TWM361902U - Framework with human body posture identification function - Google Patents

Description

M361902 五、新型說明： 【新型所屬之技術領域】 本創作係提供一種具人體姿勢辨識功能之架構，尤指 其技術上提供—種自走車上設有攝影機，攝影機擷取影: 傳至電腦可作處理及判斷人體姿勢者。 【先前技術】M361902 V. New description: [New technology field] This creation department provides a structure with human body posture recognition function, especially the technology provided. There is a camera on the self-propelled car, and the camera captures the image: It can be used to process and judge human body posture. [Prior Art]

按，隨著時代的改變，人們的觀念也漸漸改變，越來 越多的新新人類認為結婚或生育是一種請，而新新人類 都想要鬆過日子’因此’結婚與生育之人數乃年年降低， 造成倒三角形的社會，即是老人很多，而青壯年人π次之， 而新生兒及兒童很少’也就是老人時代的來臨。 老人夕半行動不便或需要人照料，此時如果沒有 多餘的人或經濟能力的不許可’均造成老人沒有人照顧， 沒有人照顧的老人如果不幸發生緊急事件或跌倒，如果沒 有緊急西4 ’冑有可能造成生命的危險，而如果有人隨時 在旁邊看護1又太浪費人力，顧人看護則花費更是非— 般人能力所能負擔。 是以，針對上述習知結構所存在之問題點，如何開發 種更具理想實用性之創新結構，實消費者所殷切企盼， 亦係相關業者須努力研發突破之目標及方向。 有4a於此，創作人本於多年從事相關產品之製造開發 ”又十驗產情上述之目標，詳加設計與審慎評估後， 3 M361902 終得-確具實用性之本創作。 【新型内容】 欲解決之技術問題 X, tL ^ " •老人多半行動不便或需要人照 枓，此日寸如果沒有多餘 ..、的人或經濟能力的不許可，均造成 老人 >又有人照顧，、力 .... 、、、'又有人照顧的老人如果不幸發生緊急事 3跌%果’又有緊急醫護’都有可能造成生命的危險，According to the changes of the times, people's ideas are gradually changing. More and more new people think that marriage or childbearing is a kind of invitation, and new and new human beings want to spend their days. Therefore, the number of people who marry and give birth is the year. The year is reduced, and the society that causes the inverted triangle is that there are many old people, while the young adults are π times, and the newborns and children are rarely 'the advent of the old age. The elderly are inconvenient or need people to take care of them. If there is no extra person or financial ability, it will cause the elderly to be taken care of. If no one cares for the elderly, if there is an emergency or falls, if there is no emergency west 4' It is possible to cause danger to life, and if someone is looking around at any time, it is too wasteful of manpower, and caregiving is more costly than ordinary people can afford. Therefore, in view of the problems existing in the above-mentioned conventional structure, how to develop an innovative structure that is more ideal and practical, the consumers are eagerly awaiting, and the relevant industry must strive to develop the goal and direction of breakthrough. There are 4a here, the creator has been engaged in the manufacture and development of related products for many years, and the above-mentioned objectives of the ten test results, after detailed design and careful evaluation, 3 M361902 will eventually have a practical and practical creation. 】Technical problems to be solved X, tL ^ " • The elderly are mostly inconvenient or need to take care of them. If there is no excess in this day, the person or the financial ability does not permit, the elderly will be caused by others. , force ....,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,

而如果有人隨時㈣邊看護，則又太浪f人力，顧人看護 則花費更是非一船人< + 非叙人迠力所能負擔。 解、門題之技術特點：提供—種具人體姿勢辨識功能 之架構中人體姿勢包括：站、坐、队。而該具人體姿 勢辨識功能之架構可實施成—自走車，該自走車上設有-固定攝影機自走車可代替人力看護，且持續地以固定 攝P機縱m ’本創作是利用無線視訊擷取影像後，將 影像傳送至電腦或自走車自行做判斷，首先制索貝爾測 邊搜尋到影像的邊緣，再利用隸法與水平垂直腐姓法， 將背景去除’進而利用處理過的圖像進行人體姿勢的判 斷，如果有緊急狀況即可立即通知救護，可替代看護及降 低人力成本。 對照先前技術之功效：現今老人分為獨居老人及有看 濩的老人，獨居老人根本請不起看護，所以也隨時處在危 險之中，而有請看護者其花費相當大，且看護只能看顧一 人有浪費人力之情形，本創作該具人體姿勢辨識功能之架 4 M361902 構可實施成一自走車，古女白本生 早边自走車上設有一固定攝影機，玆 自走車可代替人力看謨， 人 且持績地以固定攝影機追蹤辨 識，擷取影像後將影像傳 <王电驷或自走旱自行做判斷， 如果有緊急狀況即可立卽诵4 即通知救護，可替代看護及降低人 力成本者。 一有關本創作所採用之技術、手段及其功效，兹舉一較 佳實施例並配合圖式詳細說明於后，相信本創作上述之目 的、構造及特徵’當可由之得一深入而具體的瞭解。 【實施方式】 參閱第一、第二圖所示，本創作係提供—種具人體姿 勢辨識功能之架構，其包含有： -影像擷取架構（可為自走車（丄〇))，該自走車 (1 0 ) i裝設有—無線視訊鏡頭（丄i )及一影像擷取And if someone cares at any time (four), then it is too arrogant, and the care of the person is more than a boatman. + Non-narrators can afford it. The technical features of the solution and the door: provide the body posture of the human body posture recognition function: standing, sitting, team. The structure with the human body posture recognition function can be implemented as a self-propelled vehicle. The self-propelled vehicle is provided with a fixed camera self-propelled vehicle instead of a human care vehicle, and is continuously fixed by a P-machine. After the video is captured by the wireless video, the image is transmitted to the computer or the self-propelled vehicle to make its own judgment. First, the Sobel edge is searched for the edge of the image, and then the method is used to remove the background and the background is removed. The image is used to judge the posture of the human body. If there is an emergency, the ambulance can be immediately notified, which can replace the nursing and reduce the labor cost. Compared with the efficacy of the prior art: Today's elderly are divided into elderly people living alone and elderly people who have a look at them. The elderly living alone cannot afford to be cared for, so they are always at risk, and there are caregivers who spend a lot of money and care. Can only look after a person who has wasted manpower, this creation of the body posture recognition function of the 4 M361902 structure can be implemented into a self-propelled car, the ancient woman Bai Bensheng early on the self-propelled car with a fixed camera, can be replaced by a self-propelled car Manpower looks at people, and people use the fixed camera to track and identify. After capturing the image, the image is transmitted to the king or the self-driving drought. If there is an emergency, the police can be notified. Alternative care and reduce labor costs. With regard to the techniques, means and functions of the present invention, a preferred embodiment is described in detail with reference to the drawings, and it is believed that the above objects, structures and features of the present invention can be obtained from an in-depth and specific To understanding. [Embodiment] Referring to the first and second figures, the author provides an architecture with a human body gesture recognition function, which includes: - an image capture architecture (which may be a self-propelled vehicle (丄〇)), Self-propelled vehicle (10) i equipped with a wireless video lens (丄i) and an image capture

裝置（12)，該無線視訊鏡頭（11)肖該影像擷取裝 置（1 2 )電訊連結； 一電腦（2 0) ’該電腦（2Q)設有—_無線接收 器（2 1) ’該USB無線接收器（21)得以接收該自走車 (1 〇 ) t該影像擷取裝置（i 2 )之訊號，該電腦（2 0 )接收影像訊號後，利用灰階處理、索貝爾測邊、腐触 法、垂直腐#法、框A人體外框計算斜率，以判斷人體站、 坐或臥姿，計算出人體是队倒而非睡眠時，即發出訊號或 藉一網路通訊或經手機自動撥號請求協助救護。 5 M361902 上述5亥電腦（2 〇)得以直接裝設於該自走車（丄〇) 上。該自走表Γ Ί Π、》The device (12), the wireless video lens (11) is connected to the image capturing device (1 2); a computer (2 0) 'the computer (2Q) is provided with a wireless receiver (2 1) The USB wireless receiver (21) is capable of receiving the signal of the self-propelled vehicle (1 〇) t the image capturing device (i 2 ), and the computer (20) receives the image signal, and uses gray scale processing and Sobel edge measurement. , the method of decaying touch, vertical rot #法, box A human body frame to calculate the slope to determine the body standing, sitting or lying posture, calculate the human body is the team instead of sleeping, that is, send a signal or borrow a network communication or by mobile phone Automatic dialing request to assist in ambulance. 5 M361902 The above 5 Hai computer (2 〇) can be directly installed on the self-propelled vehicle (丄〇). The self-propelled watch Γ Π Π,》

v丄u)上並設有一追蹤系統，該追蹤系統得 以搜尋人體位置進行追蹤跟隨。 " ’像擷取架構為一安裝在房間的固定攝影機。 上述該影像擷取架構為一安裝在房間的網路攝影機。 本創作疋利用無線視訊鏡頭擷取影像後，將影像傳送 至電腦做判斷，首先利用索貝爾測邊搜尋到影像的邊緣， 再利用腐姓法’將背景去除，進而利用處理過的圖像進行 人體姿勢的判斷。 本創作中採用的影像擷取裝置（1 2 )，是先利用無 線擷取的方式做影像抓取’ t後是利用USB無線接收器（2 1 )將影像傳送至電腦（2 0 )。由於是利用無線的關係， 其頻率較容易受到干擾而影響到影像品質，而影像的品質 優劣則會影響到t續影像處王里的複雜度A困難度。 本文提出利用索貝爾測邊與腐蝕法，T以抓出人體外 框’進而利用此外框辨識人體姿勢，且利用精簡的程式做 〜像處理的動作’使其能夠做即時的判斷與發送訊號，達 到即時控制的效果。使用這套系統能夠代替人力看護，使 用此看護系統則能持續的追蹤騎，降低了使用人力的成 本，因此系統運算速度夠快，故能夠架設在自走車（丄〇 ) 上或者固定攝影機上’都能夠達到及時控制及回傳的功能。 影像傳至電腦後要做處理動作，因一開始傳進電腦是彩 6 M361902 色影像，本創作需利用灰階影像處理可以縮短大幅時間，之There is a tracking system on the v丄u), which tracks the body position for tracking. " ' is like a capture camera for a fixed camera installed in a room. The image capture architecture described above is a webcam installed in a room. After the image is captured by the wireless video lens, the image is transmitted to the computer for judgment. First, the edge of the image is searched by Sobel edge detection, and then the background is removed by using the rot method, and then the processed image is used. The judgment of the human body posture. The image capture device (1 2 ) used in this creation is to use the wireless capture method to capture images. Then the USB wireless receiver (2 1 ) is used to transmit images to the computer (2 0 ). Because it is a wireless relationship, its frequency is more susceptible to interference and affects the image quality, and the quality of the image will affect the complexity of the king in the image. This paper proposes to use Sobel edge measurement and corrosion method, T to capture the human body frame and then use the frame to identify the human body posture, and use the streamlined program to do ~ image processing action to enable it to make instant judgment and send signals, Achieve instant control. The use of this system can replace human care. The use of this care system can continuously track the ride and reduce the cost of using manpower. Therefore, the system can be operated fast enough to be mounted on a self-propelled vehicle (丄〇) or on a fixed camera. 'All can achieve the function of timely control and return. After the image is transmitted to the computer, the processing action is required. Since the computer is a color 6 M361902 color image, the creation of the grayscale image processing can shorten the large time.

後再利用索貝爾測邊尋找邊緣後，再利用腐蝕法將多餘的邊 緣去除將人體輪廓表現出來。 兩種深淺度 灰階影像又稱單色影像，指的是只有黑與白 的影像。一般灰度值的範圍如下式所示： 〇</(x,v)<255After using the Sobel edge to find the edge, the excess edge is removed by the etching method to express the contour of the human body. Two shades of grayscale images, also known as monochrome images, refer to images with only black and white. The range of general gray values is as follows: 〇</(x,v)<255

由於衫色影像處理起來比較複雜’如果不是需要判斷 顏色則使用灰階影像較為適合。本創作之影像辨識並非是要 辨識顏色，因此灰階影像是一個很適合本系統使用之色彩槿 型。第二圖中的影像為灰階化後的影像，可以报明顯看出影Because the color of the shirt image is more complicated to handle, it is more suitable to use grayscale images if it is not necessary to judge the color. The image recognition of this creation is not to identify the color, so the grayscale image is a color type that is very suitable for the system. The image in the second picture is a grayscale image, which can be clearly seen.

像中的像素只需要進行〇_255之間的運算，讓系、統運算量減 少’故可以進行快速的運算。 索貝爾測邊是一種士去·、户禮$人ΛΑ 種古老/瓜傳至5的影像邊緣偵測方 法。其運算遮罩有兩種。經利用第三U)圖遮單運算後可得 Gx’再利用第三⑻圖料進行内機運算而得其中代表 水平與垂直方向影像中該點灰度值的最大增”。實社 之索貝爾值GHGx| + |Gy卜利用此遮罩運算的結果如第四圖 所示。索貝爾測邊除了可以用來做影像中物體的邊緣偵測 以外，尚可用來分割物體以及辨識邊緣等。 本創作利用索貝爾邊緣偵測檢測影像中的邊緣’將人 7 丄 體邊緣顯 示出來， 進行判斷。 再利用去背景’將人體獨立出來，進 而 腐蝕是除去二元影像 化的方法。本創作中就^相干的細節，使得影像簡單 像加以腐蝕=_ 疋利用腐蝕法將索貝爾測邊後的影 出來。第五圖=::的背景能夠去除，讓人體突顯 圖，將原、、 圖’是利用3x3的遮罩蓋住原 …象素以遮罩係數侵， 的動作。圖Φ本 以達到將細部部分清除 U為遮罩孫童令， 包含有& # 、 為在遮罩底下的像素，圖中 “有衫像原點（3 〇 影像部分的像辛 、'、 1 )、在遮罩底下 素（3 2)、县目千]^4«^1· 3 )、影像〜）（3 4): 排列的遮罩係數（3 使用腐钱原理做實驗可以得到第六圖結果。 在腐蝕前的圖像較 看出The pixels in the image only need to perform the operation between 〇 255, so that the amount of the system and the system are reduced, so that fast calculation can be performed. Sobel edge detection is a kind of image edge detection method that is used to go to the family, to pay for the person, and to pass the ancient/guy to 5 image. There are two kinds of arithmetic masks. After using the third U) graph to mask the single operation, Gx' can be used to perform the internal computer operation by using the third (8) image, which can represent the maximum increase of the gray value of the point in the horizontal and vertical images. Bell value GHGx| + |Gy Bu The result of using this mask operation is shown in Figure 4. Sobel edge measurement can be used to segment objects and identify edges, in addition to edge detection of objects in images. This creation uses Sobel edge detection to detect the edge in the image to display the edge of the human body 7 and judge it. Then use the background to separate the human body, and then the corrosion is a method to remove the binary imaging. The details of the ^ coherence make the image as simple as corrosion = _ 疋 use the corrosion method to make the shadow of the Sobel edge measurement. The fifth picture =:: the background can be removed, let the body highlight the picture, the original, the picture ' The 3x3 mask is used to cover the original pixel...the mask is invaded by the mask. The figure Φ is used to remove the detail part to cover the Sunchild, including &#, which is the pixel under the mask. In the picture The shirt looks like the origin (3 〇 image part of the image, ', 1), under the mask (3 2), the county thousand] ^ 4 «^1 · 3), image ~) (3 4): Arrange The mask factor (3 using the principle of decay money to do the experiment can get the results of the sixth picture. The image before the corrosion is more visible

除，但還是殘留了二後的圓形將背景幾乎都去 &茂^ 了斗多背景物， 物品的輪廓都是直線邊 '了大多數 、人體的輪廓卻較少有直線邊的現 罩广:利用第七圖的過遽掉水平線與過遽垂直線的避 ，過錢的結果，可以明顯看出，背景物幾乎都已 去除只留下人體輪廓。 已 本創作先以視訊擷取影像，將影像㈣成圖， 索貝爾測邊將影像邊緣以白點晝出來，再利用腐钱法將, 餘的邊緣去除，接著再利用水平與垂直腐料將多餘的2 消除，最後將人體外框標出，進而判斷人體姿勢。 8 M361902 本幻作利用掃描像素點的方法將人體框選住 分為四個點，分別為最頂點、最低點m最^肢 以四點方法將人體框選1描最高點的方法是㈣ 掃描以搜尋筮 μ & 万向 …個白點，從圖像左上的第' 點為起始點， +要找=低點則是以圖像的右下角為起始點往上開始掃 、！尋sj的第—個點為最高或最低點。將掃描方式變 換成Y方向的掃描，以圖像的左上為起始點，掃描到= :個白點為最左點，若以圖像右上為起始點則是最右點， 刀別以34四個點將人體以綠色的框線框選。 盘2線框出人體外框後，求出此方框的斜率，利用斜率 與長見的關係進行判斷，因每人的體型皆不同，故本創作 利用台灣行政院衛生署2007年的平均身高調查，45歲以上 的：年人平均身高& 165·5公分因此本創作利用此平均Except, but still the second round of the circle will almost go to the background of the background, the outline of the item is a straight edge 'most, the contour of the human body is less than the straight edge of the current cover Guang: Using the seventh figure to avoid the horizontal line and the vertical line to avoid, the result of the money, it can be clearly seen that the background has almost been removed to leave only the contours of the human body. This creation has first captured the image by video and the image (4) into a picture. Sobel edged the edge of the image with white dots, and then used the rot method to remove the remaining edges, and then used horizontal and vertical rot. The excess 2 is eliminated, and finally the external body frame is marked to determine the posture of the human body. 8 M361902 This magical method uses the method of scanning pixels to divide the human body frame into four points, which are the highest vertex and the lowest point. The method of selecting the highest point by the four-point method is (4) scanning. To search for 筮μ & universal white point, starting from the first point on the top left of the image, + looking for = low point starts with the bottom right corner of the image as the starting point! Find the first point of sj as the highest or lowest point. Convert the scanning mode to the scanning in the Y direction, starting from the upper left of the image, scanning to =: the white point is the leftmost point, if the upper right point of the image is the rightmost point, the knife is 34 four points to select the human body in a green frame. After the disk 2 frame is out of the frame, the slope of the box is obtained, and the relationship between the slope and the long view is used. Since each person's body type is different, the author uses the average height of the Taiwan Administration's Department of Health in 2007. Survey, over 45 years old: average person height & 165·5 cm, so this creation uses this average

身高做出長寬比的比例。 系統利用綠色框線進行人體姿勢的判斷，利用斜率以 、長寬比刀別判斷出站、坐、臥，sl〇pe為斜率’其坐姿 :斷用斜率在。.8〜"之間’是利用平均身高上下取五公 ^所求得的斜率值，大約是纟U〜1.2之間，而站姿與执 文用1來做判斷的區間，是在斜率在角度45。時值為1，因 站安所晝出的方框其斜率會大於45%且寬乘2後小於長，而 臥安則會小於1且長乘2後小於寬，其判斷公式如下：Height makes the aspect ratio. The system uses the green frame line to judge the posture of the human body, and uses the slope to determine the station, sitting, and lying in the aspect ratio, and slspe is the slope 'the sitting position: the slope is used. .8~"between is the slope value obtained by taking the average height up and down five gong^, which is about 纟U~1.2, and the standing position and the obedience are judged by the interval of 1 and are in the slope. At angle 45. The time value is 1, because the slope of the box that the station has pulled out will be greater than 45% and the width by 2 is less than the length, while the sleep is less than 1 and the length is less than 2 and less than the width. The formula is as follows:

Width x2 < Height 且 slope〉 1 =站 M3 61902 S1〇Pe=l.5〜0·5 之間=坐Width x2 < Height and slope> 1 = station M3 61902 S1〇Pe=l.5~0·5 = sit

Wldth &gt; Helght χ2 且帅〈ι=臥 （2) 的影像評估方法有比較運算時間和它的誤差， 運真時間上當缺是抵，也&amp; ....^ …、越决越好，至於影像誤差主要會因為 光源i衣境會有所矣g ^ ’、 本創作是在室内普通光源下進行 只…下面圖示為電腦判斷的結果：Wldth &gt; Helght χ2 and handsome <ι=卧(2) The image evaluation method has a comparison operation time and its error. When the time is lost, it is also the case, and &amp; ....^ ..., the better, the better. As for the image error, it will be mainly because the light source i will be 矣g ^ ', this creation is carried out under the ordinary indoor light source only... The following figure shows the result of computer judgment:

弟八圖為坐姿的判斷’因長寬比值後算出斜率等於 1 · 432 ’在坐姿斜率範圍内，故判斷為坐姿。 弟九圖為站姿的判斷’可以报明顯的看出，索貝爾測 ^腐餘法去背景後，可將人體準確的表框出I，使用長 寬比的比料較後，長切寬的兩倍且斜率為3 _大於 1 ’故判斷為站姿。 …第十圖為坐姿的判斷，使用長寬比的比例比較後，其 斜率為G.288且寬大於長的兩倍小於卜故判斷為臥姿。 _v&gt;— 則文係針對本創作之較佳實施例為本創作之技術特徵 進行具體之說明’'丨隹’熟悉此項技術之人士當可在不脫離 本創作之精神與原則下對本創作進行變更與修改，而該等 文更/、修改’皆應涵蓋於如下申請專利範圍所界定之範疇 中。 【圖式簡單說明】 第圖.係本創作實施成自走車之立體示意圖。 第一圖.係本創作之電腦其一實施例示意圖。 10 M361902 第三（a)圖··其一運算遮罩圖。 •t - &quot; 第三（b)圖：另一運算遮罩圖。 - 第四圖：係本創作之遮罩運算結果圖。 . 第五圖：係本創作之腐蝕原理遮罩圖。 第六圖：係本創作之腐独遮罩結果圖。 - 第七圖：係本創作之過濾掉水平線與過濾垂直線的遮罩圖。 - 第八圖：係本創作之坐姿的判斷圖。 % 第九圖：係本創作之站姿的判斷圖。 第十圖：係本創作之臥姿的判斷圖。 【主要元件符號說明】 (10)自走車 (1 1 )無線視訊鏡頭 (1 2 )影像擷取裝置 (2 0 )電腦 % ( 2 1 ) USB無線接收器 (3 0 )影像原點 (31)腐姓遮罩 (3 2 )在遮罩底下影像部分的像素 (3 3 )顯示座標排列的遮罩係數 (3 4 )影像 z (X, y) 11The eighth figure of the younger brother is the judgment of sitting posture. 'The slope is equal to 1 · 432 ′ in the range of the sitting posture slope because of the aspect ratio value, so it is judged to be sitting. The figure of the nine figures is the judgment of the standing position. It can be clearly seen that after Sobel's measurement of the residual method of the rot, the accurate frame of the human body can be I, and the ratio of the aspect ratio is later, the length of the cut is longer. It is judged to be standing as if it is twice and the slope is 3 _ greater than 1 '. The tenth figure is the judgment of the sitting posture. After comparing the ratio of the aspect ratio, the slope is G.288 and the width is more than twice the length and less than the length is judged as the lying position. _v&gt;- The text is specifically described for the technical features of the present invention for the preferred embodiment of the creation. ''丨隹' Those who are familiar with the technology can change the creation without departing from the spirit and principle of the creation. And the modifications, and the texts of the texts/modifications shall be covered by the scope defined by the following patent application. [Simple description of the diagram] The first diagram is a three-dimensional diagram of the creation of the self-propelled vehicle. The first figure is a schematic diagram of an embodiment of the computer of the present creation. 10 M361902 Third (a) Figure · One of the calculation masks. • t - &quot; Third (b): Another operational mask map. - Figure 4: The result of the mask operation of this creation. Figure 5: The mask of the corrosion principle of this creation. The sixth picture: the result of the singular mask of the creation. - Figure 7: This is the mask that filters out the horizontal lines and filters the vertical lines. - Figure 8: Judgment of the sitting posture of this creation. % Ninth: A judgment chart of the stance of this creation. The tenth picture: the judgment diagram of the prone position of the creation. [Main component symbol description] (10) Self-propelled car (1 1) Wireless video lens (1 2 ) Image capture device (2 0 ) Computer% ( 2 1 ) USB wireless receiver (3 0 ) Image origin (31 ) The mask of the surname (3 2 ) The pixel (3 3 ) of the image portion under the mask shows the mask coefficient of the coordinate arrangement (3 4 ) image z (X, y) 11

Claims (1)

M361902 六、申請專利範圍： 1 一種具人體姿勢辨識功能之架構，係包含有： -影像擷取架構，該影像擷取架構上楚設有—無線視 訊鏡頭及—影像擷取裝置，該無線視訊鏡㈣該影像擷取 裝置電訊連結；以及， 該USB無線 一電腦，該電腦設有一 USB無線接收器 接收器得以接收該影像擷取架構之該影像擷取裝置之訊 號’該電腦接收影像訊號後，利用灰階處理、索貝爾測邊、 腐蝕法、垂直水平腐蝕法及框出人體外框計算斜率，以判 斷人體站、坐歧姿，計算出人體是臥倒而非睡眠時，胃卩 發出訊號或藉一網路通訊或經手機自動撥號請求協助救 護。 2 .如申請專利範圍第丨項所述之具人體姿勢辨識功 能之架構，其中該影像擷取架構得設成一自走車，該自走 %車具有-追蹤系統，該自走車配合該追縱系統得持續的追 蹤跟隨該人體移動。 3 .如申請專利範圍第2項所述之具人體姿勢辨識功 能之架構，其中該電腦得以直接裝設於該自走車上。 4 .如申请專利範圍第1項所述之具人體姿勢辨識功 能之架構，其中該影像擷取架構為一安裝在房間的固定攝 影機。 5 .如申請專利範圍第i項所述之具人體姿勢辨識功 12 M361902 能之架構，其中該影像擷取架構為一安裝在房間的網路攝 影機。 七、圖式. 如次頁M361902 VI. Scope of Application: 1 An architecture with human body recognition function includes: - an image capture architecture, the image capture architecture is provided with a wireless video lens and an image capture device, the wireless video Mirror (4) the image capture device telecommunications connection; and the USB wireless computer, the computer is provided with a USB wireless receiver receiver capable of receiving the image capture device signal of the image capture architecture, after the computer receives the image signal Using gray scale processing, Sobel edge measurement, corrosion method, vertical horizontal corrosion method and frame out of the human body frame to calculate the slope to judge the human body, sitting in a distracted posture, and calculating that the human body is lying down instead of sleeping, the stomach cramps are issued. The signal can be assisted by a network communication or by automatic dialing request by mobile phone. 2. The structure of the human body gesture recognition function as described in the scope of the patent application, wherein the image capture architecture is configured as a self-propelled vehicle, and the self-propelled vehicle has a tracking system, and the self-propelled vehicle cooperates with the self-propelled vehicle. The tracking system continues to follow the movement of the body. 3. The structure of the human body posture recognition function as described in claim 2, wherein the computer can be directly mounted on the self-propelled vehicle. 4. The structure of the human body gesture recognition function as described in claim 1, wherein the image capture architecture is a fixed camera installed in a room. 5. The structure of the human body gesture recognition function as described in claim i, wherein the image capture architecture is a network camera installed in the room. Seven, the pattern. If the next page