201123915 六、發明說明： 【發明所屬之技術領域】 ’特別是一種全周影 本發明係關於一種攝像參數的調整方法 像的攝像參數的調整方法。 【先前技術】 4為許夕人日,生活中不可或缺的交通1具。為了提高行 車的安全’各種辅助駕驶人的行車控·統不_雜出新 些行車控·統使汽車缺的安全更進—㈣提升。 °201123915 VI. Description of the Invention: [Technical Field to Which the Invention Is Alonged] </ RTI> In particular, a full-circle image method relates to a method for adjusting an imaging parameter of an image capturing parameter adjustment method. [Prior Art] 4 is an indispensable transportation in the life of Xu Xiren. In order to improve the safety of driving vehicles, the driving controls of various assisted drivers are not mixed with new traffic control systems to make the safety of car shortages more advanced. °
當汽車行經狹小|主士且、古* y一 角’容易產生碰撞的 各安裝一照相模組， 外’為了效能的考量，也可以增加照蝴_數目。將這些和 像整合成-個360。的全鄕像舰圖，此全周影像俯視^ 不於螢幕上。因為照相模_取景角度並非從上往下取景，因此， 影像合成前’需要經過_轉換的運算。驗人_此一系統， 可以像烏-樣從汽車上方俯視汽車關圍。此—_方式以用 同-顯示晝面’將汽車的周圍實際情形，清楚的顯示給駕歇人。 駕歇人可依據此-顯*的全周影像俯視關控汽車，例如：於^ 車時判H車與料格線及其轉礙物（如人行道、汽車）的^ 對位置献行驶於狹小巷弄時，能清贿補的 ^ I，以避 免碰撞意外發生’藉以提高行車安全。 在-般的環境之下’汽車四周的光線亮度/色彩可能會有極大 201123915 的差異。比如說’纽車接近牆料，汽轄近触的一側的亮 度會遠低於汽料-_亮度。鱗，若是汽車四_多個鏡頭 都使用相關攝像參數進行影像_取，其中的幾個鏡頭所揭取 的影像可能會產生過亮献树的情形。因此，所構成的全周影 像則會有亮度不均勻且在合成影像連接翁圖像不連續的斷層產 以致於難以辨識錄兩影像的銜接處的物體。並且，過亮或 是過暗的影像可能會使影像中物體的細節難以辨識。 &此外，-般汽車後方的煞車燈或是警示燈皆為紅色。因此， 汽車後方的鏡頭所拍攝出的影像與汽車前方的鏡頭所拍攝出的影 像在色彩上會树大的獨，導致合成全㈣像時，影像會非常 不自然。同樣地，當各鏡頭間的所拍攝到的色彩差異過大時，處 理器也很難正確的進行影像合成。 【發明内容】 鑑於以上的問題，本發明係提出—種全周影像的攝像參數的 調整方法’可__鶴像參數，减全鄕像整體的攝像參 數，例如：亮度值或色彩值等，能趨於一致。 本發明所提出之全觸像_像錄的婦枝包括：⑷根 據位於目標影像與相鄰影像的連接處的目標影像的影像區塊分別 對相鄰影像的影像區塊的攝像參數的差異，計算目標鏡頭分別相 對於相鄰影像的攝像參數的第，整值；（B)根據至少—第一調整 值’計算目標鏡頭的攝像參數的第二調整值;(Q重複執行步驟(A) 以及步驟(B) ’直至計算得所有鏡頭的攝像參數料二調整值·以 201123915 ()根據夕鱗—難值對軸整多個鏡獅攝像參數。 其中目標影像係為多個子影像中之任一 鏡頭中用以擷取目標影像的一铲鹿㈠ 軚鏡頭係為多個 办像的鏡碩，且相鄰影像係為多個子 於構成該全周影像的相對配置中所有與賭影像相鄰之子影像； 除此之外，本發明可做以下 ’ =:參數分別對該至少-相鄰影像的至少-：像二 至,Ί的差異，計异一目標鏡頭的該目標影像分別相對於 ，:該=影像的該攝像參數的至少一第一調整… 調整值，計算目標鏡攝像參數的-第二調整值；（C) 根據第-調整值對應調整目標鏡頭的攝像參數；以及罐複執行 =⑷、步驟_步驟(C)，直至計算得所有鏡頭的攝像參數的 第一§周整值。 在上述二個方法中，顧遞迴的機制重複運算以使於全周影 像中相互連接的二影像的連接處的差異可被明_縮小。遞迴機 制可配合下列二種中斷條件：當遞迴次數達—預設值或是直至所 有鏡頭的攝像參數的第二調整值中之皆小於預定值。 $此外，目標影像的影像區塊與相鄰影像的影像區塊可完全重 璺、部分重疊或是彼此鄰接。攝像參數可為亮度值或是色彩值。 第二調整值可為第值的算術平均或是加權平均。 藉由本發明之全周影像的攝像參數的調整方法，此全周影像 的攝像參數’例如亮度值或色彩值，能趨於—致。相互連接的二 201123915 影像之間的連接處的差異可被明顯的縮小。 以上之關於本發明内容之說明及以下之實施方式之說明係用 以示範與解釋本發明之精神與原理，並且提供本發明之專利申請 範圍更進一步之解釋。 【實施方式】 以下在μ轭方式中詳細敘述本發明之詳細特徵以及優點，其 内谷足以使任何熟習相關技藝者了解本發明之技術内容並據以實 施，且根據本說明書所揭露之内容、申請專利範圍及圖式，任何 熟習相關技藝者可輕易地理解本發明相關之目的及優點。以下之 實施例係進-步詳細說日林發明之難，但非峰何觀點限制本 發明之範轉。 ”月參照帛1圖』，『第i圖』為本發明實施例之顯示鳥蛾車 輛影像的助歡裝設位置及_翻之鳥_錢。在本實施 例中，照補齡散設置於車身⑽献車蝴圍應，例如， 將』、相模組&置於别鑛桿、後保險桿、左後視鏡、及右後視鏡 各設置-顆鏡頭112、114、116、118。這些照相模組可以拍攝廣 角度的景物影像，例如對拍攝區域112a、114a、肠、論可透 過鏡頭112、114、116、118進行拍攝。 請參照『第2圖』，『第2圖』係為本發明的系統方塊示意圖。 鏡頭ii2、m、ii6、m烟連接至―影像處理裝置 20。鏡頭 112' 114、116、11.8會將擷取到的子影像傳送至影像處理裝置2〇。根 據這些子影像，處理裝置2G即可計算出綱ii2、ii4、116、 201123915 . 118個別的攝像參數的調整值。之後，利用拍攝之數張景物影像產 生鳥陬車輛影像’並將此鳥嗽車輛影像顯現於螢幕中。 首先選取多個照相模組中的一個照相模組，並將此照相模組 定為目標鏡頭。而目標鏡頭所擷取的子影像稱為目標影像。在全 周影像中，與目標影像連接的子影像稱之為相鄰影像。一般而言， 一個目標影像會有多個相鄰影像。 接著，依據目標影像與相鄰影像的連接處定義目標影像的影 • 像區塊以及相鄰影像的影像區塊。目標影像的影像區塊與相鄰影 像的影像區塊可完全重疊、部分重疊或是彼此鄰接。 這些拍攝區域112a、114a、116a、118a中，有部分會和相鄰 的拍攝區域相重疊，這些相重疊的區域分別為U3b、115b、ii7b、 119b。 以鏡頭114為例’鏡頭114有兩台相鄰的麵，分別為鏡頭 112與綱116。若要調整鏡頭114的攝像參數，首先根據鏡頭出 »與綱114在重疊的區域113b所拍攝的影像區域，定義出個別的 影像區塊it行攝像參數的計算；透過兩相誠域組在重疊區域 附近的影像區塊的參數差異，計算鏡頭114的參數調整值。也就 是說，基於相鄰相機模組在重疊區域所拍攝後的影像，依據其影 像差異來產生第-調整值，用來調整鏡頭m所拍攝之影像的泉 數。 一 然後取伸目標影像的影像區塊的攝像參數與相鄰影像的影 像區塊的攝像錢。上義攝像參數可為亮度錢是色彩值。更/ 201123915 詳細的說，攝像參數為影像區塊中每一個像素的亮度值的平均， 或是每一個像素的色彩值的平均。此外，此攝像參數也可為一段 時間内，不同時間點所取得的攝像參數的平均值。 然而，當每個鏡頭對應的拍攝環境有別時，致使每個鏡頭的 攝像參數補’以至於完全重疊、部分重疊或是彼此鄰接的二個 影像區塊的二個攝像參數不-致。因此，計算目標影像的影像區 塊的攝像參數與每-相鄰影像的影像區塊的攝像參數之間的差 異，藉以得到目標鏡頭與每一相鄰影像的第一調整值。換言之， 若以得到之第-調整值雕目標辆賴像參數，可賴取得的 目標影像與對應此第-機值之相㈣像的連接處各自的影像區 塊的攝像參數趨於-致，即消料縮小完全重疊、部分重疊或是 彼此鄰接的二姆像區塊的二個攝像參數之間異。且^ ，了使-個目標影像與所有相鄰影像之間的差異減到最少， 可計算第—機值的平均值以得·二值，此平均值可為算 術平均值歧加權平均值4術平均值為將有的所有的第一調整 ^加總後除以第-調整值的數量所得值。加權平均則是將所 =的第一調整值個別乘以一權重後加總，並在將上述加總的總合 第lit又疋其他的照相輪組為目標鏡頭，計算該目標鏡頭的 調整值，直到所有相機模組的第二調整值都計算完畢。 之後’根據第二調整值對應調整多個綱的攝像參數。 於此’利用遞迴的機制，藉由多次重複的執行此調整參數的 201123915 方法；t爿用遞㈣機制，藉由多次重複的執行此調整參數的 、口㈣》進式地輯兩個子影像之_差距。並且，透過設定遞 勺中斷條件，來判斷遞迴終止的時機。 迴次數以達預設值時，或是全部的第二調整值皆小 於％才士止執行此方法。相反的，當判斷遞迴次數尚 執/:預=時’或是任一個第二調整值仍大於預設值時，則繼續 後二Γ 整值與攝像參數的動作。在執行最後一次遞迴 每遞迴時所得的㈣整值加總或是將 攝料的攝像參數與遞迴後的攝像參數的差異，作為 生各自對應的攝像參數的用=了針對各個鏡頭產 影像之間的色彩值/亮度值的以使此些鏡頭所擷取得的子 ^ . 差距此夠減少。由這些子影像所組合 而成的王周影像的色彩值/亮度值能夠趨於-致。 像的H係為根據本發明之第一實施例之娜 整枝之餘圖。其中，全周·係由多個子 像所耻’且乡好影像齡獅乡輯賴操取。 的塊鄰影像的連接處中，目標影像 目標影像對應相鄰影像的攝像參數的第 卿梅像差異來 201123915 再計算於步驟(S11G)中得到之財第—調整值的平均值以得 到第二調整值(S120)。 接著，判斷是否全部鏡頭的第二調整值都已計算(sl3〇)。 當判斷尚有未計算時，更換目標鏡頭為未計算第二調整值的 鏡頭(Sl4〇) ’並重複執行步驟(sll〇)、步驟即嶋步驟阳〇)。 反之，當判斷所有鏡頭都已計算時，根據第二調整值對應調整多 個鏡頭的攝像參數(S150)。 接著’判斷遞迴次數是否小於預設值(316〇)。當判斷遞迴次數 大於預設值時’即結束攝像參數的輕。#躺遞迴次數小於預 設值時，__斷是否全部第二娜值皆小於或等於預定值 (S170)。因為第二調整值可能為正數或是負數所以較佳的實施方 式為判斷衫全部第二調整_崎值皆小於預定值；亦可判斷 第二調整值的絕對值中的最大值是否小於或等於預設值。 當第二調整值魏對值巾的最大值小於預設值時，即可停止 執行此方法。相反的，當第二調整值的絕對值中的最大值大於預 設值時，則重新執行步驟(sn〇)、步驟(sl2〇)、步驟(si3〇)、步驟 (S140)、步驟(si5〇)與步驟(S160)。 雖然在此實施财，步驟(遍)與步驟(咖健兩個遞迴中斷 條件可依序觸。也就是，當兩個遞迴帽條件之中任―個條件 被達成時，即停止執行本方法。_，本伽之朗並不限於次。 步驟(S160)與步驟(S170)的執行順序亦可被調換。或者是，在步驟 (S160)或步驟(S170)之中，可只執行其中一個遞迴中斷條件。 201123915 解本發明。 假設總共有三個鏡頭(第一鏡頭、第二鏡頭與第三鏡頭)分別擷 取三個子影像(第一子影像、第二子影像與第三子影像）。第一子影 像、第二子影像與第三子影像組合成一全周影像。也就是說，第 一子影像的相鄰影像為第二子影像與第三子影像。第二子影像的 相鄰影像為第-子影像與第三子影像。第三子影像的相鄰影像為 第一子影像與第二子影像。When the car travels through a small size, the main squad and the squad are easy to collide, each of them installs a photographic module, and the external ‘the number can be increased for the sake of performance. Combine these images into a 360. The full-scale image of the ship, this full-week image overlooks ^ not on the screen. Since the camera mode finder angle is not framing from top to bottom, the image conversion process requires a _ conversion operation. Detective _ This system, like a black-like look from the top of the car overlooking the car. This—the way to clearly display the actual situation around the car with the same-display face” is given to the driver. The driver can use the full-week image of this display to look down on the car, for example: when the car is judged, the H car and the material line and its obstacles (such as sidewalks, cars) are placed in a narrow position. When you are lanes, you can clear the bribes to avoid the collision accidents, so as to improve driving safety. Under the general environment, the brightness/color of the light around the car may vary greatly from 201123915. For example, the 'New Car is close to the wall material, and the brightness of the side of the near-touch of the automobile will be much lower than the brightness of the material. Scales, if the car four or more lenses use the relevant camera parameters for image_take, the images taken by several of the lenses may produce over-brightness. Therefore, the formed full-circumference image has an unevenness in brightness and a discontinuity in the composite image connecting the image of the image, so that it is difficult to recognize the object at the junction of the two images. Also, images that are too bright or too dark may make the details of objects in the image difficult to recognize. & In addition, the brake lights or warning lights on the rear of the car are all red. Therefore, the image taken by the lens behind the car and the image taken by the lens in front of the car will be so large in color that the image will be very unnatural when the full (four) image is synthesized. Similarly, when the difference in color captured between the lenses is too large, it is difficult for the processor to perform image composition correctly. SUMMARY OF THE INVENTION In view of the above problems, the present invention proposes a method for adjusting imaging parameters of a full-circumference image, which can reduce the imaging parameters of the entire image, such as brightness values or color values. Can tend to be consistent. The full-touch image-recorded female branch of the present invention includes: (4) a difference in imaging parameters of the image blocks of the adjacent image according to the image block of the target image located at the junction of the target image and the adjacent image, Calculating a first value of the target lens relative to the imaging parameter of the adjacent image; (B) calculating a second adjustment value of the imaging parameter of the target lens according to at least the first adjustment value; (Q repeating the step (A) and Step (B) 'Under the calculation of the camera parameters of all the lenses, the second adjustment value is to 201123915 () according to the scale of the scale - difficult value for the multiple mirror lion camera parameters. The target image is any of the multiple sub-images A scooping deer in the lens for capturing the target image (1) The 軚 lens is a mirror of a plurality of images, and the adjacent image is a plurality of sub-segments adjacent to the bet image in the relative configuration constituting the full-circle image. Image; In addition, the present invention can do the following '=: parameter respectively for at least - adjacent image of at least -: like two to, Ί difference, the target image of the different target lens is respectively relative to: The photo of the image The at least one first adjustment of the parameter... adjusts the value, calculates the second adjustment value of the target mirror imaging parameter; (C) adjusts the imaging parameter of the target lens according to the first adjustment value; and the tank execution = (4), step _ step (C), until the first § week integer value of the imaging parameters of all the lenses is calculated. In the above two methods, the recursive mechanism repeats the operation to make the connection of the two images connected to each other in the full-length image. The difference can be reduced or reduced. The recursive mechanism can be combined with the following two interrupt conditions: when the number of recursive times - the preset value or until the second adjustment value of the camera parameters of all the lenses is less than the predetermined value. The image block of the target image and the image block of the adjacent image may be completely overlapped, partially overlapped or adjacent to each other. The image parameter may be a brightness value or a color value. The second adjustment value may be an arithmetic mean of the first value or By means of the adjustment method of the imaging parameters of the full-circumference image of the present invention, the imaging parameters of the full-circumference image, such as brightness values or color values, can tend to be the same. The interconnected two 201123915 images The description of the present invention and the following embodiments are intended to demonstrate and explain the spirit and principles of the present invention, and to provide further scope of the patent application of the present invention. [Embodiment] The detailed features and advantages of the present invention are described in detail below in the yoke mode, which is sufficient for any skilled person to understand the technical contents of the present invention and to implement it according to the present specification. The content, the scope of the patent application and the drawings can be easily understood by those skilled in the relevant art. The following examples are in the detailed description of the difficulties of the Japanese invention, but the non-peak view limits the present. According to the embodiment of the present invention, the image of the toy moth is displayed in the embodiment of the present invention. In this embodiment, the aging body is disposed on the vehicle body (10), for example, the 』, phase module & is placed in the other pole, the rear bumper, the left rear view mirror, and the right rear view mirror Each setting - a lens 112, 114, 116, 118. These camera modules can capture a wide range of scene images, such as shooting areas 112a, 114a, intestines, and through lenses 112, 114, 116, 118. Please refer to "Fig. 2", and "Fig. 2" is a block diagram of the system of the present invention. The lens ii2, m, ii6, m smoke is connected to the image processing device 20. The lens 112' 114, 116, 11.8 transmits the captured sub-image to the image processing device 2A. Based on these sub-images, the processing device 2G can calculate the adjustment values of the individual imaging parameters of the classes ii2, ii4, 116, and 201123915. After that, a bird's-eye vehicle image is generated using the number of captured scene images, and the bird's-vehicle vehicle image is displayed on the screen. First, one of the plurality of camera modules is selected, and the camera module is designated as the target lens. The sub-image captured by the target lens is called the target image. In a full-circumference image, a sub-image connected to a target image is called an adjacent image. In general, a target image will have multiple adjacent images. Then, the image block of the target image and the image block of the adjacent image are defined according to the connection between the target image and the adjacent image. The image block of the target image and the image block of the adjacent image may completely overlap, partially overlap, or be adjacent to each other. Some of these imaging areas 112a, 114a, 116a, and 118a overlap with adjacent imaging areas, and these overlapping areas are U3b, 115b, ii7b, and 119b, respectively. Taking the lens 114 as an example, the lens 114 has two adjacent faces, which are the lens 112 and the frame 116, respectively. To adjust the imaging parameters of the lens 114, firstly, according to the image area captured by the lens and the overlapping area 113b, the calculation of the image parameters of the individual image blocks is defined; the two-phase domain group overlaps. The parameter adjustment value of the lens 114 is calculated by the parameter difference of the image block near the area. That is to say, based on the image captured by the adjacent camera module in the overlapping area, the first adjustment value is generated according to the difference of the image, and is used to adjust the number of springs of the image captured by the lens m. Then, the imaging parameters of the image block of the target image and the image of the image block of the adjacent image are taken. The upper meaning camera parameter can be the brightness value is the color value. More / 201123915 In detail, the camera parameters are the average of the brightness values of each pixel in the image block, or the average of the color values of each pixel. In addition, the camera parameter can also be the average of the camera parameters obtained at different time points in a period of time. However, when the shooting environment corresponding to each lens is different, the imaging parameters of each lens are supplemented so that the two imaging parameters of the two image blocks that are completely overlapped, partially overlapped, or adjacent to each other are not. Therefore, the difference between the imaging parameters of the image block of the target image and the imaging parameters of the image block of each adjacent image is calculated, thereby obtaining the first adjustment value of the target lens and each adjacent image. In other words, if the target image parameter is obtained by the obtained first-adjusted value, the imaging parameters of the respective image blocks at the connection between the acquired target image and the phase (four) image corresponding to the first-machine value tend to be, That is, the difference between the two imaging parameters of the two image blocks that are completely overlapped, partially overlapped, or adjacent to each other is different. And ^, to minimize the difference between the target image and all adjacent images, the average value of the first machine value can be calculated to obtain the binary value, which can be the arithmetic mean value weighted average 4 The average value is the value obtained by adding all the first adjustments and dividing by the number of the first adjustment values. The weighted average is obtained by multiplying the first adjustment value of the = by a weight and adding the total, and calculating the adjustment value of the target lens by taking the total of the sum of the first and the other camera sets as the target lens. Until the second adjustment value of all camera modules is calculated. Thereafter, a plurality of imaging parameters are adjusted in accordance with the second adjustment value. Here, using the recursive mechanism, the 201123915 method of performing the adjustment parameter is repeated repeatedly; t爿 using the (four) mechanism, by repeatedly performing the adjustment parameter, the mouth (four) is in the form of two The gap between the sub-images. And, by setting the handoff interrupt condition, the timing of the retraction termination is judged. When the number of times is up to the preset value, or all of the second adjustment values are less than %, the method is executed. Conversely, when it is judged that the number of recursions is still /: pre-time' or any of the second adjustment values is still greater than the preset value, the action of the last two values and the imaging parameters is continued. In the last recursive execution of each recursive (4) the total value plus the total or the difference between the camera parameters of the camera and the camera parameters after recursion, as the corresponding camera parameters for the students = for each lens The color value/brightness value between the images is such that the difference between the sub-images obtained by these lenses is reduced. The color value/luminance value of the image of the Wang Zhou image combined by these sub-images can be tended to be. The H-like image is a map of the nano-pruning according to the first embodiment of the present invention. Among them, the whole week is ridiculed by a number of sub-images. In the joint of the block adjacent image, the target image target image corresponds to the difference of the image of the image of the adjacent image, 201123915, and then calculates the average value of the fiscal-adjusted value obtained in the step (S11G) to obtain the second Adjust the value (S120). Next, it is judged whether or not the second adjustment value of all the lenses has been calculated (sl3〇). When it is judged that there is still no calculation, the target lens is replaced with a lens (S14) where the second adjustment value is not calculated and the step (sll〇) is repeated, and the step is the step Yangshuo. On the other hand, when it is judged that all the lenses have been calculated, the imaging parameters of the plurality of lenses are adjusted correspondingly according to the second adjustment value (S150). Then 'determine if the number of recursions is less than the preset value (316 〇). When it is judged that the number of recursive times is greater than the preset value, the end of the imaging parameter is light. When the number of times of lying back is less than the preset value, __breaks whether all the second values are less than or equal to the predetermined value (S170). Because the second adjustment value may be a positive number or a negative number, the preferred embodiment is to determine that all the second adjustments of the shirt are less than the predetermined value; or whether the maximum value of the absolute values of the second adjustment value is less than or equal to default value. When the second adjustment value of the value of the value towel is less than the preset value, the method can be stopped. Conversely, when the maximum value of the absolute values of the second adjustment value is greater than the preset value, the step (sn〇), the step (sl2〇), the step (si3〇), the step (S140), and the step (si5) are re-executed. 〇) and step (S160). Although the money is executed here, the steps (passes) and steps (the two recursive interrupt conditions of the grievance can be sequentially touched. That is, when any of the two reversal cap conditions is fulfilled, the execution is stopped. The method _, the gamma lang is not limited to the second. The execution order of the step (S160) and the step (S170) may also be reversed. Alternatively, in the step (S160) or the step (S170), only the execution may be performed. A recursive interrupt condition. 201123915 Solving the invention. Suppose that there are three lenses (the first lens, the second lens and the third lens) respectively capturing three sub-images (the first sub-image, the second sub-image and the third sub-image) The first sub-image, the second sub-image, and the third sub-image are combined into a full-circumference image. That is, the adjacent images of the first sub-image are the second sub-image and the third sub-image. The adjacent images are the first sub-image and the third sub-image. The adjacent images of the third sub-image are the first sub-image and the second sub-image.
在第-子影像中，與第二子影像相鄰的影像區塊的攝像參數 假設為。為了方便說明，在第一子影像中，與第二子影像相鄰 的影像區塊賴像錄以R{1，2}代表，也就是說叩，2卜⑵。此 外’在第-子影像中，與第三子影像相鄰的影像區塊的攝像參數 R{1，3}=140。在第二子影像中，盘第一旦 z、弟子衫像相鄰的影像區塊的 攝像參數R{2，1}=57 ’與第:r早旦彡應士B抑 —子4相鄰的影像區塊的攝像參數 R{2,3}-55。在第二子影像中，*第一 ，、弟千〜像相鄰的影像區塊的攝 像參數R{3，1}=86，盥第-不旦/你决 R{3,2}=89。 〃—子4轉的影倾塊的攝像參數 首先’以第一子影像為目禅爭傻 π曰知衫像。第一子影像對於 像的第一調整值S{1，2}，伤η Ώ —子尽 糸為(R{2,l}-R{1，2})/2，也就是a， 一子影像對於第三子影傻 ^ 丁〜像的第一調整值S{1 (R{3，l}-R{l，3})/2，也就是_27。 ’、*' 接著，計算第一子影像的第_ ，士上 弟一㉟正值。根據上述，筮-啼屯 值可為第一職賴卿⑽驗料。此處= 201123915 為算術平均值進行說明。第一子影像的第二調整值Τ{1}係為S{1，2} 與S{1，3}的算術平均，也就是丁⑴二如。 接著，以第二子影像為目標影像，並計算第二子影像的第— 調整值S{2，1}與S{2,3}。根據與第一子影像相同的方法可得 叩，1卜32及S{2,3卜17。之後，再計算SR1}與S(2,3}的算術平 均，也就是T{2}=25。 最後，以第三子影像為目標影像，並計算第三子影像的第一 調整值Sftim 。根據與第一子影像相同的方法，可得魯 S{3，1}=27及S{3,2}=-17。之後’再計算扣…與叩，3}的算術平 均，也就是T{3}=5。 當在計算完上述三個子影像之後，即根據第二調整值更新鏡 頭的攝像參數。在攝像參數被更新後，可得新的攝像參數 R{1，2}=9卜 R{1，3}=110、卜82、R{2 3}=8〇、r⑽=91 且 R{3,2}=94。 接著，重複利用上述的方法進行遞迴運算。為了簡化起見， 計算所得的數值以底下的表格表示之。In the first sub-image, the imaging parameters of the image block adjacent to the second sub-image are assumed to be. For convenience of explanation, in the first sub-image, the image block adjacent to the second sub-image is recorded by R{1, 2}, that is, 叩, 2 (2). Further, in the first sub-image, the imaging parameter R{1, 3}=140 of the image block adjacent to the third sub-image. In the second sub-image, the disc is once z, the disciple's shirt is like the image parameter R{2,1}=57 ' of the adjacent image block and the first: r: The imaging parameters of the image block are R{2, 3}-55. In the second sub-image, *first, brother thousand ~ like the image parameter R{3,1}=86 of the adjacent image block, 盥第-不旦/你决R{3,2}=89 . The camera parameters of the shadow-dumping of the 〃-sub 4 turns First, the first sub-image is used as a meditation for idiots. The first sub-image is for the first adjustment value S{1, 2} of the image, and the η Ώ 子 子 子 is (R{2, l}-R{1, 2})/2, that is, a, one child The first adjustment value S{1 (R{3, l}-R{l, 3})/2 of the image for the third sub-shadow is the same as the image, that is, _27. ‘,*' Next, calculate the _th of the first sub-image, and the positive value of the priest-35. According to the above, the 筮-啼屯 value can be the first job of Lai Qing (10). Here = 201123915 describes the arithmetic mean. The second adjustment value Τ{1} of the first sub-image is the arithmetic mean of S{1, 2} and S{1, 3}, that is, D (1) is as follows. Next, the second sub-image is used as the target image, and the first adjustment values S{2, 1} and S{2, 3} of the second sub-image are calculated. According to the same method as the first sub-image, 卜, 1 卜 32 and S{2, 3 卜 17 can be obtained. Then, calculate the arithmetic mean of SR1} and S(2,3}, that is, T{2}=25. Finally, the third sub-image is used as the target image, and the first adjustment value Sftim of the third sub-image is calculated. According to the same method as the first sub-image, you can get Lu S{3,1}=27 and S{3,2}=-17. After that, 'calculate the arithmetic mean of deduction...and 叩,3}, that is, T {3}=5. After the above three sub-images are calculated, the imaging parameters of the lens are updated according to the second adjustment value. After the imaging parameters are updated, a new imaging parameter R{1, 2}=9 is obtained. R{1,3}=110, Bu 82, R{2 3}=8〇, r(10)=91 and R{3,2}=94. Next, repeat the above method for recursive operation. For the sake of simplicity , the calculated value is expressed in the table below.
J、於 R{1，2} 初始值 121 第一次遞迴 91 第二次遞迴 84 第三次遞迴 82 假設遞_ +斷條件為第二 12 201123915 苐二次遞迴 ^ H)1 89 87 93 96 假設遞迴的中斷條件為第二調整值的絕對值的最大值小於 5。在第三次遞迴之後，三個第二調整值分別為T{1}=_2 、Τ{2}=1 且Τ{3}=1，也就是已達到中斷條件。此時，即停止此方法。 從上面的表格可看出，R{1，2}與R{2，1}的初始值差距為 m_57=64。而再經過三次的遞迴運算後，R{1，2}與叩，^的差距 為89-82=7。也就是說’完全重疊、部分重疊或是彼此鄰接的二個 影像區塊的攝像參數之間的差異明顯的被縮小。 另一方面，若是平均值為加權平均值，假設在計算T{〗丨時， R{1，2}的權重為0.45，且R{1，3}的權重為〇 55，則1{1}為_29。在 冲算T{2}時，假設R{2,1}的權重為〇 %，且R{2,3} _重為〇 65， 則T{2}為22。在計算T{3}時，假設:^”丨的權重為〇 45，且R{3,2} 的權重為0.55，則T{3}為3。 當在計算完上述三個子影像之後，即根據第二調整值更新鏡 頭的攝像參數。在攝像參數被更新後’可得新的攝像參數 R{U}=92、R{1，3卜⑴、R{21}=79、R{2 3}=77、R{3 i}=89 且 R{3,2}=92。 運算 接著，重複利用上述的方法以及上述的權重進行遞迴 ~~--—- R{1,2} R{1,3} 初始值 —--- 121 140 57 R{2,3} R{3，1} R{3,2} 55 86 89 13 201123915 第一次遞迴 92 —~^ 111 79 77 89 92 第二次遞迴 83 102 86 84 90 93 第三次遞迴 80 99 —-- 88 86 90 ------ 93 假設遞迴的中斷條件為第二調整值的絕對值的最大值小於 5。在第三次遞迴之後，三個第二調整值分別為 且T{3}=0，也就疋已達到中斷條件。此時，即停止此方法。 從上面的表格可看出，R{1，2}與R{2，1}的初始值差距為 121-57=64。而再經過三次的遞迴運算後，尺{1,2丨與尺卩，”的差距 為88-80=8。也就是說，完全重疊、部分重疊或是彼此鄰接的二個 影像區塊的攝像參數之間的差異明顯的被縮小。 此外’可在异㈣二調整值之後，即接著根據第二調整值調 整對應之目標鏡頭的攝像參數。當調整完目標鏡頭的攝像參數 後’目標鏡頭才會更換為其他的鏡頭。請參照『第4圖』，係為本 發明之第二實關之流糊。此全郷像賴像參數賴整方法 包括：根據位於目標影像與補影像的連接處中目標影像的影像 區塊分別對相鄰影像的影像區塊的攝像參數的差異，分別計算目 標鏡頭的目標影像對應相鄰影像的攝像參數的第—調整值 (S210)彳算第—調整值的平均值以制第二調整鄉调、根據 第二調整值對應調整目標鏡頭的攝像參數(S230)以及判斷是否全 部鏡頭的第一調整值都已計算(S240)。 田判斷尚有鏡頭未計算時，更換目標鏡頭為未計算第二調整 值的鏡頭(S25〇)，並重複執行步驟⑻1〇)、步驟(S220)、步驟(S230) 201123915 預认值(S260)。當判斷遞迴次數大於預設值時，即結束攝像參數的 調整。當判斷遞迴次數小於預設值時，則繼續判斷是否全部第二 调正值（即對應各個鏡頭的第二調整值）皆小於或等於預定值 (7〇)田判斷任一第二調整值大於預定值時，則重新執行步驟 (S210)步驟(S22〇)、步驟阳〇)、步驟_〇)、步驟卿〇)與步驟 (S260)。 在此，同樣假設叩^卜⑵、叩，3卜⑽、叩，㈣7、 R{2,3}=55、R{3,1}=86 且 r(3 2卜89。 首先乂第一„周整值為第一調整值的算術平均做說明。在第一 遞k中位於5周整後的第一子影像的兩側的影像區塊的攝像參 /別為R{1，2}=91及R{1,3}=11〇。對於第二子影像而言，在計 =與第-子影像相關的第—調整值叩，1}時，會以調整後的第一 ^像的攝像參離{1，2㈣)進行計算。接著，再根據第一調整 調整第一}===第一調整值聊並且以第二調整值耶 4-鏡頭的攝像參數。調整後的叩，1}=74及卿卜a。之 像：根據:?後的第一子影像與第二子影像計算第丄鏡頭的攝 像參數，調整後可得叩，1}，及印，2}哨。 R{1,2} μ…Π Γ—--J, at R{1, 2} Initial value 121 First recursive 91 Second recursive 84 Third recursive 82 Assume that the _ + break condition is the second 12 201123915 苐Second recursive ^ H)1 89 87 93 96 Assume that the recurring interrupt condition is that the absolute value of the second adjusted value has a maximum value of less than 5. After the third recursion, the three second adjustment values are T{1}=_2, Τ{2}=1, and Τ{3}=1, that is, the interrupt condition has been reached. At this point, stop this method. As can be seen from the above table, the initial value difference between R{1, 2} and R{2, 1} is m_57=64. After three recursive operations, the difference between R{1, 2} and 叩, ^ is 89-82=7. That is to say, the difference between the imaging parameters of the two image blocks which are completely overlapped, partially overlapped or adjacent to each other is remarkably reduced. On the other hand, if the average value is a weighted average, it is assumed that when calculating T{〗, the weight of R{1, 2} is 0.45, and the weight of R{1, 3} is 〇55, then 1{1} For _29. When calculating T{2}, it is assumed that the weight of R{2,1} is 〇%, and R{2,3} _ is 〇65, then T{2} is 22. When calculating T{3}, suppose that the weight of ^"丨 is 〇45, and the weight of R{3,2} is 0.55, then T{3} is 3. After calculating the above three sub-images, Updating the imaging parameters of the lens according to the second adjustment value. After the imaging parameters are updated, the new imaging parameters R{U}=92, R{1, 3b(1), R{21}=79, R{2 3 are available. }=77, R{3 i}=89 and R{3,2}=92. The operation is then repeated using the above method and the above weights to recurse ~~---- R{1,2} R{ 1,3} Initial value—121 210 57 57 R{2,3} R{3,1} R{3,2} 55 86 89 13 201123915 First recurs 92 —~^ 111 79 77 89 92 Second recursive 83 102 86 84 90 93 Third recursive 80 99 —-- 88 86 90 ------ 93 Assume that the recurring interrupt condition is that the absolute value of the second adjustment value is less than 5 After the third recursion, the three second adjustment values are respectively and T{3}=0, that is, the interrupt condition has been reached. At this point, the method is stopped. As can be seen from the above table, R The difference between the initial value of {1,2} and R{2,1} is 121-57=64. After three recursive operations, the difference between the ruler {1,2丨 and the ruler,” It is 88-80=8. That is to say, the difference between the imaging parameters of the two image blocks which are completely overlapped, partially overlapped or adjacent to each other is remarkably reduced. Furthermore, the imaging parameters of the corresponding target lens can be adjusted after the different (four) two adjustment values, that is, according to the second adjustment value. When the camera parameters of the target lens are adjusted, the target lens will be replaced with another lens. Please refer to "Figure 4" for the second real thing of this invention. The method for determining the image of the full image includes: calculating the target of the target lens according to the difference of the imaging parameters of the image block of the adjacent image in the image block of the target image at the connection between the target image and the complementary image. The image corresponds to the first adjustment value of the imaging parameter of the adjacent image (S210), calculates the average value of the first adjustment value to make the second adjustment, adjusts the imaging parameter of the target lens according to the second adjustment value (S230), and determines Whether the first adjustment value of all the lenses has been calculated (S240). When the field judges that the lens is not calculated, the target lens is replaced with a lens that does not calculate the second adjustment value (S25〇), and the steps (8) and 1) are repeated, and the step (S220) and the step (S230) 201123915 are recognized (S260). . When it is judged that the number of recursive times is greater than the preset value, the adjustment of the imaging parameters is ended. When it is determined that the number of recursive times is less than the preset value, it is further determined whether all the second adjusted values (ie, the second adjusted values corresponding to the respective shots) are less than or equal to a predetermined value (7〇), and any second adjusted value is determined. When it is greater than the predetermined value, the step (S210), the step (S22), the step (step), the step (step), and the step (S260) are re-executed. Here, the same assumption is made that 叩^(2), 叩, 3 (10), 叩, (4) 7, R{2, 3}=55, R{3, 1}=86 and r(3 2b 89. First 乂 first „ The weekly integer value is an arithmetic mean of the first adjustment value. The imaging parameter of the image block located on both sides of the first sub-image after 5 weeks in the first recursion k is R{1, 2}= 91 and R{1,3}=11〇. For the second sub-image, when the first adjustment value 叩, 1} associated with the first sub-image is used, the adjusted first image is The camera is separated from {1, 2 (four)) for calculation. Then, according to the first adjustment, the first}=== first adjustment value is talked and the second adjustment value is used to adjust the camera parameters of the lens. }=74 and Qingbu a. Image: According to the first sub-image and the second sub-image after the calculation, the camera parameters of the second lens are calculated. After adjustment, you can get 叩, 1}, and India, 2) whistle. {1,2} μ...Π Γ—--
彳打表之結果 15 201123915 第二次遞迴 [ΊΤ~ 第三次遞迴 97 81 79 83 81 88 88 91 91 大值小於 5。在第三次遞迴之後’三個第二調整值分別為叩卜3、印㈣ 且T{3}=0，也就是已達到中斷條件。此時，即停止此方法。The result of the beating table 15 201123915 The second recursion [ΊΤ~ The third recursion 97 81 79 83 81 88 88 91 91 The large value is less than 5. After the third recursion, the three second adjustment values are 叩3, 印(四), and T{3}=0, that is, the interruption condition has been reached. At this point, stop this method.
從上面的表格可看出，叩,2}與RR1丨的初始值差距為 121_57=64。而再經過三次的遞迴運算後’叩，2齡糾的差距 為83-78=5。也就是說，完全重疊、部分重疊或是彼此鄰接的二個 影像區塊的攝像參數之間的差異明顯的被縮小。 另一方面，若是平均值為加權平均值，則遞迴計算的結果如 下表所述。 '*-------- R{1，2} R{1，3} R{2，1} R{2,3} R{3，1} R{3,2} 初始值 121 140 57 55 86 89 第一次遞迴 92 111 74 72 87 90 第二次遞迴 81 100 81 79 87 90 第三次遞迴 77 96 84 82 87 90 一 ----- 此時計算所用的權重與前述的例子相同。As can be seen from the above table, the initial value difference between 叩, 2} and RR1丨 is 121_57=64. After three more recursive operations, the gap between the two ages is 83-78=5. That is to say, the difference between the imaging parameters of the two image blocks which are completely overlapped, partially overlapped or adjacent to each other is remarkably reduced. On the other hand, if the average is a weighted average, the result of the recursive calculation is as described in the table below. '*-------- R{1,2} R{1,3} R{2,1} R{2,3} R{3,1} R{3,2} initial value 121 140 57 55 86 89 First rewind 92 111 74 72 87 90 Second recursive 81 100 81 79 87 90 Third re-entry 77 96 84 82 87 90 One----- The weight used in the calculation The foregoing examples are the same.
假設遞迴的中斷條件為第二調整值的絕對值的最大值小於 5。在第三次遞迴之後’三個第二調整值分別為T⑴=_4、T{2}=3 且Τ{3}=〇 ’也就是已達到中斷條件。此時，即停止此方法。 從上面的表格可看出，R{1，2}與R{2，1}的初始值差距為 121-57=64。而再經過三次的遞迴運算後’ 丨與的差距 16 201123915 為84-77=7。也就是言兒，完全重疊、部分重疊或是彼此鄰接的二個 影像區塊的攝像參數之間的差異明顯的被縮小。 藉由本發明實關所提出的方法，可獅於汽車四周的全周 影像的影像處理。從上述的數據可看出，相互連接的二影像之間 的連接處的差異可湘_削、。也紋，本發啊使此全周影 像的攝像參數，例如亮度值或色彩值，能趨於一致。It is assumed that the recurring interrupt condition is that the maximum value of the absolute value of the second adjustment value is less than 5. After the third recursion, the three second adjustment values are T(1) = _4, T{2} = 3, and Τ {3} = 〇 ', that is, the interrupt condition has been reached. At this point, stop this method. As can be seen from the above table, the initial value difference between R{1, 2} and R{2, 1} is 121-57=64. And after three recursive operations, the gap between the 16 and 16 16 201123915 is 84-77=7. That is to say, the difference between the imaging parameters of the two image blocks that are completely overlapped, partially overlapped, or adjacent to each other is significantly reduced. By the method proposed by the present invention, the image processing of the whole week image of the lion can be performed. As can be seen from the above data, the difference in the connection between the two connected images can be reduced. Also, this hair makes the camera parameters of this full-length image, such as brightness value or color value, tend to be consistent.
雖然本發明以前述之實施例揭露如上，並翻以限定本 發明。在不脫離本發明之精神和範_，所為之更動與潤飾，均 屬本發明之專娜護細。齡本發賴界定之倾朗請參考 所附之申請專利範圍。 ^ 【圖式簡單說明】 第1圖係為根據本發明實施例之顯示鳥瞰車輛影像的照相模 組裝設位置及拍攝範圍之鳥瞰示意圖。 …、果 第2圖係為根據本發明之系統方塊示意圖。 第3圖係為根據本發明之第一實施例之全周影像的攝像表數 的調整方法之流程圖。 第4圖係為根據本發明之第二實施例之全周影像的攝像參數 的調整方法之流程圖。 【主要元件符號說明】 車身 車輛周圍 鏡頭 110 110a 112、114、116、118Although the invention has been disclosed above in the foregoing embodiments, the invention is defined. Without departing from the spirit and scope of the present invention, the modifications and retouchings are all the details of the present invention. Please refer to the attached patent application scope for the definition of the age limit. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a bird's-eye view of a camera module assembly position and a shooting range for displaying a bird's-eye view vehicle image according to an embodiment of the present invention. Fig. 2 is a block diagram of a system according to the present invention. Fig. 3 is a flow chart showing a method of adjusting the number of cameras of the full-circumference image according to the first embodiment of the present invention. Fig. 4 is a flow chart showing a method of adjusting imaging parameters of a full-circumference image according to a second embodiment of the present invention. [Main component symbol description] Body Vehicle around the lens 110 110a 112, 114, 116, 118
17 201123915 112a ' 114a、116a、118a 拍攝區域 113b、115b、117b、119b 重疊的區域 20 影像處理裝置17 201123915 112a ' 114a, 116a, 118a Areas where the imaging areas 113b, 115b, 117b, 119b overlap 20 Image Processing Unit
1818