201036424 六、發明說明: 【發明所屬之技術領域】 [0001]本發明係關於一種攝像裝置及其動態偵測方法。 【先前技術】 酬當前,用在安全監控領域(比祕行金庫、企業機密場 所等)之攝職,其動H物體制方法是透過對攝像機 取得之連續場景之圖像進行特徵值分析來進行。然,傳 統之攝像機在動態偵測之過程中,攝像機之機身與攝像 頭之相對位置必須固定,即攝像頭不能任意移動。若在 偵測之過程中,攝像頭處於移動欺態’如攝影頭或攝影 機被人移動,即攝像視角發生了改變,則無法偵測出圖 像中之動態部分,不能達到動態偵測之目的,是故,給 安全監控工作帶來隱患。 【發明内容】 [0003] 鑒於前述内容,有必要提,一種攝像裝置及其動態偵測 方法’可在攝像裝置於移動狀態下進行動態偵測。 [0004] —種攝像裝置,用於在移動狀感對一監控物件進行動 態偵測,包括: [0005] —圖像獲取單元,用於連續獲取該監控物件之複數圖像 9 [00〇6] —資料處理單元,用於計算該圖像獲取單元所獲取之複 數圖像之特徵值; [0007] —記憶單元,用於存儲該資料處理單元所得到之特徵值 098110018 表單編號Α0101 第4貢/共17頁 0982016443-0 201036424 [0008] 一圖像匹配單元,用於對複數圖像中相鄰兩張圖像之特 徵值進行模糊匹配,根據相鄰兩張圖像中之後一張圖像 之特徵值搜尋其與前一張圖像中之相符區域;以及 [0009] —動態偵測單元,用於對該相鄰之兩張圖像之相符區域 之特徵值進行精確比對,並將特徵值不同之部分標示為 動態區域。 [0010] 一種使用於一攝像裝置對一監控物件之動態偵測方法, 其包括以下步驟: 〇 [0011] 獲取該監控物件之一第一時間之圖像; [0012] 計算並存儲該第一時間之圖像之特徵值; [0013] 獲取該監控物件之一第二時間之圖像; [0014] 計算並存儲該第二時間之圖像之特徵值; [0015] 將該第一時間與第二時間之圖像之特徵值進行模糊匹配 ,按照該第二時間之圖像之特徵值搜尋其與第一時間之 Q 圖像之相符區域;以及 [0016] 將該第一時間與第二時間之圖像之相符區域之特徵值進 行精確比對,將該第一時間與第二時間之圖像之相符區 域之特徵值不同之部分標示為動態區域。 [0017] 前述攝像裝置及其動態偵測方法透過該攝像裝置在移動 狀態下連續獲取圖像,將獲取之兩張相鄰之圖像進行特 徵值模糊匹配找出相符區域,再對連續獲取之兩張相鄰 之圖像之相符區域之特徵值進行精確比對,得到相符區 域特徵值不同之部分即為動態區域,從而完成攝像裝置 098110018 表單編號A0101 第5頁/共17頁 0982016443-0 201036424 之動態偵測。 【實施方式】 [0018] [0019] 請參閱圖1,本發明攝像 # ^ , 氓置U之較佳實施方式包括一圖 像獲取早m — 一 ®你nr 爽理早兀200、—記憶單元300 獲取單配單元彻以及—動態偵·元_。該圖像 疋0用於連續獲取_監控物 料處理單元20。與該圖像獲取單元1〇(u魏圖像,該資 圖後心~ _ 丨職早^*1_連,用於計算該 錄之複㈣叙龍值;該記憶單 單元200所Γ料處理早疋200相連,用於存儲該資料處理 一 ㈣之特徵值;該圖像K配單細〇與該記憶 單元300相連’用於對複數圏像中相鄰之兩㈣像之特徵 值進行_匹配祕⑽中之後-張圖像中 某一區域之特徵值搜尋前__像中之相频域,如將 特徵值相同度在8〇%,%之圖像區域視為相符區域,在其 他實施方式中,該特徵值相同度亦可根據實際需要被擴、 大或縮小;該動態_單元5QQ與該圖像·單元4〇〇相 連,用於對該相鄰之兩張圖像之相符區域之特徵值進行 精確比對,並將特徵值不同之部分標示為該監控物件之 動態區域,以提示操作者此時晝面出現變動,或者透過 一電腦系統對變動之畫面進行錄影。 其中,該圖像獲取單元100可為一CCD (Charge Coupled Device,電荷耦合裝置)。該特徵值可為圖像 之幾何特徵、顏色特徵、紋理特徵等透過該資料處理單 元200進行傅立葉變換得到之資料。 下面將舉例詳細描述: 098110018 表單編號A0101 第6頁/共17頁 0982016443-0 [0020] 201036424 [0021] 請一併參閱圖2及圖3,該攝像裝置Cl在移動狀態下對監 控物件如一動態場景s 1連續獲取圖像。該動態場景s 1中 存在一動態物體,如—瓢蟲40 ’圖像10表示該攝像裝置 C1在一第一時間A時所獲取之圖像;圖像20表示該攝像裝 置C1在一第二時間B時所獲取之圖像;圖像30表示該攝像 裝置C1在一第三時間c時所獲取之圖像。其中,該圖像1〇 、20及30是該攝像裝置C1連續拍攝之三張圖像。該攝像 裝置C1中之資料處理單元200計算該三張圖像1〇、20、 30之特徵值並將其存儲在該記憶單元300中。 Ο [0022] 請一併參閱圖4 ’該攝像裝置C1中之圖像匹配單元4〇〇將 該圖像10與圖像20之特徵值透過自動關聯(Auto Correlation) 之方式進行模糊匹配 ,按該圖像 2〇 之特徵值 搜尋其與該圖像10之相符區域12 ;同理,將該圖像20與 圖像30之特徵值透過自動關聯之方式進行模糊匹配,按 該圖像30之特徵值搜尋其與該圖像20之相拜區域23。 [0023] Ο 請一併參閱圖5,該動態偵測單元&00將該圖像1〇與圖像 20之相符區域12之特徵值進行精確比對以找出動態區域 ,即將該相符區域12在該圖像10中之特徵值與其在該圖 像20中之特徵值不同之區域,標示為動態區域,該動態 區域即為瓢蟲40所在之區域。同理,該動態偵測單元500 將該圖像20與圖像30之相符區域23之特徵值進行精確比 對’即將該相符區域23在該圖像20中之特徵值與在該圖 像30中之特徵值不同之區域,標示為動態區域,該動態 區域亦為瓢蟲40所在之區域。 請一併參閱圖6,一種應用本發明攝像裝置C1之動態偵測 098110018 表單煸衆 A010! 第7買/共17頁 0982016443-0 [0024] 201036424 方法之步驟如下: [0025] #驟S1 :該圖像獲取單元1〇〇獲取該動態場景以在一第— 時間A之圖像1〇 ; [0026] 步驟S2,該資料處理單元2〇〇計算該圖像1〇之特徵值,並 將其存儲於該記憶單元3〇〇内; _7] #驟S3 :該圖像獲取單元剛獲取該動態場景财一與該 第一時間A連續之第二時間β之圖像2〇 .201036424 VI. Description of the Invention: [Technical Field of the Invention] [0001] The present invention relates to an image pickup apparatus and a motion detecting method thereof. [Prior Art] Currently, it is used in the field of security monitoring (before secret bank vaults, corporate secret places, etc.), and its dynamic H system method is to perform eigenvalue analysis on the images of continuous scenes obtained by the camera. . However, in the process of motion detection, the relative position of the camera's body and camera must be fixed, that is, the camera cannot move freely. If the camera is in a moving state during the detection process, if the camera or the camera is moved, that is, the camera angle changes, the dynamic part of the image cannot be detected, and the motion detection cannot be achieved. Therefore, it brings hidden dangers to the safety monitoring work. SUMMARY OF THE INVENTION [0003] In view of the foregoing, it is necessary to provide that an image pickup apparatus and a motion detection method thereof can perform motion detection while the image pickup apparatus is in a moving state. [0004] A camera device for dynamically detecting a monitored object in a moving sense, comprising: [0005] an image acquiring unit for continuously acquiring a plurality of images of the monitored object 9 [00〇6 a data processing unit for calculating a feature value of the complex image acquired by the image acquisition unit; [0007] a memory unit for storing the feature value obtained by the data processing unit 098110018 Form No. 1010101 / Total 17 pages 0982016443-0 201036424 [0008] An image matching unit for performing fuzzy matching on the feature values of two adjacent images in a complex image, according to the next image in the adjacent two images The feature value is searched for the matching region with the previous image; and [0009] - a motion detecting unit for accurately comparing the feature values of the matching regions of the two adjacent images, and The part with different eigenvalues is indicated as a dynamic area. [0010] A method for detecting motion of a monitoring object for an imaging device, comprising the steps of: [0011] acquiring an image of the first time of the monitoring object; [0012] calculating and storing the first The feature value of the image of the time; [0013] acquiring an image of the second time of the monitored object; [0014] calculating and storing the feature value of the image of the second time; [0015] Performing fuzzy matching on the feature values of the image of the second time, searching for the matching region with the Q image of the first time according to the feature value of the image of the second time; and [0016] the first time and the second time The feature values of the coincident regions of the image of time are accurately compared, and the portion of the coincidence region of the image of the first time and the second time is marked as a dynamic region. [0017] The camera device and the motion detection method thereof continuously acquire images through the camera device in a moving state, and perform fuzzy matching of the acquired two adjacent images to find a matching region, and then sequentially acquire two adjacent frames. The eigenvalues of the matching regions of the image are accurately compared, and the difference between the eigenvalues of the matching regions is obtained as the dynamic region, thereby completing the motion detection of the camera device 098110018 Form No. A0101 Page 5 of 17 page 0982016443-0 201036424 . [Embodiment] [0019] Referring to FIG. 1, the preferred embodiment of the present invention includes an image acquisition early m-a® you nr refreshing early 200, a memory unit 300 Get the single-match unit and the dynamic detection element_. This image 疋0 is used to continuously acquire the _ monitoring material processing unit 20. And the image acquisition unit 1 〇 (u Wei image, the picture after the heart ~ _ 丨 job early ^ * 1_, used to calculate the record of the complex (four) Xu Long value; the memory unit 200 The processing block 200 is connected to store the feature value of the data processing one (four); the image K is associated with the memory unit 300 for performing the feature values of the adjacent two (four) images in the complex image _ matching secret (10) after the eigenvalue of a certain area in the image is searched for the phase frequency domain in the __ image, if the eigenvalue is equal to 8〇%, the image area of % is regarded as the matching area, In other embodiments, the feature value of the same degree may be expanded, reduced or reduced according to actual needs; the dynamic_unit 5QQ is connected to the image unit 4〇〇 for the two adjacent images. The eigenvalues of the matching regions are accurately compared, and the different parts of the eigenvalues are marked as dynamic regions of the monitored object to prompt the operator to change at this time, or to record the changed picture through a computer system. The image acquisition unit 100 can be a CCD (Charge Coupled Device). The feature value may be a data obtained by performing Fourier transform on the geometrical features, color features, texture features, etc. of the image through the data processing unit 200. The following is a detailed description: 098110018 Form No. A0101 Page 6 of 17 Page 0982016443-0 [0020] 201036424 [0021] Please refer to FIG. 2 and FIG. 3 together, the camera device C1 continuously acquires images of the monitoring object such as a dynamic scene s1 in a moving state. There is one in the dynamic scene s1. The dynamic object, such as the ladybug 40' image 10, represents the image acquired by the camera C1 at a first time A; the image 20 represents the image acquired by the camera C1 at a second time B. The image 30 indicates an image acquired by the imaging device C1 at a third time c. The images 1〇, 20, and 30 are three images continuously captured by the imaging device C1. The imaging device C1 The data processing unit 200 calculates the feature values of the three images 1〇, 20, and 30 and stores them in the memory unit 300. [0022] Please refer to FIG. 4 'the image in the camera C1 Image matching unit 4 〇〇 the image 10 Performing fuzzy matching with the feature value of the image 20 by means of Auto Correlation, searching for the matching region 12 with the image 10 according to the feature value of the image 2; similarly, the image 20 is The feature value of the image 30 is subjected to fuzzy matching by means of automatic association, and the image area of the image 30 is searched for its area of worship 23 with the image 20. [0023] Ο Please refer to FIG. 5 together, the dynamic detection The measuring unit & 00 accurately compares the image 1 to the feature value of the matching region 12 of the image 20 to find a dynamic region, that is, the feature value of the matching region 12 in the image 10 and its An area having a different feature value like 20 is indicated as a dynamic area, which is the area where the ladybug 40 is located. Similarly, the motion detecting unit 500 accurately compares the image 20 with the feature value of the matching region 23 of the image 30, that is, the feature value of the matching region 23 in the image 20 and the image 30. The area with different eigenvalues is marked as a dynamic area, which is also the area where the ladybug 40 is located. Please refer to FIG. 6 together, a dynamic detection of the camera device C1 of the present invention, 098110018, a form of A010! 7th buy/total 17 pages 0982016443-0 [0024] 201036424 The steps of the method are as follows: [0025] #STEPS1: The image acquisition unit 1 〇〇 acquires the dynamic scene to an image of the first time A; [0026] step S2, the data processing unit 2 calculates the characteristic value of the image 1〇, and It is stored in the memory unit 3〇〇; _7] #STEPS3: The image acquisition unit just acquires the image 2〇 of the second time β of the dynamic scene and the first time A.
[0028]步驟S4 :該資料處理單元2〇〇計算該圖像2〇之特徵值’並 將其存儲於該記憶單元30〇内; [〇〇29]步驟S5 :該圖像匹配單元400將該圖像10及圖像20之特 徵值透過自動關聯之方式進行模糊匹配,按照該圖像20 之特徵值搜尋其與圖像1〇之相符區域12 ; [0030] 步驟S6 :該動態偵測單元5〇〇將該相符區域12在該圖像 10中之特徵值與相符區域12在該圖像2〇種之特徵值進行 精確比對,將該相符區域12在該圖像1〇與圖像2〇中特徵 值不同之部分標示為動態區域,從而得到瓢蟲40之所在 區域,完成動態偵測。本發明攝像裝置及其動態偵測方 法亦可用在攝像裝置C1不動之情況下進行動態偵測。 [0031] 前述攝像裝置及其動態偵測方法透過該攝像裝置1〇〇在移 動狀態下連續獲取圖像,將獲取之兩張相鄰之圖像進行 特徵值模糊匹配找出其相符區域,再對連續獲取之兩張 圖像之相符區域之特徵值進行精確比對,找出特徵值不 同之。P 77即為動態物體所在之區域,從而完成攝像袈置 098110018 表單編號A0101 第8 1/共17頁 0982016443-0 201036424 1 0 0之動態偵測。 [0032] 綜上所述,本發明符合發明專利要件,爰依法提出專利 申請。惟,以上所述者僅為本發明之較佳實施例,舉凡 熟悉本案技藝之人士,在爰依本發明精神所作之等效修 飾或變化,皆應涵蓋於以下之申請專利範圍内。 【圖式簡單說明】 [0033] 圖1係本發明攝像裝置之較佳實施方式之結構方框圖。 [0034] 圖2係應用本發明攝像裝置對一監控物件進行連續攝像之 Ο 示意圖。 [0035] 圖3係圖2中本發明攝像裝置所取得之三個連續圖像。 [0036] 圖4係應用本發明攝像裝置及其動態偵測方法所獲得之兩 相鄰圖像之相符區域之示意圖。 [0037] 圖5係應用本發明攝像裝置及其動態偵測方法獲得動態區 域之示意圖。 [0038] ·圖6係應用於本發明攝像裝置之動態偵測方法之較佳實施 方式之流程圖。 【主要元件符號說明】 攝像裝置 C1 圖像獲取單元 100 資料獲取單元 200 記憶單元 300 圖像匹配單元 400 動態偵測單元 500 瓢蟲 40 動態場景 S1 圖像 10-30 相符區域 12,23 098110018 表單編號Α0101 第9頁/共17頁 0982016443-0[0028] Step S4: the data processing unit 2 calculates the feature value ' of the image 2〇 and stores it in the memory unit 30〇; [〇〇29] Step S5: the image matching unit 400 The feature values of the image 10 and the image 20 are subjected to fuzzy matching by means of automatic association, and the matching region 12 of the image 1 is searched according to the feature value of the image 20; [0030] Step S6: the motion detection The unit 5 精确 accurately compares the feature value of the matching region 12 in the image 10 with the feature value of the matching region 12 in the image 2, and the matching region 12 is in the image 1 The part with different eigenvalues in 2〇 is marked as a dynamic area, thereby obtaining the area where the ladybug 40 is located, and performing motion detection. The camera device of the present invention and the motion detecting method thereof can also be used for motion detection while the camera device C1 is not moving. [0031] The camera device and the motion detecting method thereof continuously acquire images through the camera device 1 in a moving state, and perform fuzzy matching of the acquired two adjacent images to find a matching region thereof, and then continuously The eigenvalues of the matching regions of the obtained two images are accurately compared to find out that the feature values are different. P 77 is the area where the dynamic object is located, thus completing the camera setup 098110018 Form No. A0101 Page 8 1/17 Page 0982016443-0 201036424 1 0 0 Motion detection. [0032] In summary, the present invention complies with the requirements of the invention patent and submits a patent application according to law. However, the above description is only the preferred embodiment of the present invention, and equivalent modifications or variations made by those skilled in the art of the present invention should be included in the following claims. BRIEF DESCRIPTION OF THE DRAWINGS [0033] FIG. 1 is a block diagram showing the configuration of a preferred embodiment of an image pickup apparatus of the present invention. 2 is a schematic diagram of continuous imaging of a monitored object by using the image pickup apparatus of the present invention. 3 is a view showing three consecutive images obtained by the image pickup apparatus of the present invention in FIG. 2. 4 is a schematic diagram of a matching region of two adjacent images obtained by applying the image pickup apparatus of the present invention and a motion detecting method thereof. 5 is a schematic diagram of obtaining a dynamic region by applying the image pickup apparatus of the present invention and a motion detecting method thereof. [0038] Fig. 6 is a flow chart showing a preferred embodiment of a motion detecting method applied to the image pickup apparatus of the present invention. [Description of main component symbols] Camera C1 Image acquisition unit 100 Data acquisition unit 200 Memory unit 300 Image matching unit 400 Motion detection unit 500 Ladybug 40 Dynamic scene S1 Image 10-30 Matching area 12, 23 098110018 Form number Α0101 Page 9 of 17 0982016443-0