TWI241535B - Focusing method for a moving object - Google Patents
Focusing method for a moving object Download PDFInfo
- Publication number
- TWI241535B TWI241535B TW90132874A TW90132874A TWI241535B TW I241535 B TWI241535 B TW I241535B TW 90132874 A TW90132874 A TW 90132874A TW 90132874 A TW90132874 A TW 90132874A TW I241535 B TWI241535 B TW I241535B
- Authority
- TW
- Taiwan
- Prior art keywords
- image
- template
- moving object
- histogram
- focusing
- Prior art date
Links
Abstract
Description
1241535 ^ 五、發明說明（1) 【發明之應用領域】 、 本發明係關於一種移動物體的 運用模板匹配技術來達到跟縱移:銘叙特別是一種 方法。 物版之移動物體的聚焦 【發明背景】 P返著多媒體通訊技術的發 ， & 在電腦功能越來越強大的條件許可恶冷的影像處理， 九領域。而動態物體的影像，、^成為熱門的研 呈現為動態影像(FuU—M丄；二現都以序列圖像的方式 處理逐漸成為影像處理技術：：焦^。0)’於是，序列圖像 由於序列圖像包含了一 $ 的目標物進行辨識成A s =、圖像，對這些圖像當中 針對序列圖ΐ像處理的重點。而其應用上，可 應用在許多1抒、目軲進行實際分割、識別與跟蹤，而可 缩、生產ί:;，例如：自動監視、體育報導、視頻壓 此f卜線广以及智能交通管理等。 最新的視頻圖2 η處理技術當中的目標跟蹤技術也成為 ㈣士正^標準4當中的關鍵技術，其中 的圖像操作。因用1目橾跟蹤和分割技術，提供了基於内容 各項應用且右^ t ’對這項技術進行深入的研究，開發其 /、頁冰运的意義。 目月1】’已細右今n夕 較為廣泛的是5午夕目標識別與跟蹤算法，其中應用得 matching) ^^^於^圖像特徵的模板匹配（Pattern 前圖像特徵和新异法。這一算法主要是利用所提取的當 ?跟^目標模板的特徵，兩者加以比對，而1241535 ^ V. Description of the invention (1) [Application field of the invention] The present invention relates to a moving object using template matching technology to achieve vertical movement: inscription is especially a method. Focusing on moving objects in the object version [Background of the Invention] Back to the development of multimedia communication technology, in the field of increasingly powerful computers, permitting cold image processing, nine fields. The image of a dynamic object has become a popular research and presented as a dynamic image (FuU-M 丄; both are now processed as a sequence of images and gradually become an image processing technology :: focus ^ .0) Because sequence images contain a $ target and are identified as A s =, images, the focus of these images on the processing of sequence image artifacts. And its application can be applied to many real-time segmentation, identification and tracking, but can be scaled down and produced; for example: automatic monitoring, sports reports, video downloads, and intelligent traffic management Wait. The latest video Figure 2 The target tracking technology in the η processing technology has also become the key technology in the standard 4 of the fighters, including the image manipulation. Due to the use of 1-track tracking and segmentation technology, it provides content-based applications and in-depth research on this technology to develop its significance. Eye month 1] 'The most widely used is the 5 pm target recognition and tracking algorithm, which is used for matching) ^^^ ^ Template matching for image features (pre-pattern image features and new exotic methods This algorithm mainly uses the extracted features of the target template and the target template to compare them, and
第4頁 1241535 五、發明說明（2) 採用的特徵通常是圖像的邊緣特徵。進行目標模板與當前 圖像之間的匹配，並在區配的過程之中完成模板的更新工 作。其中，基於Hausdorff距離的模板匹配方法是最近發 ^起來的一種效果較好的方法。但由於採用這種演算法 1，匹配過程中的搜索空間會隨著圖像大小和模板大小的 曰力而急遽增大，進而導致搜索所需的記憶體盥哭 ”腦負荷大幅增…外，圖像特徵的提取並；是二】 ::: :作，#圖像大小與模板大小增加時，即會導致搜 索一跟蹤的效率降低，很難達到實際處理的要求。 禮，如何能在目標跟蹤上，能讓跟蹤的結果更加準 亚犯貫現自動跟蹤，且能提高跟蹤的 研究者所亟欲解決的研究課題。 …莫不成為 【發明之目的及概述】 蓉於以上^知技術的問題，本發明&日+ 種移動物體的聚焦方法，利用模板=、目=於提供- 的自動跟蹤與拍攝。 -末達成對移動物體 本發明的另一個目的在於提供一 法，利用序列圖像當中的灰度直；圖：體1聚焦方 資訊：來達到目標物辨認以及模板更新的圖投影圖的 為了達到上述目的，本發明揭一、 焦方法，應、用-攝影設備攝取一移 ：：物體的聚 攝影設備透過-模板匹配方式加 ：：衫像，並將該 步驟：設定-最佳匹配模板位置；操；：：：，包含下列 像序列；依據該圖像序列取得該 2動物體之-圖 刃肢之一初始模板並Page 4 1241535 V. Description of the invention (2) The features used are usually the edge features of the image. Match the target template with the current image, and update the template during the matching process. Among them, the template matching method based on the Hausdorff distance is a better method recently developed. However, with this algorithm 1, the search space during the matching process will increase sharply with the power of the image size and the template size, which will cause the memory required for searching to cry. The extraction of image features is two] :::: 作, # When the image size and template size increase, it will cause the search-tracking efficiency to decrease, and it is difficult to meet the actual processing requirements. On tracking, the results of tracking can be more accurate and automatic tracking can be realized, and it can improve the research problems that the tracking researchers urgently want to solve.... It will be [the purpose and summary of the invention]. The present invention & Japanese + a method for focusing on a moving object, using automatic tracking and shooting of template =, mesh = provided by-. Finally, a moving object is achieved. Another object of the present invention is to provide a method using sequence images The gray level is straight; Figure: Volume 1 Focusing Party Information: To achieve the target object identification and template update map projection. In order to achieve the above purpose, the present invention discloses a method for focusing, which should be used-photographed. The device captures one move :: the object's poly photography device through-template matching method plus :: shirt image, and this step: set-the best matching template position; operation; ::: contains the following image sequence; according to the image Sequentially obtain the initial template of one of the two animal body-graph blade limbs and
第5頁 I241535Page 5 I241535
發明說明（3) 後得—初始模板位置；及 — …板位置以調整該攝影設：二置與該最佳 此外，本發明更揭霖—拍攝角度。 :下列步驟：擷取該移動物：：體的聚焦方法’其包 ^依據該圖像序列進：模=-初始模板位 取佳匹配位置；依據該最佳匹配位c移動物體之-—更新模板並獲得一更新模板位署置以取得該移動物體之 位置進行調整 P 、板位置，及，依據該更新模板 发 71正3攝衫没備之一拍攝角度。 -圖像據下Γ驟而得：掏取該圖像序列之 方圖投影圖十二亥圖像之：灰度直方圖；計算該圖像之一直 取得該初#模板=據錢度直方圖與該直方圖投影圖， 口振板與该初始模板位置。 詳細月的特徵與實[兹配合圖示作最佳實施例 【發明之詳細說明】 圖像基礎的影像擷取上，攝影機-方面是以 影 丄式來產生圖像，另一方面則可以透過調整攝 整攝影；透過處理圖像序列來達到直接調 、，风的攝影角度的目的。 與跟;m明採用了一種基於灰度直方圖的目標識別 幅；r圖平二=垂直投影。同時’綜合考慮到運動預測和 /、固之間）的相似性，以確定目標的位置。此外，Description of the invention (3) The following is obtained-the initial template position; and ... the plate position to adjust the photographic device: two sets and the best In addition, the present invention further reveals-the shooting angle. : The following steps: capture the moving object: the focus method of the volume 'its package ^ according to the image sequence: modul =-initial template position to take the best matching position; move the object according to the best matching position-update The template is obtained and an updated template position is set to obtain the position of the moving object to adjust the position of P and the board, and according to the updated template, one of the shooting angles is 71. -The image is obtained according to the following steps: Extract the histogram of the image sequence, the projection image of the twelve-hine image: the grayscale histogram; calculate the image all the time to obtain the initial #template = according to the degree of histogram With the histogram projection, the mouth plate and the initial template position. The characteristics and realities of the detailed month [The best embodiment is described with the illustration [Detailed description of the invention] On the image-based image capture, the camera-on the one hand, generates the image in a shadow mode, on the other hand, it can be transmitted through Adjust photography; adjust the shooting angle by directly processing the image sequence. And follow; m Ming uses a target recognition frame based on gray histogram; r picture flat two = vertical projection. At the same time, the similarity between motion prediction and / or solidity is comprehensively considered to determine the position of the target. In addition,
第6頁 在獲得圖像的灰度直方投影圖的基礎上，來Page 6 Based on the gray histogram of the image,
1241535 五、發明說明（4) 本發明更結合模板更新八 、 的模板，因而可實現對目尸==適應地調整跟蹤目標 、 、凡對目铱的自動跟蹤。 為了使跟縱的結果更加準確， 強的預處理，以提高目巧盥北旦 毛月另刼用了局部增 發明亦採用了兩個衡量r柘：厅、f曰的對比t。此外，本 別與跟蹤的精確度。亦即 ：柘來提高目標識 方法，使本發明的方法對突以； = J =更新與匹配的 性。由於灰度直方圖报容 :、 的即時反應 比較少，實驗證明可幸I南斤以，本發明的計算量 a』以達到實際處理的要求。 成·· s -卜、dr被跟蹤物體的運動分為兩個部分組 變。這兩部分ϊ動：j 第二、物體幾何形狀的改 性變形，由於視角變化：空間内中物體的非剛 物體的幾何形狀與；！，化等去都可以歸人 $ &久#& 、， 頦之中。在延裡，可考慮被跟蹤目 "'.. ，亚假定物體幾何形狀的變化是連續而緩慢 二=設在通常的情況下是正確的，本發明的目標識 別入自動跟蹤，其具體的操作如下所述： 本,與跟蹤主要包含兩部分，一是模板匹 圖傻在獲得初始模板後，在以後的每-幅 二Η傻：肉^ Ϊ ^目標的最佳匹配位置’找到後，利用該 這樣模板匹配和模板更：：作=行下—幅圖像匹配， 像序列的目標識別盥跟：工進、直到完成整個圖 = 跟琰工作。在匹配中，充分利用圖像 ' σ心Μ後得被跟蹤目標的潛在匹配點，輔以運動預 12415351241535 V. Description of the invention (4) The present invention further updates the template of 、 and 结合 in combination with the template, so that the automatic tracking of the target iris can be realized by adaptively adjusting the tracking target. In order to make the results more accurate, the strong pre-treatment to improve the eyesight. Mao Yue also used a local increase. The invention also used two measures r 柘: hall, f, contrast t. In addition, the accuracy and accuracy of tracking. That is: 柘 to improve the target identification method, so that the method of the present invention can be used to highlight; = J = update and matching. Because the gray-scale histogram reports a relatively small amount of instant response, experiments have shown that it is fortunate that the amount of calculations in the present invention can meet the requirements of actual processing. The motion of the tracked object is divided into two parts. These two parts move: j Second, the modified deformation of the object's geometry, due to the change in perspective: the geometry of the non-rigid object in the space; , Hua, etc. can be attributed to $ & 久 # & In Yanli, the tracked object can be considered. "Assumes that the change in the geometry of the object is continuous and slow. 2 = It is correct to set it under normal circumstances. The target recognition of the present invention is automatically tracked, and its specific The operation is as follows: This and tracking mainly consist of two parts. The first is that the template matches the image. After obtaining the initial template, every subsequent frame will be stupid: Meat ^ Ϊ ^ After the best matching position of the target is found, Use this template matching and template more :: make = line below-image matching, target recognition of image sequences: work until the entire picture is completed = follow up work. In matching, make full use of the potential matching points of the tracked target after the image 'σ heart M is supplemented by the motion pre- 1241535
1241535 玉、發明說明（6) 〜 *--- 圖其包含了下列步驟：擷取一幅圖像（步驟1 2 1 )、計算 灰度直方圖（步驟122)、計算直方圖投影圖（步驟Mg)與取 得模板（步驟1 2 4 )。 ’ 只要從圖像序列當中擷取一幅圖像（步驟121)，即可 依…、圖像§中所有的資訊來計算灰度直方圖（步驟1 2 2 )， ，灰度直^圖的兩個峰值對應了物體與背景的位置；接 者’再計算直方圖投影圖（步驟123)，即可對應出物體的 位置，進而可取得模板（步驟丨2 4 )。 事貫上，本發明亦可用另一種方式來說明本發明的聚 ί方ί，請參：「第3圖」，本發明之移動物體的聚焦方 〉之弟二具體實施例流程圖，其包含下列步驟：擷取圖像 歹J (步驟3 1 0 )、取得初始模板（步驟3 2 〇 )、尋找最佳四配 =，（步驟3 3 0 )、取得更新匹配模板（步驟34〇)、調整拍攝 又（步「驟πο)與是否為最後一幅圖像？（步驟36〇)。 、，「第3圖」的流程中，更進一步第揭露了本發明如 :::核板匹配的方式來作為本發明的聚焦方〉去。首先，同 過攝影機來擷取圖像序列（步驟31〇)，所擷取到的圖 =即為杈板取的的基礎。接著，必須先取得目標物的 =、十° ^板，此即步驟32(3，模板的取得方式肖「第2圖」所 ::不再說明。取得初始模板後，即以此處使模 $田二吴板匹配的比對基礎，於接下來的圖像當中，來尋 =土匹配位置（步驟33〇)，也就是，利用初始模板來進 r^匹配的動作。在模板匹配的過程當巾，可運用初始 吴"十出圖像序列當中的下一幅圖像的目標物位置，此1241535 Jade, description of the invention (6) ~ * --- The picture contains the following steps: capture an image (step 1 2 1), calculate the gray level histogram (step 122), calculate the histogram projection image (step Mg) and obtaining the template (steps 1 2 4). 'As long as an image is taken from the image sequence (step 121), the gray level histogram (step 1 2 2) can be calculated based on all the information in the image § (step 1 2 2), The two peaks correspond to the positions of the object and the background; the receiver 'then calculates the histogram projection (step 123), which can correspond to the position of the object, and then obtain the template (step 丨 2 4). As a matter of course, the present invention can also use another way to explain the invention of the present invention. Please refer to: "Figure 3", the second embodiment of the present invention focusing method flowchart, which includes The following steps: capture an image 歹 J (step 3 1 0), obtain an initial template (step 3 2 0), find the best four match =, (step 3 3 0), obtain an updated matching template (step 34), Adjust the shooting (step "step πο") and whether it is the last image? (Step 36). In the process of "Figure 3", the invention is further disclosed such as: Way to be the focus of the present invention. First, the image sequence is captured by the same camera (step 31), and the captured image is the basis for the board acquisition. Next, you must first obtain the target object's =, ten ° ^ plate, which is step 32 (3, the template acquisition method Xiao "Figure 2": no longer explained. After the initial template is obtained, the template is used here The comparison basis of $ 田 二 吴 板 matching, in the next image, is to find the position of soil matching (step 33), that is, the initial template is used to perform the r ^ matching action. During the template matching process As a towel, the target position of the next image in the initial Wu image sequence can be used.
1241535 五、發明說明（7) '^ --- 即步驟3 3 0的意含。在尋找到下一幅圖像的最佳匹配位置 後，即可取％更新匹配模板（步驟3 4 〇 )，接著，即可依據 初始模板的位置與更新後的匹配模板位置來調整拍攝角度 (步驟3 5 0)。接下來，即判斷此次的圖像與是否為最後一 幅圖像（步驟3 6 0 )，如果不是，則以更新後的匹配模板作 為初始模板，回到步驟3 3 〇，尋找下一幅圖像的最佳匹配 位置’繼續進行模板匹配與拍攝角度的調整動作，直到目 標物消失，不再有圖像序列的產生。 「第4圖」則更詳盡地說明了本發明的移動物體的聚 焦方法，其包含下列步驟：輸入圖像序列（步驟4 〇 5 )、操 取第一幅圖像（步驟410)、擷取下一幅圖像（步驟415)、局 部增強處理（步驟42 0 )、計算直方圖投影圖並將之二值化 (步驟42 5 )、獲得候選目標的中心點集合（步驟43 0 )、是否 集合為空集？（步驟435 )、處理判斷數據（步驟44 0 )、取得N 值（步驟445)、N是否小於5?(步驟4 5 0 )、產生模板（步驟 4 5 5 )、輸出目標的位置與大小（步驟4 6 0 )、到達最後_幅？ (步驟46 5 )與設定N = 0(步驟470 )。 · 同樣地，一開始為輸入圖像序列（步驟4 〇 5 )，接著掏 取第一幅圖像（步驟410)，隨即產生模板（步驟4 5 5 )。產生 模板的用途同樣為模板匹配之用，產生模板的方式同樣為 「第2圖」所說明的方式。接著，從圖像序列當中操取下… 一幅圖像（步驟4 1 5 )，並將此次擷取的圖像加以局部增強 處理（步驟42 0 )。局部增強處理是為了增大物體與背景之 間的對比度，當灰度直方圖中出現了兩個峰值的情況，可1241535 V. Description of the invention (7) '^ --- That is, the meaning of step 3 3 0. After finding the best matching position for the next image, you can take% to update the matching template (step 34). Then, you can adjust the shooting angle based on the position of the initial template and the position of the updated matching template (step 3 5 0). Next, it is judged whether the current image is the last image (step 360), if not, the updated matching template is used as the initial template, and the process returns to step 3300 to find the next image. The best matching position of the image 'continues to perform template matching and adjustment of the shooting angle until the target disappears and no more image sequences are generated. "Figure 4" explains the focusing method of the moving object of the present invention in more detail, which includes the following steps: input image sequence (step 4 05), manipulate the first image (step 410), capture The next image (step 415), local enhancement processing (step 42 0), calculating and binarizing the histogram projection (step 42 5), obtaining the center point set of candidate targets (step 43 0), whether The collection is empty? (Step 435), processing the judgment data (step 44 0), obtaining the value of N (step 445), is N less than 5? (Step 4 5 0), generating a template (step 4 5 5), outputting the position and size of the target ( Step 4 6 0), reach the last frame? (Step 46 5) and setting N = 0 (Step 470). · Similarly, start with the input image sequence (step 4 05), then extract the first image (step 410), and then generate a template (step 4 5 5). The purpose of generating templates is also for template matching, and the method of generating templates is also the method described in "Figure 2". Then, from the image sequence, an image is taken (step 4 1 5), and the captured image is locally enhanced (step 42 0). The local enhancement process is to increase the contrast between the object and the background. When two peaks appear in the grayscale histogram, you can
第10頁 1241535Page 10 1241535
以假設分別對應著圖像中的物體和背景，一般情況不都是 如此’則可以對相應的灰度級段進行拉伸、厪縮、或是其 他處理，即直方圖均衡，使得圖像背景與物體間的對比度 增大。進行這種處理的第一步是在原來圖像灰度直方圖的 基礎上，獲得要進行處理的灰度級段，以及確定相應的處 理方法。 將所擷取的第二幅圖像作局部增強處理後，即開始處 理第二幅圖像的影像資料，亦即，計算直方圖投影圖並將 之二值化（步驟4 2 5 )。在直方圖投影圖當中，可獲得候選 目標的中心點集合（步驟43 0 )，也就是目標物的部分。接 著，必須判定目標的中心點集合狀況，是否集合為空 集？（步驟4 3 5 )，如果不是，則表示目標物存在，則進行處 理判斷數據（步驟440 );如果集合為空集，則取得N值（步 驟445 )，亦即，取得目標的中心點的灰度值。在步驟445 之後，隨即檢查N的值，判斷是否N小於5(步驟45 0 )，如果 是，則表示目標以消失；如果不是，再回到步驟4 1 5繼續 擷取下一幅圖像以作判定。 在步驟440之進行處理判斷數據，係將步驟45 5所產生 的模板加以比對並匹配，即可進行輸出目標的位置與大小 (步驟4 6 0 )，同時，將此次所產生的目標位置與大小作為 新的模板，亦即，回到步驟4 5 5。接著，判定是否到達最 後一幅（步驟4 6 5 )，如果是的話，即結束；如果不是，則 設定N = 0(步驟47 0 )，並回到步驟41 5繼續判定下一幅圖 像0It is assumed that they correspond to the objects and background in the image, which is not always the case. 'You can stretch, shrink, or perform other processing on the corresponding gray level segments, that is, the histogram equalization, so that the image background The contrast with the object increases. The first step in this kind of processing is to obtain the gray level segments to be processed based on the gray histogram of the original image, and determine the corresponding processing method. After the captured second image is subjected to local enhancement processing, the image data of the second image is processed, that is, the histogram projection image is calculated and binarized (step 4 2 5). In the histogram projection, the center point set of the candidate target can be obtained (step 43 0), that is, the part of the target. Next, we must determine the status of the center set of the target. Is the set an empty set? (Step 4 3 5), if not, it indicates that the target exists, then process the judgment data (step 440); if the set is an empty set, obtain N value (step 445), that is, obtain the center point of the target grayscale value. After step 445, then check the value of N to determine whether N is less than 5 (step 45 0). If it is, it means that the target disappears; if not, go back to step 4 1 5 and continue to capture the next image to Make a decision. In step 440, the judgment data is processed. The template generated in step 45 5 is compared and matched to output the position and size of the target (step 4 60). At the same time, the target position generated this time is output. With the size as the new template, that is, go back to steps 4 5 5. Next, it is determined whether the last frame is reached (step 4 6 5), if it is, the end; if not, then set N = 0 (step 47 0), and return to step 4 5 to continue to determine the next image 0
1241535 五、發明說明（9) 在每一次的模板產生與更新模板的過程當中，都會相 應地產生模板的位置資訊。透過此一不斷更新的位置資 訊，即可達到對目標物跟蹤的目的。並且，透過此一不斷 更新的位置資訊，即可調整攝影機的拍攝角度。 雖然本發明以前述之較佳實施例揭露如上，然其並非 用以限定本發明，任何熟習相關技藝者，在不脫離本發明 之精神和範圍内，當可作些許之更動與潤飾，因此本發明 之專利保護範圍須視本說明書所附之申請專利範圍所界定 者為準。1241535 V. Description of the invention (9) During each template generation and update, the location information of the template will be generated accordingly. Through this constantly updated position information, the purpose of tracking the target can be achieved. And, through this constantly updated position information, the camera's shooting angle can be adjusted. Although the present invention is disclosed in the foregoing preferred embodiments as above, it is not intended to limit the present invention. Any person skilled in the relevant arts can make some changes and decorations without departing from the spirit and scope of the present invention. The patent protection scope of the invention shall be determined by the scope of the patent application scope attached to this specification.
第12頁 1241535 圖式簡單說明 第1圖為本發明之移動物體的聚焦方法之第一具體實 施流程圖； 第2圖為本發明之移動物體的聚焦方法之模板取得流 程圖； 第3圖為本發明之移動物體的聚焦方法之第二具體實 施例流程圖；及 第4圖為本發明之移動物體的聚焦方法之具體實施例 之詳細流程圖。 【圖示符號說明】 步驟11 0 設 定 曰 取 佳 模 板 匹 配 位置 步驟1 2 0 擷 取 圖 像 序 列 步驟1 2 1 擷 取 一 幅 圖 像 步驟1 2 2 計 算 灰 度 直 方 圖 步驟1 2 3 計 算 直 方 圖 投 影 圖 步驟1 2 4 取 得 模 板 步驟1 3 0 取 得 初 始 模 板 與 其 位置 步驟1 4 0 調 整 拍 攝 角 度 步驟1 5 0 是 否 為 最 後 一 幅 圖 像？ 步驟3 1 0 擷 取 圖 像 序 列 步驟3 2 0 取 得 初 始 模 板 步驟330 尋 找 最 佳 匹 配 位 置 步驟340 取 得 更 新 匹 配 模 板 步驟3 5 0 調 整 拍 攝 角 度 步驟3 6 0 是 否 為 1^7 取 後 一 幅 圖 像？Page 121241535 Brief description of the diagram. Figure 1 is the first specific implementation flowchart of the focusing method of the moving object of the present invention; Figure 2 is the flowchart of obtaining the template of the focusing method of the moving object of the present invention; Figure 3 is A flowchart of a second specific embodiment of a method for focusing a moving object according to the present invention; and FIG. 4 is a detailed flowchart of a specific embodiment of a method for focusing a moving object according to the present invention. [Illustration of Symbols] Step 11 0 Set the best template matching position Step 1 2 0 Capture image sequence Step 1 2 1 Capture an image Step 1 2 2 Calculate gray level histogram Step 1 2 3 Calculate histogram Figure Projection Step 1 2 4 Obtain template Step 1 3 0 Obtain initial template and its position Step 1 4 0 Adjust shooting angle Step 1 5 0 Is it the last image? Step 3 1 0 Capture the image sequence Step 3 2 0 Get the initial template Step 330 Find the best matching position Step 340 Get the updated matching template Step 3 5 0 Adjust the shooting angle Step 3 6 0 Whether it is 1 ^ 7 Take the next picture image?
第13頁 1241535 圖式簡單說明 步驟4 0 5 步驟41 0 步驟4 1 5 步驟4 2 0 步驟4 2 5 步驟4 3 0 步驟4 3 5 步驟440 步驟445 步驟4 5 0 步驟4 5 5 步驟4 6 0 步驟4 6 5 步驟4 7 0 輸入圖像序列 擷取第一幅圖像 擷取下一幅圖像 局部增強處理 計算直方圖投影圖並將之二值化 獲得候選目標的中心點集合 是否集合為空集？ 處理判斷數據 取得N值 N是否小於5? 產生模板 輸出目標的位置與大小 到達最後一幅？ 設定N = 0Page 131241535 Schematic description of step 4 0 5 step 41 0 step 4 1 5 step 4 2 0 step 4 2 5 step 4 3 0 step 4 3 5 step 440 step 445 step 4 5 0 step 4 5 5 step 4 6 0 Step 4 6 5 Step 4 7 0 Input image sequence capture First image capture Next image Local enhancement processing Calculate histogram projection and binarize to obtain whether the set of center points of candidate targets is set Is an empty set? Process judgment data. Get N value. Is N less than 5? Generate template. Position and size of output target. Reach the last one? Set N = 0
第14頁Page 14
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW90132874A TWI241535B (en) | 2001-12-28 | 2001-12-28 | Focusing method for a moving object |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW90132874A TWI241535B (en) | 2001-12-28 | 2001-12-28 | Focusing method for a moving object |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| TWI241535B true TWI241535B (en) | 2005-10-11 |
Family
ID=37013981
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW90132874A TWI241535B (en) | 2001-12-28 | 2001-12-28 | Focusing method for a moving object |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TWI241535B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI557689B (en) * | 2013-04-23 | 2016-11-11 | 鴻海精密工業股份有限公司 | System and method of computing focuses |
-
2001
- 2001-12-28 TW TW90132874A patent/TWI241535B/en not_active IP Right Cessation
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI557689B (en) * | 2013-04-23 | 2016-11-11 | 鴻海精密工業股份有限公司 | System and method of computing focuses |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Gracker: A graph-based planar object tracker | |
| CN101416219B (en) | Foreground/background segmentation in digital images | |
| JP5366756B2 (en) | Information processing apparatus and information processing method | |
| CN103337077B (en) | A kind of based on multi-scale division and the visible ray of SIFT and the method for infrared image registration | |
| JP4961965B2 (en) | Subject tracking program, subject tracking device, and camera | |
| WO2019042426A1 (en) | Augmented reality scene processing method and apparatus, and computer storage medium | |
| JP2011090413A (en) | Image recognition apparatus, processing method thereof, and program | |
| CN109635686B (en) | Two-stage pedestrian searching method combining human face and appearance | |
| US20080013837A1 (en) | Image Comparison | |
| JP2009522591A (en) | Method and apparatus for controlling autofocus of a video camera by tracking a region of interest | |
| CN105706108A (en) | Frequency spectrum resource scheduling device, method and system | |
| WO2019174544A1 (en) | Image synthesis method and apparatus, and computer storage medium and electronic device | |
| JP2011511977A (en) | Detection of facial expressions in digital images | |
| JPWO2012077286A1 (en) | Object detection apparatus and object detection method | |
| WO2012068902A1 (en) | Method and system for enhancing text image clarity | |
| WO2022103684A1 (en) | Face-aware person re-identification system | |
| JP5503510B2 (en) | Posture estimation apparatus and posture estimation program | |
| US20190364196A1 (en) | Method and Apparatus for Generating Shot Information | |
| CN110941992A (en) | Smile expression detection method and device, computer equipment and storage medium | |
| Nguyen et al. | Combined YOLOv5 and HRNet for high accuracy 2D keypoint and human pose estimation | |
| TW201421101A (en) | Method for automatically focusing on specific movable object, photographic apparatus including automatic focus function, and computer readable storage media for storing automatic focus function program | |
| TWI241535B (en) | Focusing method for a moving object | |
| JP2001119622A (en) | Image-puckup device and control method therefor | |
| CN114494732A (en) | Gait recognition method and device | |
| JP2012234500A (en) | Pattern identification device, pattern identification method and program |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MM4A | Annulment or lapse of patent due to non-payment of fees |