M443156 101年9月14日修正替換頁 五、新型說明·· 1--— 一 【新型所屬之技術領域】 本創作為種結合RGB-D攝影機例如KINECT的量測身體或 物體三維尺寸的裝置，特別是指KINECT放置於旋轉平台前方約一 到兩公尺處的適當方向位置，使KINECT可以連續拍攝受測者儲 存三維座標餘，每秒3G張，藉由讎上的標記可以辨認受測者 被拍攝時的旋轉角，之後，可以由旋轉角進行雲點座標轉換，將 鲁 角度的曲面合成身體三維尺寸。 【先前技術】 習知’傳統人體或物體三維重建技術是利用雷射掃描的方 式，雷射掃礙雷射級㈣—條雷射線在人體或物體上由下至 上^吏用具有雷射光親的攝影機捕抓計算出每個雷射點的三維 座I再將所有點貼合加以呈現，但是因為雷射掃描需要較高的M443156 Correction and replacement page on September 14, 101. V. New description·· 1--- A new technical field of the invention. This creation is a device for measuring the three-dimensional size of a body or object in combination with an RGB-D camera such as KINECT. In particular, the KINECT is placed in the appropriate direction position about one to two meters in front of the rotating platform, so that the KINECT can continuously record the three-dimensional coordinates of the subject, 3G per second, and the subject can be identified by the mark on the cymbal. The rotation angle at the time of shooting, after which the cloud point coordinate conversion can be performed by the rotation angle, and the curved surface of the Lu angle is combined into the three-dimensional size of the body. [Prior Art] Conventional 'Three-dimensional reconstruction of human body or object is a laser scanning method. The laser sweeps the laser level (4) - the thunder beam is on the human body or object from the bottom to the top. The camera captures the three-dimensional seat I of each laser point and then puts all the points together for presentation, but because the laser scanning needs a higher
$知’ RGB-D攝影機本身就可以掃描整面物體的三維資訊， 使用方式是彻紅外線投絲投射出紅外線於物體上在心線 ，影機捕抓紅外線反射的點純計算三維魅，但這樣的掃描為 早面二維雲點1要全身三維尺寸模型的話，就要_ 8台:上 的=GB-D攝影機擺設在不同角度下拍攝，拍攝距離大約需要 兩公尺，故整體拍攝站地A約需要四如平方公尺的面積 的拍攝需要足夠的場地和8台攝影機與4台電腦以上的成本。7 3 基於上述這些缺點，可知目前的習知技術，對於傳統31)掃描 機所存在問題就是佔很大的空間而且須高成本設備。 本創作主要在於提供一有效的方法，透過RGB-D攝影器材和 一旋轉平台組合’拍攝人體或物體同時拍攝旋轉平台上的定位標 記，定位標記上的特徵點對應實際之深度距離，藉由此深度值轉 換成- RGB-D攝影機為原點的三維座標，特徵點的三維座標可叶 ^出拍攝時旋轉平台的轉角，此轉角與當時拍攝的人體或物 體行座標轉換，將各角度取得的身體曲面合成完整的身 【新型内容】 裝置如圖脇2雜是_種取得影像找鏡度資訊 又0，疋、、，口 δ 一彩色攝影機2與一紅外魂摄玛機s 建异T以快棘得影像深度資訊。 本創作將RGB-D攝影機結合於= 是有旋棘羊A nX —'維知Γ描系統’特] 括或物體原地旋轉-圈。本創作 -RGB-D攝影機例如KINECT，_旋棘半Α， 及一電腦。其中該RGB_D攝影機 灰轉千口，， 轉平台的適當方向位置，以擷取受: 可不捉轉千台即時影像與三維 資訊和平台上的Μ標記’利用定的 過三角函齡电认1 I 5己的二維座標ί 气點餘疋公式計算轉盤爽角，將夾角斑三: -.·订座標轉換後角度的曲面合成身體三維尺^ M443156 【實施方式】 1〇1年9月14曰修正替換頁 ..將一種.D攝影機放置在與受拍攝人物或物體間固定的距 2彻RGB D攝職的紅外線舰^制深度影像，藉此深度 影像轉換成三維雲點儲存，拍攝人體或物體多角度的曲^儲存= 維雲點’讀4合人體或物體的三維尺寸。而為了拍攝多角度: 面，使用-旋轉平台如圖2，包括平面板4、鐵柱5和驅動馬達6, 當馬達6啟動㈣平面板4轉動。在平面板4貼上定位標記，再 # 藉由RGB-D攝影機中_色影像去_平面板上定位標記與機台 的固定基準點’ _出的兩點轉換成三維座標，兩點三維座標代 入餘奴公式計算出夾角。最後將每個取像的夾角與儲存雲點在經 過座彳示轉換，就能將人體或物體的三維尺寸重建回原樣。 如圖3所示，將rgb-D攝影機放置在旋轉平台前約丨到2公 尺’利用RGB-D攝影機之彩色攝影機每秒30張資訊。平台上有三 個不同標記71、72和73,三個不同標記是為了區分一圈三百六十 φ 度，分別範圍為第一個標記71涵蓋零度到一百二十度、第二個標 記72涵蓋一百二十度到兩百四十度、第三個標記73涵蓋兩百四 十度到三百六十度’此方法是可以避免單一標記被受測者擋住而 偵測不到。計算角度9的步驟如下：先只偵測畫面上圓盤四分之 一前端的區域，可能偵測到的標記71，、72, 73。最接近攝影機的 標記點與固定基準點8兩點晝素座標經過RGB-D攝影機轉換成三 維座標’利用兩點三維座標_經過餘旋公式計算出角度9」.假如是第 一個標記71，角度9就是由第一個標記71與固定基準點8兩點計 5 M443156 101年9月14日修正替換頁 异得出’如果是第二個標記72，就是第固 8一兩點計算的角度9再加上⑽度，要是第三個標記73，就是第 -個U 73與固定基準點8賴計算的角度9再加上⑽度。經 過上述_程就能即時知道轉盤轉到哪個角度，以方便三刺續 疊合時的計算。 圖4所示，為兩台RGB_D攝影機拍攝受測者示意圖。整 個系統若是彻兩台RGB娜雜1G與丨台電腦13連結來取得 影像’兩台·-D攝频的取·完全畔，可以大幅增加解析 度，也可縮短攝影機與受測者π的拍攝距離，讓兩台以上rgb_d 攝影機1G分別設置於上下段，拍攝受測者上半部與下半部的身 軀。量測的方法包括下列步驟： .步驟- ’系統啟動時’先進行校正，取得旋轉平台的中心座知' RGB-D camera itself can scan the three-dimensional information of the whole surface, the way is to use infrared ray to project infrared rays on the object in the heart line, the camera captures the infrared reflection point purely calculated three-dimensional charm, but this Scanning for the early 2D cloud point 1 to the full body 3D size model, it is necessary to _ 8 sets: the upper = GB-D camera is placed at different angles, the shooting distance takes about two meters, so the overall shooting station A Approximately four square meters of area required to shoot enough space and cost of 8 cameras and 4 computers. 7 3 Based on these shortcomings, it can be seen that the current conventional technology has a large space for the conventional 31) scanner and requires high cost equipment. The main purpose of this creation is to provide an effective method for capturing a human body or an object through a combination of RGB-D photographic equipment and a rotating platform while simultaneously photographing the positioning marks on the rotating platform, and the feature points on the positioning marks correspond to the actual depth distances. The depth value is converted into - the RGB-D camera is the three-dimensional coordinates of the origin, and the three-dimensional coordinates of the feature point can be used to convert the rotation angle of the rotating platform when shooting, and the rotation angle is converted with the coordinates of the human body or object taken at that time, and the angle is obtained. Body surface synthesis complete body [new content] Device as shown in Figure 2 Miscellaneous is _ kind of image to find the mirror information and 0, 疋,,, port δ a color camera 2 and an infrared soul camera s built a different T Fast image depth information. This creation combines the RGB-D camera with = a spine sheep A nX - 'visual scanning system' or the object's in-situ rotation - circle. This creation - RGB-D camera such as KINECT, _ spine half a scorpion, and a computer. Among them, the RGB_D camera is grayed out for thousands of passes, and the appropriate position of the platform is transferred to capture: Can not capture thousands of real-time images and three-dimensional information and the mark on the platform 'Using the fixed triangle to recognize the age 1 I 5 own two-dimensional coordinates ί gas point Yu 疋 formula to calculate the turntable cool angle, will be the angle of the spot three: -.. After the coordinate conversion of the angle of the surface of the composite body three-dimensional rule ^ M443156 [Embodiment] 1〇1 September 14曰Correction replacement page: Place a .D camera in an infrared ship's depth image fixed by 2 RGB D from the subject or object being photographed, thereby converting the depth image into 3D cloud point storage, shooting the human body or Multi-angle song storage of objects = Dimensional cloud point 'Read 4' the three-dimensional size of the human body or object. In order to shoot multiple angles: face, use-rotation platform as shown in Fig. 2, including plane plate 4, iron column 5 and drive motor 6, when motor 6 starts (four) plane plate 4 rotates. Positioning marks are placed on the flat plate 4, and then two points are converted into three-dimensional coordinates by two-point coordinates of the _ color image on the RGB image camera and the fixed reference point of the machine table. Substitute the Yunu formula to calculate the angle. Finally, the angle between each image and the storage cloud point can be reconstructed from the three-dimensional size of the human body or object. As shown in Figure 3, the rgb-D camera was placed approximately 2 meters in front of the rotating platform. Using a color camera of the RGB-D camera, 30 pieces of information per second. There are three different markers 71, 72 and 73 on the platform. The three different markers are used to distinguish one circle of three hundred and sixty degrees, ranging from zero to one hundred twenty degrees for the first marker 71 and 72 for the second marker. Covering one hundred and twenty degrees to two hundred and forty degrees, the third mark 73 covers two hundred and forty degrees to three hundred and sixty degrees. This method can prevent a single mark from being blocked by the subject and cannot be detected. The procedure for calculating the angle 9 is as follows: First, only the area of the front end of the disc on the screen is detected, and the marks 71, 72, 73 may be detected. The point closest to the camera and the fixed reference point 8 are converted to a three-dimensional coordinate by the RGB-D camera. 'Using two-point three-dimensional coordinates _ through the co-rotation formula to calculate the angle 9". If it is the first mark 71, The angle 9 is determined by the first mark 71 and the fixed reference point 8 5 M443156 The revised page of September 14, 2011 is different. If it is the second mark 72, it is the angle calculated by the 8th point. 9 plus (10) degrees, if the third mark 73, is the angle 9 calculated by the first U 73 and the fixed reference point 8 plus (10) degrees. After the above _-step, you can instantly know which angle the turntable is to be used to facilitate the calculation of the three-spinning overlap. Figure 4 shows a schematic diagram of the subject being photographed for two RGB_D cameras. If the entire system is connected to two RGB Naza 1G and the Taiwan computer 13 to obtain the image 'two ·-D video capture · complete, can greatly increase the resolution, can also shorten the camera and the subject π shooting For the distance, two or more rgb_d cameras 1G are placed in the upper and lower sections, and the upper and lower half of the subject are photographed. The measurement method includes the following steps: . Step - ' When the system is started', the calibration is performed first, and the center of the rotating platform is obtained.
步驟二’受測者u站在旋轉平台12上保持不動，讓旋轉平 ^ 12旋轉—圈’兩則㈣攝影機個別取得錄影像儲存於電腦 後，受測者11即可離開旋轉平台12。 ς二’電腦接_至少複紐影像，複數張影像較佳是 刀度’例如三張影像，每張間隔120度，取出分別加以處 理’相對於下_ R㈣攝频所取得的影像的，因為同時可 2得旋轉台上的標記’藉由不同的標記，分別測出零度到一百 :度、—百二十度到兩百四十度、兩百四十度到三百六十产的 範圍。有了每張影像拍攝日_度，就可以根魏轉平台的中心 6 M443156 1〇1年9月14日修正替換頁 棘’縣縣像取得的三維無，依 還原成相對於旋轉平台中心座標的三維雲點，之後合In step 2, the subject u stands still on the rotating platform 12, and the rotation is rotated by 12 rotations - the circle 'two' (4). After the camera separately records the image and stores it in the computer, the subject 11 can leave the rotating platform 12. ς二' computer connection _ at least complex image, a plurality of images is preferably a knife degree 'for example, three images, each interval 120 degrees, take out the image separately processed relative to the lower _ R (four) camera, because At the same time, the mark on the rotating table can be measured by using different marks to measure from zero to one hundred degrees, from one hundred and twenty degrees to two hundred and forty degrees, two hundred and forty degrees to three hundred and sixty. range. With each image shooting day _ degrees, you can root the center of the platform 6 M443156 1 〇 1 September 14 revision of the replacement page thorn ' County 3D no image, according to the coordinates of the center of the rotating platform 3D cloud point, after
三錐Μ刑。 巧V 步驟四，由於上半身㈣像_與下半相步取得， 模型的合成可_下半身模型合成_序進行； ^五，糊撕位(registmiQ侧技術，將 下半身的二維模型加以融合成受測者上下身的三 實施例一 、 如圖4所示’於三維掃描系統上，受測者u站在旋轉平二 =上’讓兩台以上RGB_D嶋1()她胁上下段，拍攝受測 者上+部與下半部的身躺，將受測者的三維資訊儲存，再斑叶算 2角度-起進行座標轉換，人_三軟寸觀完整重建原樣 f附件ΤΓ附件顯相4中較财11三維重建後的正面與 ΚΙΝ:=Γ:轉45度的三賴形。雖然這個實施例以上下兩台 KINECT為基礎，但並非以此為限，其實兩台以上也可以。 概算本實施例所需成本只要約美金—千元以下。占地約貺 =5⑽即可。在，_姻合中，_了及其電 以嵌入於牆面，峻轉台可獨立軸，需要時在 要*之，本糾提供_種量科β三軟 ::=」:接於一_，一旋轉盤;第-台心 。用於取得腰部以下的人體三維尺寸，將第-與第 M443156 二台KINECT合成。受測者將兩手貼於頭部，在KINECT*i^_ ^-^ • 至二公尺之間，站在—旋轉盤上，讓旋轉盤由馬達驅動自動繞一 圈，使KINECT可以連續拍攝受測者，每秒3〇張，藉由旋轉盤的 定位標記可以辨認受測者被拍攝時的旋轉角，之後，可以由旋轉 角進行座標轉換，將各角度的曲面合成身體三維尺寸。 . 軸本發明僅以身體三維尺寸為例說明，不難想知，掃猫同 時也可以取得受測者身高’如圖4所示，進—步將體脂肪計2q放 # 置於旋轉平台上，可同時取得身體三維外型以及體重、體脂肪， 有助於針對身體過胖之處，提供脂肪碰的建議。該體脂肪計2〇 可以配備藍牙等無線通訊魏與本發明之電腦無線連結，避免旋 轉平台對有線連結的電線造成纏繞的困擾。 紅上所述，三維掃描系統雖然只是以人體為例，但是不以此 為限’三維掃描系統可以掃描各種器具、動物、植物等等物體。 對於會成長_働，受财透過糊作可峨察比較該動植物 • 經過一段時間其身軀三維尺寸的變化。 雖然本創作已以一較佳實施例揭露如上，然其並非 用以限定本創作，任何熟習此技藝者，在不脫離本創作 之精神和範圍内，當可作各種之更動與潤飾，因此本創 作之保濩範圍當視後附之申請專利範圍所界定者為準。 8 M443156 _ 101年9月14日修正替換頁 【圖式簡單說明】 - -· 圖1 RGB-D攝影機包含彩色攝影機、紅外線攝影機和紅 外線投光器示意圖。 圖2旋轉平台結合馬達整體示意圖。 ' 圖3利用RGB-D攝影機偵測旋轉平台所旋轉角度示意 - 圖。 圖4兩台RGB-D攝影機拍攝受測者示意圖。 【主要元件符號說明】 8. 固定基準點 9. 角度 10. RGB-D攝影機 11. 受測者 12. 旋轉平台 13. 電腦 20.體脂肪計 1. 紅外線投光器 2. 彩色攝影機 3. 紅外線攝影機 4. 平面板 5. 鐵柱 6. 馬達 71. 第一個標記 72. 第二個標記 73. 第三個標記Three cones gauntlet. Q Step V, because the upper body (four) image _ and the lower half of the phase, the synthesis of the model can be _ lower body model synthesis _ order; ^ five, paste tear position (registmiQ side technology, the second half of the lower body model is merged into The third embodiment of the upper and lower body of the tester is as shown in Fig. 4. On the three-dimensional scanning system, the subject u stands on the rotation level = upper one, so that two or more RGB_D嶋1() are threatened by the upper and lower sections, and the shooting is subject to On the tester, the + part and the lower part of the body are lying, the three-dimensional information of the subject is stored, and then the variegated leaves are counted as 2 angles - the coordinates are converted, and the human _ three soft inch view is completely reconstructed as the original f attachment ΤΓ attachment appearance 4 The positive and negative after the 11-dimensional reconstruction of the Chinese Finance: =Γ: The three-dimensional shape of 45 degrees. Although the above two KINECTs are based on this embodiment, it is not limited to this. In fact, more than two can be used. The cost required in this embodiment is only about US$-1000. The land area is about 5=5(10). In the _ marriage, _ and its electricity are embedded in the wall, and the turret can be independent. When needed, To be *, this correction provides _ seed volume section beta three soft:: =": connected to a _, a rotating disk; For obtaining the three-dimensional size of the human body below the waist, the first and the M443156 two KINECT are combined. The subject puts both hands on the head, between KINECT*i^_ ^-^ • to two meters, standing On the rotating disc, let the rotating disc be automatically driven by the motor to make a circle, so that the KINECT can continuously shoot the subject, 3 每秒 per second, and the rotation angle of the rotating disc can recognize the rotation angle of the subject when shooting. After that, the coordinate transformation can be performed by the rotation angle, and the curved surface of each angle can be combined into the three-dimensional size of the body. The present invention is only described by taking the three-dimensional size of the body as an example, and it is not difficult to know that the sweeping cat can also obtain the height of the subject. As shown in Figure 4, the body fat meter 2q put # on the rotating platform, can simultaneously obtain the body's three-dimensional appearance and weight, body fat, help to provide a fat touch for the body overweight The body fat meter 2 can be equipped with wireless communication such as Bluetooth and the wireless connection of the computer of the present invention, so as to avoid the trouble that the rotating platform causes entanglement of the wired connecting wires. Red, although the three-dimensional scanning system is only for the human body , but not limited to this, 'three-dimensional scanning system can scan a variety of objects, animals, plants and other objects. For growth _ 働, by the money through the paste can be compared to the animal and plant • After a period of three-dimensional changes in the body Although the present invention has been described above in a preferred embodiment, it is not intended to limit the present invention, and any person skilled in the art can make various changes and refinements without departing from the spirit and scope of the present invention. The scope of this creation is subject to the definition of the scope of the patent application. 8 M443156 _ September 14, 2011 revised replacement page [simple description] - -· Figure 1 RGB-D camera includes color camera , infrared camera and infrared light projector schematic. Figure 2 is a schematic view of the rotating platform combined with the motor. Figure 3 shows the angle of rotation of the rotating platform using an RGB-D camera. Figure 4 shows two subjects of the RGB-D camera. [Main component symbol description] 8. Fixed reference point 9. Angle 10. RGB-D camera 11. Subject 12. Rotating platform 13. Computer 20. Body fat meter 1. Infrared light projector 2. Color camera 3. Infrared camera 4 Flat plate 5. Iron column 6. Motor 71. First mark 72. Second mark 73. Third mark