200529092 玖、發明說明: 【發明所屬之技術領域】 本發明係關於一種從車輛影像中擷取車牌區域的方法,特別是可 應用於車牌自動辨識糸統及父通監控、車輛門禁的車牌區域影像之切 取。 【先前技術】 .按,車輛影像中之車牌區域的定位及切岐車牌自動觸的前置 處理程序,因此其準確性影響車牌自動辨識的整體性能;此外,在交 通監控上乃至於車㈣禁管理上,管理人„要透職賴及顯示器 觀看來往車輛的車牌號碼,若能自動將車祕_影像娜並顯示出 來’將可方便m員觀看。因此,如何自動且有效的從車輛影像中 找出車牌的位置所在、切取出來,便是車牌自動辨識系統及交通監控、 車輛門禁等應用的重要課題。 先前技術對於車輛影像中之車牌區域蚊位及姆,—般係使用 像素明暗度(IntensitM^g算處理並將所有像素之運算結果的可能 情形繪出柱狀圖(Histog·)再從柱狀圖中計算出欲將所有像素之運 算結果二值化所需的臨界值(ThreshGld),所以很容易因柱狀圖上下起 落無跡可循而無法找到合適之臨界值以致二值化效果不佳而影響車牌 &域定位及切取的準確性;此外,由於先前技術多未對可能的車牌區 域分析其高值像素的分布情形,以致找出錯誤之車牌區域的機會較大。 【發明内容】 本發明之目的即在於提供一種從車輛影像中擷取車牌區域的方 200529092 法’可應用於車牌自動辨識及交通監控、車輛門禁。 本發明主要係使用像素之明暗度(丨时⑼以办)或對數明暗度 (Logarithmic intensity)來運算處理並將所有^象素之運算結果的可能 情形繪出累積柱狀圖(Cumulative histogram),並從累積柱狀圖中計 算出將所有像素之運算結果二值化所需的臨界值(Thresh〇ld);因為累 積柱狀圖係呈遞增上升狀態而不會有一般柱狀圖上下起落無跡可循的 現象,較易計算出欲將所有像素之運算結果二巧化所需的臨界值,本 發明以逼近法獲得累積柱狀圖遞增上升轉折或趨緩最明顯的—處做為_ 一值化的臨界值,可得到較佳的二值化結果。 此外,本發明對每個車牌候選區域分別切成數個小區塊並分析所 有小區塊所包含之高值像素的平均個數及標準差(Standard deviation),車牌候選區域之區塊高值個數標準差太高者,可被剔除 而不纟忍疋為車牌區域。 【實施方式】 茲為便於貴審查委員能更進一步對本發明之構造、使用及其特鲁 徵有更深一層,明確、詳實的認識與瞭解,發明人舉出較佳之實施例, 配合圖式洋細說明如下: 首先請參閱圖一所示係本發明之實施例流程圖,主要係由以下步 驟所組成: (1)計算水平梯度1 :計算車輛影像之水平梯度(H〇riz〇ntai gradient)來產生水平梯度影像;其計算水平梯度的方式係將相鄰的像 素之明暗度相減或將明暗度取對數後成為對數明暗度再相減。 200529092 ⑵水平梯度二值化2 :接著將水平梯度影像扣二值化成水平梯 度二值化影像使得影像中每個像素的數值只為高值㈣或低 值(Low value) ’可行的做法是計算水平梯度影像之所有可能之水平梯 度值的出現次數以繪出水平梯度值的累積柱狀圖,累積柱狀圖的橫轴 是所有水平做值由小到大依序排列,縱軸是對每—斜梯度值而言 小於或等於該水平梯度值的累積出現次數,在最低水平梯鱼最言 水平梯度值之間找-水平梯度值,使得以最低水平梯度值的出現次" 數、小於鱗於此水平梯度值的累積出現次數、小於鱗於最大水平 梯度值的累積出現-人數二者戶斤構成的線性折線與水平梯度值累積柱狀 圖的差距最小(即最ϋ近累積柱狀圖),將此水平梯度值定為水平梯度 值的二值化轉值並將大於臨界制水平梯度蚊為高值、小於或等 於臨界值的水平梯度值定為低值。 (3) 計算垂直梯度3:計算車輛影像之垂直梯度(Verticalgradient) 來產生垂直梯度影像;其計算垂直梯度的方式係將相鄰的像素之明暗 度相減或將明暗度取對數後成為對數明暗度再相減。 (4) 垂直梯度二值化4 ··並將垂直梯度影像予以二值化成垂直梯度 二值化影像使得影像中每個像素的數值只為高值或低值;做法是計算 垂直梯度影像之所有可能之垂直梯度值的出現次數以繪出垂直梯度值 的累積柱狀圖’累積柱狀圖的橫軸是所有垂直梯度值由小到大依序排 列’縱軸是對每一垂直梯度值而言小於或等於該垂直梯度值的累積出 現人數’在最低垂直梯度值與最高垂直梯度值之間找一垂直梯度值, 200529092 使得以最低垂直梯度值的出現次數、小於或等於此垂直梯度值的累積 出現次數、小於或等於最大垂直梯度值的累積出現次數三者所構成的 線性折線與垂直梯度值累積柱狀圖的差距最小(即最逼近累積柱狀 圖),將此垂直梯度值定為垂直梯度值的二值化臨界值並將大於臨界值 的垂直梯度值定為高值、小於或等於臨界值的杳直梯度值定為低值。 (5) 合併水平梯度二值化影像與垂直梯度二值化影像5 :將水平梯 度二值化影像及垂直梯度二值化影像合併成整合梯度二值化影像,可 行做法為將水平梯度二值化影像及垂直梯度二值化影像做”⑽"運 算,亦即水平梯度二值化影像與垂直梯度二值化影像中同一座標位置 的兩個像素只要有一個是呈現高值,則整合梯度二值化影像中同一座 標位置的像素亦呈現高值;若水平梯度二值化影像與垂直梯度二值化 影像中同一座標位置的兩個像素皆呈現低值,則整合梯度二值化影像 中同一座標位置的像素亦呈現低值。 (6) 將鄰近的高值像素群聚成車牌候選區域6 :將整合梯度二值化 影像中鄰近的高值像素群聚成車牌候選區域(Candidate regi〇n) ··可 行做法為整合梯度二值化影像中之高值像素間的距離若小於一特定 值,則將這些咼值像素標示相同之標籤,再將相同標籤之所有像素的 所在區域定為車牌候選區域。 (7) 以高值像素的分布情形來確定車牌候親域是否為車牌區域 7 :將每個|牌候選區域依照其區域内之高值騎的分布情形來確定是 否為車牌區域:圖三所示的是將每個車牌候選區域分別切成數個小區 200529092 塊並分析所有小區塊所包含之高值像素的平均她及標準差的實施例 示意圖,車牌候選區域之區塊高值個數標轄太高者,可被剔除而不 涊定為車牌區域,並產出確定後的車牌區域影像。 本創作之另-實施例,如圖四_,可只針對車姉像的水平梯 度來二值化影像,主要包含以下步驟: ⑴計算水平梯度丨:計算車姉像之水平梯⑽⑽歷加 gradient)來產生水平梯度影像;其計算水平梯度的方式係將相鄰的像 素之明暗度械或㈣暗度取後成鱗數明暗度再相減。 ⑵水平梯度二值化L接祕水平影像付二值化成水平梯 度二值化f彡像使得影像巾每轉素的數H高值(Η_㈤此)或低 值(Low Value) ’可行的做法是計算水平梯度影像之所有可能之水平梯 度值的出現次數以—水平梯度值的累積柱狀圖,累積柱狀圖的橫轴 是所有水平梯度值由小到大依序排列,縱軸是對每一水平梯度值而古 小於或等於該水平做值的累《現次數,在最财平做值與最高 水平梯度值之職-水平梯度值,使得以最低水平梯度值的出現次 數、小於或等於此水平梯度值的累積出現次數、小於或等於最大水平 梯度值的累《歡數三者賴成的_折線與水平梯度值累積柱狀 圖的差距最小(即最逼近累積柱狀圖),將此水平梯度值定為水平梯度 值的二值減界值麟大舰界值的水平梯度值定為高值、小於或等 於臨界值的水平梯度值定為低值。 (3)將鄰近的高值像素群聚成車牌候選區域6 ··將整合梯度二值化 200529092 影像中鄰近的高值像素群聚成車牌候選區域(Candidate region);可 行做法為整合梯度二值化影像中之高值像素間的距離若小於一特定 值’則將這些高值像素標示相同之標籤,再將相同標籤之所有像素的 所在區域定為車牌候選區域。 (4)以高值像素的分布情形來確定車牌候選區域是否為車牌區域 7:將每個車牌候選區域依照其區域内之高值像素的分布情形來確定是 否為車牌區域:圖三所示的是將每個車牌候選區域分別切成數個小區 塊並分析所有小區塊所包含之高值像素的平均個數及標準差的實施例翁 示意圖,車牌候選區域之區塊高值個數標準差太高者,可被剔除而不 認定為車牌區域,並產出確定後的車牌區域影像。 本創作之又一實施例,如圖五所示,可只針對車輛影像的垂直梯 •度來二值化影像,主要包含以下步驟: (1)计算垂直梯度3· ό十算車輛影像之垂直梯度(ver^cai gra(jient) 來產生垂直梯度影像;其計算垂細度財式係將相鄰的像素之明暗 度相減或將明暗度取對數後成為對數明暗度再相減。 修 ⑵垂直梯度二值化4 :將垂直梯度影像料二值化成垂直梯度二 值化影像使得影像中每個像素的數值只為高值或低值;做法是計算垂 直梯度衫像之所有可能之垂直梯度值的出現次數以繪出垂直梯度值的 累積柱狀圖,累積柱狀圖的橫軸是所有垂直梯度值由小到大依序排 列’縱軸是對每-垂直梯度值而言小於或等於該垂直梯度值的累積出 見人數在最低垂直梯度值與最高垂直梯度值之間找一垂直梯度值, 10 200529092 使得以最低垂直梯度值的出現次數、小於或等於此垂直梯度值的累積 出現次數、小於或等於最大垂直梯度值的累積出現次數三者所構成的 線性折線與垂直梯度值累積柱狀圖的差距最小(即最逼近累積柱狀 圖),將此垂直梯度值定為垂直梯度值的二值化臨界值並將大於臨界值 的垂直梯度值定為高值、小於或等於臨界值的垂直梯度值定為低值。 (3)將鄰近的高值像素群聚成車牌候選區域6 ••將整合梯度二值化 影像中鄰近的高值像素群聚成車牌候選區域(Candidate邮⑻,·可 打做法為整合梯度二值化影像巾之高值像素間的距離若小於一特定 值,則將高值像素標示相同之誠,再斜_職之所有像素的 所在區域定為車牌候選區域。 ⑷以高值像素的分树形來確定轉闕區域是料車牌區域 7 :將每個車牌候選區域依照其區域内之高值像素的分布情形來確定是 否為車賴域··圖三所稀是將每解義舰域㈣切成數個小區 塊並分析财小區朗包含之高值像麵平均錄及辟差的實施例 不意圖 車牌候選區域之區塊高值個數標衫太高者,可被剔除而不 認定為車賴域,並產$較後的車魏域影像 本發明所提供之從車輛影料擷取車牌區域的方法,與其它習用 技術相互比較時,具有以下的優點: 1·以逼近法獲得累積柱狀圖遞增上 、一 上开轉折或趨緩最明顯的一處做 為二值化的臨界值,可得雜麵f彡像二值化結果。 2.對車牌簡區域分析其高值像素齡布情形,彳匈錯誤之車牌 200529092 區域的機會較小。 惟以上所述者,僅為本發明之較佳實施例而已,當不能以此限定 本發明實施之範圍’即大凡依本發明申請專利範圍及發明說明書内容 所作之簡單的等效變化與料,皆應仍屬本發明專利涵蓋之範圍内。 系不由上對本發明之技術結構詳細揭述,本發明之構造確實能達到 獲知較佳之影像二值化效果的目的,且未見於業界相關產品當中,應 符口發明專利之要件’中請人爰依專利法之規定,肖鈞局提起發明 專利之巾請,Μ請早日_本料利5實感德惠。 【圖式簡單說明】 凊參閱以下有關本發明—較佳實施例之詳細說明及其_,將可 進-步瞭解本發明之技術内容及其目的功效;有關該實施例之附圖為: 圖-為本發明從車輛影像中擷取車牌區域的方法之實施例流程 圖;以及 ..冑二為本發明從水平梯度值(或垂直梯度值)的累積柱狀圖得出水 平梯度值(讀飾度值)之二值化臨界值的實施_意圖;以 及 圖三為本發明對每個車牌候選區域分別切成數個小區塊並分析所 有小區塊所包含之高值像素的平均健及標準差的實施例示意 圖。 圖四為本㈣從車姉像的水平梯度來擷取車牌區_方法之實 施例流程圖; 圖五為本發明從車輛影像的垂直梯度來擷取車牌區域的方法之實 i2 200529092 施例流程圖; 【圖式中之符號說明】 1 一計算水平梯度 2— 水平梯度二值化 3— 計算垂直梯度 4— 垂直梯度二值化 5— 合併水平梯度二值化影像與垂直梯度二值化影像 6— 將鄰近的高值像素群聚成車牌候選區域 7— 以高值像素的分布情形來確定車牌候選區域是否為車牌區域200529092 发明 Description of the invention: [Technical field to which the invention belongs] The present invention relates to a method for capturing a license plate area from a vehicle image, and is particularly applicable to a license plate area image for automatic license plate identification system, parental monitoring, and vehicle access control. Take it. [Previous technology]. Press, the positioning of the license plate area in the vehicle image and the pre-processing program for the automatic touch of the Qiqi license plate, so its accuracy affects the overall performance of automatic recognition of the license plate; in addition, in traffic monitoring and even car bans In terms of management, the manager „requires the license plate number of the passing vehicle through the monitor and the monitor. If the car secret_image is automatically displayed and displayed ', it will be convenient for m members to watch. Therefore, how to automatically and effectively from the vehicle image Finding out the location of a license plate and cutting it out are important topics for applications such as automatic license plate recognition systems, traffic monitoring, and vehicle access control. The previous technology generally used pixel lightness and darkness for license plate areas in vehicle images. IntensitM ^ g calculates and draws a histogram (Histog ·) for the possible situation of the calculation results of all pixels, and then calculates the critical value (ThreshGld) required to binarize the calculation results of all pixels from the histogram. , So it is easy to affect the license plate & domain because the histogram has no trace up and down and no suitable threshold can be found. Location and cutting accuracy; moreover, because the prior art has not analyzed the distribution of high-value pixels for possible license plate areas, there is a greater chance of finding the wrong license plate area. [Abstract] The purpose of the present invention is to Provide a method for capturing license plate area from vehicle image. 200529092 method can be applied to automatic license plate recognition, traffic monitoring, and vehicle access control. The present invention mainly uses the lightness and darkness of pixels (logarithm) or logarithmic lightness (Logarithmic intensity) to calculate and plot the possible situation of the operation results of all ^ pixels into a Cumulative histogram, and calculate the threshold required to binarize the operation results of all pixels from the cumulative histogram Value (Thresh〇ld); because the cumulative histogram is increasing, there is no phenomenon that the general histogram can go up and down and there is no trace. It is easier to calculate the need to double the calculation results of all pixels. The critical value of the present invention is obtained by the approximation method of the cumulative histogram with the most obvious ascending turning or slowing-off. In addition, the present invention cuts each license plate candidate area into several small blocks and analyzes the average number and standard deviation of high-value pixels contained in all the small blocks. The license plate candidate area Those blocks with high standard deviations that are too high can be eliminated without being a license plate area. [Embodiment] To facilitate your review committee to further the structure, use of the invention and its Trude sign A deeper, clearer and more detailed understanding and understanding, the inventor cites a preferred embodiment, which will be described in detail with reference to the figure: First, please refer to FIG. 1 which is a flowchart of an embodiment of the present invention, which is mainly caused by the following steps. Composition: (1) Calculate horizontal gradient 1: Calculate the horizontal gradient of the vehicle image to generate a horizontal gradient image; the method of calculating the horizontal gradient is to subtract the lightness and darkness of adjacent pixels or reduce the lightness and darkness. After taking the logarithm of the degree, it becomes the logarithmic lightness and darkness and then subtracts. 200529092 ⑵Horizontal Gradient Binarization 2: Subsequent binarization of the horizontal gradient image into a horizontal gradient binarized image so that the value of each pixel in the image is only a high value ㈣ or a low value (Low value) 'The feasible way is to calculate The number of occurrences of all possible horizontal gradient values of the horizontal gradient image is used to draw a cumulative histogram of horizontal gradient values. The horizontal axis of the cumulative histogram is that all horizontal values are arranged in order from small to large, and the vertical axis is —The cumulative number of occurrences of the oblique gradient value that is less than or equal to the horizontal gradient value. Find the horizontal gradient value between the lowest horizontal gradient and the lowest horizontal gradient value, so that the number of occurrences of the lowest horizontal gradient value is less than, The cumulative number of occurrences of this horizontal gradient value is less than the cumulative occurrence of the maximum horizontal gradient value-the number of households with a linear polyline and the horizontal gradient value cumulative histogram has the smallest difference (that is, the closest cumulative column) State diagram), set this horizontal gradient value as the binary conversion value of the horizontal gradient value, and set the horizontal gradient value that is higher than the critical horizontal gradient mosquito to be high and less than or equal to the critical value to low value. (3) Calculate the vertical gradient 3: Calculate the vertical gradient of the vehicle image to generate a vertical gradient image; the method of calculating the vertical gradient is to subtract the lightness and darkness of adjacent pixels or take the logarithm to the logarithm. Degrees are subtracted again. (4) Binary Vertical Gradient 4 ... Binary the vertical gradient image into a vertical gradient binarized image so that the value of each pixel in the image is only a high value or a low value; the method is to calculate all of the vertical gradient image The number of possible vertical gradient values to draw the cumulative histogram of vertical gradient values. 'The horizontal axis of the cumulative histogram is that all vertical gradient values are arranged in order from small to large.' The vertical axis is for each vertical gradient value. The cumulative number of people who appear to be less than or equal to the vertical gradient value is to find a vertical gradient value between the lowest vertical gradient value and the highest vertical gradient value. 200529092 makes the number of occurrences of the lowest vertical gradient value less than or equal to this vertical gradient value. The difference between the linear polyline formed by the cumulative number of occurrences and the cumulative occurrence number less than or equal to the maximum vertical gradient value and the cumulative histogram of the vertical gradient value is the smallest (that is, the closest approximation to the cumulative histogram). Is the binarization threshold of the vertical gradient value, and the vertical gradient value greater than the critical value is set to a high value, and the straight gradient value less than or equal to the critical value is set to be low . (5) Merging the horizontal gradient binarized image and the vertical gradient binarized image 5: The horizontal gradient binarized image and the vertical gradient binarized image are combined into an integrated gradient binarized image. The feasible method is to combine the horizontal gradient binarized image Image and vertical gradient binarized image for “⑽” operation, that is, if one of the two pixels at the same coordinate position in the horizontal gradient binarized image and the vertical gradient binarized image has a high value, the gradient binarized image is integrated. Pixels at the same coordinate position in the binarized image also show high values; if both pixels at the same coordinate position in the horizontal gradient binarized image and vertical gradient binarized image show low values, the same The pixels at the coordinate position also show a low value. (6) Group neighboring high-value pixels into a license plate candidate area 6: Group neighboring high-value pixels in the integrated gradient binarized image into a license plate candidate area (Candidate regi〇n ) ·· The feasible method is to integrate the distance between the high-value pixels in the gradient binarized image if the distance is less than a specific value. The area where all the pixels of the label are located is the license plate candidate area. (7) The distribution of high-value pixels is used to determine whether the license plate candidate area is the license plate area. 7: Each candidate area of the | To determine whether or not it is a license plate area: Figure 3 shows an example of cutting each license plate candidate area into a number of districts 200529092 and analyzing the average and standard deviation of the high-value pixels contained in all small blocks. Schematic diagram, if the number of blocks in the license plate candidate area is too high, it can be removed instead of being determined as the license plate area, and an image of the determined license plate area will be produced. Another-embodiment of this creation is shown in Figure 4 _, Can only binarize the image for the horizontal gradient of the car sister image, mainly including the following steps: ⑴Calculate the horizontal gradient 丨: calculate the horizontal ladder of the car sister image plus gradient) to generate a horizontal gradient image; it calculates the horizontal gradient The method is to subtract the lightness and darkness or the darkness of the adjacent pixels into a scale number and then subtract the lightness and darkness. ⑵Horizontal gradient binarization The f image can be changed to a high value (素 _㈤) or a low value of the number of pixels per revolution of the image towel. The feasible method is to calculate the number of occurrences of all possible horizontal gradient values of the horizontal gradient image to the horizontal gradient value. The horizontal axis of the cumulative histogram is that all horizontal gradient values are arranged in order from small to large, and the vertical axis is the cumulative number of times that each horizontal gradient value is less than or equal to that level. Do the job at the most economic level and the highest horizontal gradient value-the horizontal gradient value, so that the lowest number of horizontal gradient values, the cumulative number of times less than or equal to this horizontal gradient value, the cumulative value of less than or equal to the maximum horizontal gradient value The gap between the polyline and the cumulative histogram of the horizontal gradient value is the smallest (that is, the closest approximation to the cumulative histogram). This horizontal gradient value is defined as the binary minus value of the horizontal gradient value. The value of the horizontal gradient value is set to a high value, and the horizontal gradient value less than or equal to the critical value is set to a low value. (3) Clustering adjacent high-value pixel groups into license plate candidate regions 6. Binarizing integrated gradients in the 200529092 image Clustering neighboring high-value pixel groups into license plate candidate regions; a feasible approach is to integrate gradient binary values If the distance between the high-value pixels in the image is less than a specific value, the high-value pixels are marked with the same label, and the area where all pixels of the same label are located is selected as the license plate candidate area. (4) Determine whether the license plate candidate area is a license plate area based on the distribution of high-value pixels. 7: Determine whether each license plate candidate area is a license plate area according to the distribution of high-value pixels in its area. It is a schematic diagram of the embodiment of cutting each license plate candidate area into several small blocks and analyzing the average number and standard deviation of the high-value pixels contained in all the small blocks. If it is too high, it can be removed without identifying it as a license plate area, and an image of the determined license plate area will be produced. In another embodiment of this creation, as shown in Figure 5, the image can be binarized only for the vertical gradient of the vehicle image, mainly including the following steps: (1) Calculate the vertical gradient of the vehicle image. Gradient (ver ^ cai gra (jient) to generate a vertical gradient image; its calculation of vertical fineness is to subtract the lightness and darkness of adjacent pixels or take the logarithm of the lightness and darkness to logarithmic lightness and darkness and then subtract. Vertical gradient binarization 4: Binary the vertical gradient image into a vertical gradient binarized image so that the value of each pixel in the image is only high or low; the method is to calculate all possible vertical gradients of the vertical gradient shirt image The number of occurrences of the value to draw a cumulative histogram of vertical gradient values. The horizontal axis of the cumulative histogram is that all vertical gradient values are arranged in ascending order. The vertical axis is less than or equal to each vertical gradient value. The cumulative number of people seeing the vertical gradient value finds a vertical gradient value between the lowest vertical gradient value and the highest vertical gradient value. 10 200529092 makes the number of occurrences of the lowest vertical gradient value less than or equal to this vertical gradient value The difference between the linear polyline formed by the cumulative number of occurrences and the cumulative occurrence number less than or equal to the maximum vertical gradient value and the cumulative histogram of the vertical gradient value is the smallest (that is, the closest approximation to the cumulative histogram). Is the binarization threshold of the vertical gradient value, and the vertical gradient value greater than the critical value is set to a high value, and the vertical gradient value less than or equal to the critical value is set to a low value. (3) Clustering of adjacent high-value pixel groups into License plate candidate area 6 •• Group adjacent high-value pixels in the integrated gradient binarized image into a license plate candidate area (Candidate). The method is to integrate the distance between the high-value pixels of the integrated gradient binarized image towel. If it is less than a specific value, the high-value pixels are marked with the same sincerity, and the area where all the pixels are slanted is determined as the license plate candidate area. ⑷ The tree structure of the high-value pixels is used to determine that the transition area is the material license plate area 7 : Each license plate candidate area is determined by the distribution of high-value pixels in its area. Whether it is a car domain or not ... Figure 3 is a slicing of each ship domain into several small blocks and analysis of financial districts. The embodiment of Lang's high-value image-area average recording and making a difference does not intend that the number of high-value blocks in the license plate candidate area is too high, which can be eliminated without identifying it as the vehicle domain, and the later $ Che Weiwei Image The method provided by the present invention for retrieving the license plate area from vehicle shadows has the following advantages when compared with other conventional techniques: 1. Obtaining the cumulative histogram incrementally upwards and downwards by the approximation method Or the most obvious slowdown is used as the threshold for binarization, and the binarized f-image binarization result can be obtained. 2. Analysis of the high-value pixel age distribution of the simple area of the license plate. The opportunity of the region is small. However, the above are only the preferred embodiments of the present invention. When the scope of the implementation of the present invention cannot be limited in this way, that is, what is simply done according to the scope of the patent application and the content of the invention specification Equivalent changes and materials should still fall within the scope of the invention patent. The technical structure of the present invention is not disclosed in detail above. The structure of the present invention can indeed achieve the purpose of obtaining a better image binarization effect, and it is not found in related products in the industry. It should meet the requirements of the invention patent. In accordance with the provisions of the Patent Law, Xiao Jun Bureau filed a patent for invention patents, please ask as soon as possible. [Brief description of the drawings] 凊 Refer to the following detailed description of the preferred embodiment of the present invention and its _, to further understand the technical content of the present invention and its purpose and effect; the drawings related to this embodiment are: -This is a flowchart of an embodiment of a method for capturing a license plate area from a vehicle image according to the present invention; and .. 2 is a horizontal gradient value (read from a cumulative histogram of horizontal gradient values (or vertical gradient values) of the present invention (read The implementation of the threshold of the binarization threshold) _ and the figure; and Fig. 3 is an example of the invention which cuts each license plate candidate area into several small blocks and analyzes the average health and standard of the high-value pixels included in all the small blocks. Poor example schematic. FIG. 4 is a flowchart of an embodiment of a method for retrieving a license plate area from a horizontal gradient of a car sister image. FIG. 5 is a flowchart of an embodiment of the method for retrieving a license plate area from a vertical gradient of a vehicle image. Figures; [Description of symbols in the diagram] 1-Calculate horizontal gradient 2-Horizontal gradient binarization 3-Calculate vertical gradient 4-Vertical gradient binarization 5-Combine horizontal gradient binarized image and vertical gradient binarized image 6— Group neighboring high-value pixels into a license plate candidate area 7—Use the distribution of high-value pixels to determine whether the license plate candidate area is a license plate area
1313