =0.3 085 p4 = F ( β ) =0.5 此外，VI、V2、V3、Vm之可取之有效區域 Vmin、 Vmax，係依據經驗從基準之木材等求取，例如，設定成 V m i η = 0.2 5、V m a X = 0 · 9 〇 a)於Vl、V2、V3、Vm內，檢索存在於有效區域內 之相當於 V2: F(Vm-a ) =0.1587 及 Vm: F(Vm) =0.5 之亮度位置，即可求取推算平均値x = Vm及標準偏差σ v (參照第8圖）。 b )然而，亮度分佈小於x=//而使分佈形狀崩潰時等 ，Vm位於有效區域外。此時，求取存在於有效區域內之 VI : F ( Vm-2.0 σ v )及 V2 : F ( Vm-Ι ·0 σ v )， -17- 200842342 (15) σ v = V2-VlZt[x?y] = Zc[x?y]-fZv[x5y] The luminance histogram cannot be correctly obtained because of camera characteristics, etc., rather than normal distribution (for example, the brightness is close to 1·〇, the bright part is saturated, etc.) Take the average 値Vm and the standard deviation σν. At this time, the histogram is a standard normal distributor -16- 200842342 (14), and the normalized cumulative probability distribution number F (x) can be expressed as follows. [Equation 1] 1 (x-/i) 2 F(x)= J f(x)dx=--J exp{—-- }dx ηΐπσ1) 2σ2 (here, x is brightness, // is brightness The average 値 and σ are standard deviations. The luminance histogram is calculated from one of the lower luminances, and the pl, p2 is obtained by dividing the total 値 by the 画 (cumulative frequency) of the full-pixel prime cut. The brightness of P3 and p4 is regarded as V1, V2, V3, and Vm. P1=F ( // -2.0 σ ) =0.0228 p2 = F ( μ -1.0α ) =0.1587 p3=F ( // -0.05 α ) =0.3 085 p4 = F ( β ) =0.5 In addition, VI, V2 The effective areas Vmin and Vmax of V3 and Vm are obtained from the reference wood according to experience, for example, set to V mi η = 0.2 5, V ma X = 0 · 9 〇a) at Vl, V2, V3 In Vm, the brightness position corresponding to V2: F(Vm-a) = 0.1587 and Vm: F(Vm) = 0.5 existing in the effective area is retrieved, and the estimated average 値x = Vm and the standard deviation σ can be obtained. v (Refer to Figure 8). b) However, when the luminance distribution is smaller than x=// and the distribution shape is collapsed, etc., Vm is outside the effective area. At this time, find VI: F ( Vm-2.0 σ v ) and V2 : F ( Vm-Ι ·0 σ v ) existing in the effective area, -17- 200842342 (15) σ v = V2-Vl
Vm = V2+cr ν 而可推算σν及Vm(參照第9圖）。 或者，亮度異常部相對較大時， • VI : F ( Vm-2.0 σ v )及 V2 : F ( Vm-1.0c ' 效區域外。此時，由存在於有效區域內之亮度Vm = V2+cr ν and σν and Vm can be estimated (refer to Fig. 9). Or, when the brightness abnormality is relatively large, • VI : F ( Vm-2.0 σ v ) and V2 : F (Vm-1.0c ' outside the effective area. At this time, the brightness existing in the effective area
Vm— 〇·5σ ) =0.3085 及 Vm: F (Vm) =0.5 可 Vm-V3 ) X2，故可推算σ v及Vm (參照第10 I ® 使用該方法，未使用基準木材，此外，與 關，也可求取健全部的平均値及標準偏差（此 準之木材時，不執行求取處理，而採用該平均 差）。 7 )其次，爲了求取實際缺陷部份的區域 間之與標準相隔之色的畫素，必須只選擇由原 素g[x，y]所形成之局部塊的畫素。其可以使用 I 畫素的色，除去不連續之孤立點，強調偏差較 處理，亦即，可以使用影像處理技法之一般被 之方法等。 其實例如下所示，亦即，利用將色彩及亮 差値Zt[x，y]視爲初始標記之驰緩法（參照第 )，決定色彩及亮度異常所造成之缺陷區域。 到目則爲止’木材之自動品質檢查時，只 之明暗、或指定特定色來檢測缺陷，該方法時 明亮部份及色變化時，故全部爲良品。 「v)位於有 値 V3 : F ( 得到σ v=( 囊）° 分佈形狀無 外，使用基 値及標準偏 ，色分佈空 來之木材畫 考慮周邊之 大之部份的 稱爲弛緩法 度之綜合偏 4圖之說明 利用表面色 ，無法對應 -18- 200842342 (16) 木材的表面色當中，該肉眼有不調和感之對品質會產 生影響之部份，通常不會是木材原本之自然色，而會以3 次元色空間之色分佈之差異呈現出來。利用以分離檢測該 差異之統一方法’可以良好精度檢測缺陷部份。 ' 此外’對木材品質會造成影響之黴菌等，會因爲木材 • 的種別及產地而爲不同之色。 以單一之方法很難以良好精度實施全部之檢測。因此 ，依據本發明，即使檢查對象之木材種類不同，無需變更 ® 基準中心軸座標値即可進行檢測。而且，即使因爲樹種而 使檢測精度變差，只要變更初始値之3次元色分佈的基準 中心軸座標，即可恢復檢測精度。 此外’樹脂及樹皮等，到目前爲止，難以利用影像處 理之外觀檢查之檢測，亦可實現該等檢測。 因爲可以良好精度檢測樹皮等黑化之部份，判定是否 殘留黑化之樹皮，很容易即可判定生木節瘤、死木節瘤。 ^ ( 6 )驰緩法之說明 第4圖係弛緩法之說明圖。以下，係依照第4圖之處 理S1〜S3進行說明。 •本處理時，針對對象之影像的各畫素g ( X，y )，設定 缺陷機率P i ( X，y )。此處，P i ( X，y )係針對對第1次重 複後之畫素g ( x，y )之缺陷機率。 S 1 :影像處理裝置針對對象之影像之各畫素g ( x,y ) ’附與初始機率P〇(x，y) (0〜1.0)，進入處理S2。此 -19- 200842342 (17) 處，P0 ( x，y)係依據色偏差値Z ( x，y)的値而爲如下所 不 ° P0(x，y) =Z(x，y) ： 0<Z< 1.0 1.0： Z ^ 1.0 S2 :影像處理裝置針對全部畫素， 於（0.0<Pi(x，y) < 1.0 )時， 求取Pi(x，y)之鄰近畫素的機率平均値<p：>，並以 Pi+ 1 =Pi+ a ( <P>-Pi ) (α係周邊畫素的影響係數，爲l〜4程度） 更新機率，進入處理S3。 (Pi(x，y) S0.0 或 Pi(x，y) -1·〇)時，Vm— 〇·5σ ) =0.3085 and Vm: F (Vm) =0.5 can be Vm-V3 ) X2, so σ v and Vm can be calculated (refer to the 10th I ® using this method, the reference wood is not used, in addition, It is also possible to obtain the average 値 and standard deviation of the health (when this standard wood is used, the average difference is not used for the processing) 7) Secondly, in order to obtain the inter-regional standard of the actual defect For pixels that are separated by color, only the pixels of the local block formed by the prime g[x, y] must be selected. It can use the color of the I pixel to remove the discontinuous isolated points, emphasizing the deviation processing, that is, the general method of the image processing technique can be used. An example of this is as follows, that is, a defect region caused by an abnormal color and brightness is determined by a relaxation method (refer to the first) in which the color and the luminance 値Zt[x, y] are regarded as initial marks. At the time of the automatic quality inspection of wood, only the darkness or the specific color is specified to detect the defect. In this method, when the bright part and the color change, all are good. "v) is located in the presence of 値V3 : F (the σ v = ( sac) ° distribution shape is not used, the use of the base and standard deviation, the color distribution of the wood painting to consider the surrounding part of the large part of the called relaxation law The description of the comprehensive partial 4 map uses the surface color and cannot correspond to -18- 200842342 (16) Among the surface colors of wood, the part of the naked eye that has an unconformed effect on the quality will usually not be the original natural color of the wood. It will be presented in the difference of the color distribution of the 3 dimensional color space. The unified method of separating and detecting the difference can be used to detect the defective part with good precision. 'In addition, the mold which affects the quality of wood, etc., will be due to wood. • Different types and origins are different colors. It is difficult to carry out all tests with good precision in a single method. Therefore, according to the present invention, even if the type of wood to be inspected is different, it is not necessary to change the reference center axis coordinate mark to perform detection. Moreover, even if the detection accuracy is deteriorated due to the tree species, the detection center can be restored by changing the reference center axis coordinate of the initial 3 次 color distribution. In addition, 'resin and bark, etc., so far, it is difficult to detect the visual inspection using image processing, and these tests can also be realized. Since the blackened part of the bark can be detected with good precision, it is determined whether or not the blackened bark remains. It is easy to determine the rhythm and dead wood nodules. ^ (6) Description of the relaxation method Fig. 4 is an explanatory diagram of the relaxation method. Hereinafter, the description will be made according to the processing of S1 to S3 in Fig. 4. In this process, the defect probability P i ( X, y ) is set for each pixel g ( X, y ) of the target image. Here, P i ( X, y ) is for the painting after the first repetition. The probability of the defect of g (x,y). S 1 : The image processing device attaches to the initial probability P〇(x,y) (0~1.0) for each pixel g (x,y) of the image of the object. Process S2. This -19- 200842342 (17), P0 ( x, y) is based on the color deviation 値Z ( x, y) and is not as follows P0 (x, y) = Z (x, y ) : 0<Z< 1.0 1.0: Z ^ 1.0 S2 : The image processing device for all pixels, at (0.0<Pi(x,y) < 1.0 ), find the adjacent painting of Pi(x,y) Probability average <p:>, and update the probability with Pi + 1 = Pi + a ( < P > - Pi ) (the influence coefficient of the α-peripheral pixels, which is about 1 to 4), and enter the process S3. (Pi(x, y) When S0.0 or Pi(x,y) -1·〇),
Pi+l=Pi 不更新機率，進入處理S3。 S 3 :影像處理裝置調查收斂條件。 針對 P i ( X，y )， 重複次數I大於指定數時（1>指定數）， 若相對於Pi = 〇.〇及pi=1.0之畫素數的全部畫素之比例大 於指定率（ >指定率），則結束處理。 若非如此，重複處理S 2。 此處，重複之指定數爲1〇次程度，針對Pi = 〇.〇及 pi=l.〇之畫素數的全邰畫素之指定率爲99 %程度。 (7 ) HSV色空間之色分佈的說明 第5圖係HSV色空間之色分佈的說明圖。第5圖中 -20- 200842342 (18) ，向上方向係亮度（V:此處，ν = 0·0〜1·0 )，相同亮度 平面之直徑方向彩度（S:此處，S = 0.0〜1.0)，圓周方 向爲色相（Η :此處，H = 0°〜3 60° )。健全木材之色分佈 具有較大之上下的色分佈區域，該色分佈的中心軸（基準 * 中心軸）以向上方向之箭頭表示。 ' 此外，黴菌等之變色部之色彩異常部的色分佈，係以 右側之色彩異常區域來表示。此外，乾燥機所造成之焦黑 等之亮度異常部以下側之較小色分佈區域來表示。 (8 )特定亮度平面之畫素分佈的說明 第6圖係特定亮度v平面之畫素分佈的說明圖。第6 圖中，係特定亮度v平面之色彩異常部的畫素分佈。此處 ，色彩異常部的畫素分佈係分佈於標準偏差σ c ( v )之特 定區域（參照網狀部）。 ^ ( 9 )中心軸方向的畫素分佈之說明 第7圖係正規分佈的累積頻率F(x)之說明圖。第7 圖中，以虛線表示一般正規分佈（分佈機率），累積頻率 (累積機率分佈涵數）F ( X )係將木材的健全部之積算値 除以全部畫素數N之規格化者。此處，pi =0.0228 (// -2(7 )、ρ2 = 0.1 5 8 7 ( β -1.0α ) 、 ?3 = 0.3 08 5 ( β -0.5 σ )、 ρ4 = 0·5(平均値=#)。 第8圖係中心軸方向之畫素分佈（正常分佈形狀）的 說明圖。第8圖中，係3次元色分佈之基準中心軸（亮度 -21 - 200842342 (19) 軸）方向的直方圖。該亮度直方圖時，相當於前述pi、 p 2、p 3、ρ 4之売度分別爲 V1、V 2、V 3、V m，此外，木 材表面的健全部之可取之有效區域爲Vmin、Vmax。該圖 時，檢索相當於存在於有效區域內之 V2 : F ( Vm- σ ) = 0.1 5 87及Vm : F ( Vm) =0·5之亮度位置，並求取推算 平均値x = Vm及標準偏差σ ν。 第9圖係分佈形狀異常時之用以推算平均値Vm的說 明圖。第9圖中，係3次元色分佈之基準中心軸（亮度軸 )方向的直方圖。該亮度直方圖係木材表面的健全部之亮 度分佈小於x= //而使分佈形狀崩潰時等，Vm位於有效區 域外。此時，求取存在於有效區域內之VI : F ( Vm-2.0 σ ν )及 V2:F(Vm-1.0av)，利用 av = V2-Vl 及 Vm = V2 + σν推算σν及Vm。 第10圖係亮度異常部面積較大時的說明圖。第1〇圖 中，爲3次元色分佈之基準中心軸（亮度軸）方向的直方 圖。該亮度直方圖於木材表面的亮度異常部面積相對較大 時，VI: F(Vm-2.0av)及 V2: F(Vm-l.Oav)位於有 效區域外。因此，從存在於有效區域內之亮度値V3 : F ( νηι-0·5σν) =0.3085 及 Vm: F(Vm) =0.5 而得到 σν=( Vm-V3) Χ2，而可推算出σν及Vm。 如此，亮度直方圖因爲攝影機特性等而爲非正規分佈 時（例如，亮度接近1.0之部份之感測感度特性爲非線性 )，有時無法正確求取平均値Vm、標準偏差σ ν。此時’ 直方圖係標準正規分佈者，可以分佈之底部的該直方圖（ -22- 200842342 (20) 2個點）來推算整體的分佈，並求取平均値Vm 差σ v。 藉此，影像處理裝置可使用有效區域內 VmaX)之點VI〜Vm內之2點來求取（推算） * 平均値Vm及標準偏差σν。使用之2點的優先 • 均値Vm及其他點（VI〜V3的1點），平均値 於有效區域內時，利用V 1〜V3之2點。 ® ( 1 〇 )程式安裝的說明 以程式構成影像處理裝置（影像處理手段） 機控制裝置（分選機控制手段）2、攝影機影像 盤1 1 a、1 1 b、1 1 c、主電腦1 4等，由主控制部（ 行’係儲存於主記憶之物。該程式係由電腦執行 。該電腦係由主控制部、主記憶、檔案裝置、顯 之輸出裝置、輸入裝置等之硬體所構成。 φ 將本發明之程式安裝於該電腦。該安裝係將 記憶於軟式磁碟、光磁碟片等之移動型記錄（記 ’介由對電腦具備之記錄媒體存取之驅動器 LAN等之網路，安裝至配設於電腦之檔案裝置 很容易提供可利用色分佈正確地檢測對木材品質 響之木材表面的變色所造成之缺陷部份木材的檢 【圖式簡單說明】 第1圖係本發明的單板篩選裝置的說明圖。 及標準偏 (Vmin 〜 健全部的 順位爲平 Vm不位 1、分選 取得用基 CPU)執 處理之物 示裝置等 該等程式 憶）媒體 裝置、或 。藉此， 會產生影 查裝置。 -23- 200842342 (21) 第2圖係本發明的影像處理裝置的說明圖。 第3圖係將本發明的影像g各點的色變換至η S平面 上之說明圖。 第4圖係本發明的弛緩法的說明圖。 第5圖係本發明的HSV色空間之色分佈的說明圖。 第6圖係本發明的特定亮度ν平面之畫素分佈的說明 圖。 第7圖係本發明的正規分佈的累積頻率ρ ( X )的說 明圖。 第8圖係本發明的中心軸方向的畫素分佈（正常分佈 形狀）的說明圖。 第9圖係本發明的分佈形狀異常時所推算之平均値 V m之說明圖。 第10圖係本發明的亮度異常部的面積較大時之說明 圖。 【主要元件符號說明】 1 :影像處理裝置（影像處理手段） 2 :分選機控制裝置 3 :操作盤 4 :帶式輸送機 5 :透射光用照明（照明手段） 6 :反射光用照明（照明手段） 7 :等級別分配裝置 -24- 200842342 (22) 8 :生產線感測攝影機（攝影手段） 9 :單板（木材）Pi+l=Pi does not update the probability and proceeds to process S3. S 3 : The image processing device investigates convergence conditions. For P i ( X,y ), when the number of repetitions I is greater than the specified number (1 > specified number), the ratio of all pixels of the pixel number relative to Pi = 〇.〇 and pi=1.0 is greater than the specified rate ( > ; specified rate), then the processing ends. If not, repeat S 2 . Here, the specified number of repetitions is 1 degree, and the designation rate of the full-time pixels for the pixel numbers of Pi = 〇.〇 and pi=l.〇 is about 99%. (7) Explanation of the color distribution of the HSV color space Fig. 5 is an explanatory diagram of the color distribution of the HSV color space. In Fig. 5, -20- 200842342 (18), the brightness in the upward direction (V: here, ν = 0·0~1·0), the diopter saturation of the same brightness plane (S: here, S = 0.0) ~1.0), the circumferential direction is the hue (Η: here, H = 0°~3 60°). The color distribution of the sound wood has a color distribution area with a large upper and lower, and the central axis of the color distribution (reference * center axis) is indicated by an upward arrow. Further, the color distribution of the color abnormal portion of the discolored portion such as mold is indicated by the color abnormal region on the right side. Further, it is represented by a small color distribution region on the lower side of the brightness abnormal portion such as burnt black caused by the dryer. (8) Description of pixel distribution of a specific luminance plane Fig. 6 is an explanatory diagram of a pixel distribution of a specific luminance v plane. In Fig. 6, the pixel distribution of the color anomaly portion of the specific luminance v plane is shown. Here, the pixel distribution of the color abnormality portion is distributed in a specific region of the standard deviation σ c ( v ) (refer to the mesh portion). ^ (9) Explanation of the pixel distribution in the central axis direction Fig. 7 is an explanatory diagram of the cumulative frequency F(x) of the normal distribution. In Fig. 7, the general normal distribution (distribution probability) is indicated by a broken line, and the cumulative frequency (cumulative probability distribution metric) F (X) is calculated by dividing the total of the wood's health by the total number of pixels N. Here, pi =0.0228 (// -2(7), ρ2 = 0.1 5 8 7 (β -1.0α ) , ?3 = 0.3 08 5 ( β -0.5 σ ), ρ4 = 0·5 (average 値 = #). Fig. 8 is an explanatory diagram of the pixel distribution (normal distribution shape) in the central axis direction. In Fig. 8, the reference center axis of the 3 dimensional color distribution (luminance-21 - 200842342 (19) axis) Histogram. The luminance histogram corresponds to the pi, p 2, p 3, and ρ 4 degrees of V1, V 2, V 3, and V m, respectively, and the effective area of the surface of the wood. It is Vmin, Vmax. In this figure, the brightness position corresponding to V2 : F ( Vm - σ ) = 0.1 5 87 and Vm : F ( Vm) =0·5 existing in the effective area is retrieved, and the estimated average is obtained.値x = Vm and standard deviation σ ν. Fig. 9 is an explanatory diagram for estimating the average 値Vm when the shape is abnormal. In Fig. 9, the direction is the direction of the center axis (brightness axis) of the 3 dimensional color distribution. The brightness histogram is that the brightness distribution of the whole surface of the wood surface is less than x= // and the distribution shape collapses, etc., and Vm is outside the effective area. At this time, the solution exists in the effective area. VI : F ( Vm-2.0 σ ν ) and V2: F (Vm-1.0av), using σ = V2-Vl and Vm = V2 + σν to estimate σν and Vm. Figure 10 shows the area where the brightness anomaly is large. Fig. 1 is a histogram in the direction of the reference central axis (brightness axis) of the 3 dimensional color distribution. When the luminance histogram is relatively large in the brightness abnormal portion of the wood surface, VI: F(Vm- 2.0av) and V2: F(Vm-l.Oav) is outside the effective area. Therefore, the brightness 値V3 : F ( νηι-0·5σν) =0.3085 and Vm: F(Vm) = from the effective area 0.5 and σν=( Vm-V3) Χ2, and σν and Vm can be derived. Thus, when the luminance histogram is irregularly distributed due to camera characteristics or the like (for example, the sensing sensitivity characteristic of the portion where the luminance is close to 1.0 is Non-linear), sometimes the average 値Vm and the standard deviation σ ν cannot be correctly obtained. At this time, the histogram is a standard normal distribution, and the histogram at the bottom of the distribution can be distributed ( -22- 200842342 (20) 2 points) To estimate the overall distribution, and to obtain the average 値Vm difference σ v. Thereby, the image processing device can use the VmaX) point in the effective area VI~Vm To strike (estimated) * Average Zhi Vm and standard deviation σν. The priority of using 2 points • When averaging Vm and other points (1 point of VI~V3), when averaging within the effective area, use 2 points of V 1~V3. ® ( 1 〇) program installation instructions to form an image processing device (image processing means) machine control device (sorting machine control means) 2, camera image disk 1 1 a, 1 1 b, 1 1 c, host computer 1 4, etc., by the main control unit (line 'system is stored in the main memory. The program is executed by the computer. The computer is the main control unit, main memory, file device, display device, input device, etc. φ The program of the present invention is installed in the computer. The installation system is stored in a portable type such as a flexible disk or a magneto-optical disk (indicating a drive LAN for accessing a recording medium to a computer, etc.) The network, installed to the file device equipped with the computer, is easy to provide a part of the wood that can be used to accurately detect the discoloration caused by the discoloration of the wood surface due to the color distribution. [Simplified drawing] Figure 1 An explanatory diagram of a veneer screening device according to the present invention, and a standard device (such as a memory device that is processed by a Vmin-to-health is a Vm-in-one 1 and a sort-and-acquisition-based CPU). Or, a recording device will be generated. -23- 200842342 (21) Fig. 2 is an explanatory view of the image processing device of the present invention. Fig. 3 is a diagram for converting the color of each point of the image g of the present invention to η S 4 is an explanatory diagram of a relaxation method of the present invention. Fig. 5 is an explanatory diagram of a color distribution of an HSV color space of the present invention. Fig. 6 is a pixel of a specific luminance ν plane of the present invention. Fig. 7 is an explanatory diagram of the cumulative frequency ρ ( X ) of the normal distribution of the present invention. Fig. 8 is an explanatory diagram of the pixel distribution (normal distribution shape) in the central axis direction of the present invention. Fig. 10 is an explanatory diagram of the average 値V m estimated when the distribution shape of the present invention is abnormal. Fig. 10 is an explanatory diagram showing a case where the area of the luminance abnormal portion of the present invention is large. [Description of main component symbols] 1: Image processing apparatus (Image processing means) 2: Sorter control device 3: Operation panel 4: Belt conveyor 5: Illumination light (illumination means) 6: Illumination light (illumination means) 7: Equal-level distribution device-24 - 200842342 (22) 8 : Production line sensing camera ( Photographic means) 9 : veneer (wood)
-25--25-