【正文】
n forward direction, the behavior of each layer only has influence on the next layer and if the output is not ideal, the process will turn into backward direction, which return the error signal along the connection paths and transmit it to input layer by modifying the weight. Repeat this process until the error meets the demand. In backward direction, the weight is adjusted by this formula. ? pω ji = ηδ pj o pi (6) Where η is learning rate and its value should be selected between zero and one. When the nodes are in output layer, following definition is used, otherwise equation (8) is used. δ pj = (t pj ? o pj ) f 39。它可以真實,生動地代表物體的 3D和顏色信息。為了驗證校準精度,進行了一次以為 。此外, 3D之間的映射關(guān)系的坐標( XW, YW,度 Zw)及其相應的像素彩色圖像坐標得到了彩色 CCD可以通過顏色傳感器校準獲得。顏色傳感器是由 BP 神經(jīng)網(wǎng)絡(luò)校準上述輸入和輸出選擇為的三維坐標的校準點的和其相應的二維象素座標為圖 1示出。 網(wǎng)絡(luò)的訓練過程開始訓練樣本的制備,它包含輸入樣本與理想的輸出樣本。這就是線性分配法的基本概念,使用這種方法可以使測量誤差顯著減少。如果校準點的數(shù)目足夠, M完全可以通過由對象的空間坐標和它們的對 應像素坐標創(chuàng)建的求解線性方程系統(tǒng)來確定。并且 3D 顏色傳感器數(shù)字化得到的 3D 坐標和顏色信息匹配可以由內(nèi)部照相機來校準實現(xiàn)。校準的過程主要包括公式計算、求解過程以及信息匹配方法的詳細討論。截至目前,各種基于不同原理的技術(shù) [1]被提出,并廣泛應用于許多領(lǐng)域,如 CAD和 CAM,逆向工程,快速原型,虛擬現(xiàn)實,人體工程學和文物保護等 [2, 3]。 對象的空間坐標( XW, YW, ZW)和它們的對應像素坐標( Xf為, YF)之間的映射關(guān)系,可以成從圖像捕獲過程得到的形式如下面等式 (1)所示的齊此坐標配制而成的矩陣方程。 但是,若匹配校準只是簡單地按照上述方法進行計算,將會導致很多錯誤,因為該方法并沒有考慮到鏡頭等非線性因素的失真情況。隱藏層和輸出層的行為模式可以被公式 (4)描述。 3rd International Symposium on Instrumentation Science and Technology Aug. 18~22, 2020, Xi’an. China 1806 其中,η是學習速率,其值應 0 和 1 之間進行選擇。 3D之間的匹配坐標和其顏色信息可通過以下過程來實現(xiàn)。W,彩色 CCD用 8毫米鏡頭、 768 576 的分辨率采樣 CCD,兩者的面積都為一個 7 毫米的鏡頭。圖 3 是物體的彩色圖像 。因此,該方法可以有效解決信息匹配的問題,這為未來的 3D彩色重建和紋理映射奠定了良好的基礎(chǔ)。W CCD to space coordinate (XW, YW). With another coordinate Zw got from scanning system, the whole 3D information of the object can be expressed by (XW, YW, Zw). Furthermore, the mapping relationship between 3D coordinate (XW, YW, Zw) and its corresponding pixel coordinate on color image got from color CCD can be obtained by color sensor calibration. The RGB value of the pixel can be got from the true color image. So, the process mentioned above can realize the matching between 3D information and its corresponding color information. 3rd International Symposium on Instrumentation Science and Technology Aug. 18~22, 2020, Xi’an. China 1812 3. EXPERIMENTAL RESULTS Based on the above theory, coplanar and noncoplanar calibration points are used to calibrate Bamp。 phone: 02227402594。W CCD with a 6 mm lens is sampled by an image capture board with a resolution of 640480 and the color CCD with a 8 mm lens is sampled by a resolution of 768576. The laser is SNF501L670 from LASIRIS Company in Canada, the light is linestructured and the wavelength is 670nm. The Bamp。 camera calibration, information matching, linear partition calibration, BP neural work, color point cloud 1. INTRODUCTION Acquisition of 3D and color information of real object is a research focus in digitization field. Up to now, different kinds of techniques based on different principles are proposed[1], which are widely applied in many fields such as CAD and CAM, reverse engineering, rapid prototype, virtual reality, human engineering and preservation of cultural relics and so on[2, 3]. Among these techniques, noncontact optical method, especially structured light method is bee more and more popular due to its simple principle, rapid measurement, non contact and high precision characteristics. Real object can be digitized with color information by 3D color sensor whose key parts are black and white (Bamp。圖 5 是從顏色傳感器校準以及信息獲得的 3D 彩色點云的匹配過程。 黑白圖象是由大部分在 1 4的形式內(nèi)的電和 4的區(qū)域劃分得到的變換矩陣組成。與另一個坐標 ZW接到掃描系統(tǒng)中,對象的整個三維信息可以由( XW, YW,度 Zw)來表示。 其中, TPJ 是理想的輸出, OPJ 是實際產(chǎn)出。 fj是激勵功能應該是微無處不在 BP神經(jīng)網(wǎng)絡(luò),所以使用 S形函數(shù)總是偏好,如以下公式 (5)所示。由該方法可以得到若干轉(zhuǎn)換矩陣,它們在需要測量時將被用于某些基于區(qū)域劃分分類規(guī)則的數(shù)據(jù)輸入。顯而易見的我們可以從上面的公式 (1)中發(fā)現(xiàn),矩陣 M中包含了所有3rd International Symposium on Instrumentation Science and Technology Aug. 18~22, 2020, Xi’