如果需要處理的原圖及程式碼,請移步小編的GitHub地址
傳送門:請點選我
如果點選有誤:https://github.com/LeBron-Jian/ComputerVisionPractice
1,簡單幾何影像繪製
簡單幾何影像一般包括點,直線,矩陣,圓,橢圓,多邊形等等。
下面學習一下 opencv對畫素點的定義。影像的一個畫素點有1或3個值,對灰度影像有一個灰度值,對彩色影像有3個值組成一個畫素值,他們表現出不同的顏色。
其實有了點才能組成各種多邊形,才能對多邊形進行輪廓檢測,所以下面先練習一下簡單的幾何影像繪製。
1.1 繪製直線
在OpenCV中,繪製直線使用的函式為 line() ,其函式原型如下:
def line(img, pt1, pt2, color, thickness=None, lineType=None, shift=None): # real signature unknown; restored from __doc__ """ line(img, pt1, pt2, color[, thickness[, lineType[, shift]]]) -> img . @brief Draws a line segment connecting two points. . . The function line draws the line segment between pt1 and pt2 points in the image. The line is . clipped by the image boundaries. For non-antialiased lines with integer coordinates, the 8-connected . or 4-connected Bresenham algorithm is used. Thick lines are drawn with rounding endings. Antialiased . lines are drawn using Gaussian filtering. . . @param img Image. . @param pt1 First point of the line segment. . @param pt2 Second point of the line segment. . @param color Line color. . @param thickness Line thickness. . @param lineType Type of the line. See #LineTypes. . @param shift Number of fractional bits in the point coordinates. """ pass
可以看到這個函式主要接受引數為兩個點的座標,線的顏色(其中灰色圖為一個數字,彩色圖為1*3的陣列)。
實踐程式碼如下:
import cv2 import numpy as np import matplotlib.pyplot as plt # 生成一個空灰度影像 img1 = np.zeros((400, 400), np.uint8) img1 = cv2.line(img1, (0, 0), (400, 400), 255, 5) # 生成一個空彩色影像 img3 = np.zeros((400, 400, 3), np.uint8) img3 = cv2.line(img3, (0, 0), (400, 400), (0, 255, 0), 5) titles = ['gray line image', 'color line image'] res = [img1, img3] for i in range(2): plt.subplot(1, 2, i+1) plt.imshow(res[i]), plt.title(titles[i]) plt.xticks([]), plt.yticks([]) plt.show()
效果如下:
注意1:在這裡再強調一下,由於cv和matplotlib的讀取影像通道不同,導致灰度圖和彩色圖的顏色不一樣,如果想分開看,可以直接使用cv2.imshow()。
注意2:繪製影像是在原圖上繪製,這裡我們寫的是專門在原圖上繪製,後面draw輪廓的話,可能需要 img.copy()了。不然我們的原圖會存在畫的輪廓。
1.2 繪製矩陣
在OpenCV中,繪製直線使用的函式為 rectangel() ,其函式原型如下:
def rectangle(img, pt1, pt2, color, thickness=None, lineType=None, shift=None): # real signature unknown; restored from __doc__ """ rectangle(img, pt1, pt2, color[, thickness[, lineType[, shift]]]) -> img . @brief Draws a simple, thick, or filled up-right rectangle. . . The function cv::rectangle draws a rectangle outline or a filled rectangle whose two opposite corners . are pt1 and pt2. . . @param img Image. . @param pt1 Vertex of the rectangle. . @param pt2 Vertex of the rectangle opposite to pt1 . . @param color Rectangle color or brightness (grayscale image). . @param thickness Thickness of lines that make up the rectangle. Negative values, like #FILLED, . mean that the function has to draw a filled rectangle. . @param lineType Type of the line. See #LineTypes . @param shift Number of fractional bits in the point coordinates. rectangle(img, rec, color[, thickness[, lineType[, shift]]]) -> img . @overload . . use `rec` parameter as alternative specification of the drawn rectangle: `r.tl() and . r.br()-Point(1,1)` are opposite corners """ pass
引數解釋
- 第一個引數img:img是原圖
- 第二個引數pt1:(x,y)是矩陣的左上點座標
- 第三個引數pt2:(x+w,y+h)是矩陣的右下點座標
- 第四個引數color:(0,255,0)是畫線對應的rgb顏色
- 第五個引數thickness:2是所畫的線的寬度
cv2.rectangle(img, (10, 10), (390, 390), (255, 0, 0), 3),需要確定的就是矩形的兩個點(左上角與右下角),顏色,線的型別(不設定就預設)。
程式碼如下:
import cv2 import numpy as np import matplotlib.pyplot as plt # 生成一個空灰度影像 img1 = np.zeros((400, 400), np.uint8) img1 = cv2.rectangle(img1, (40, 40), (350, 350), 255, 5) # 生成一個空彩色影像 img3 = np.zeros((400, 400, 3), np.uint8) img3 = cv2.rectangle(img3, (40, 40), (350, 350), (0, 255, 0), 5) titles = ['gray rectangle image', 'color rectangle image'] res = [img1, img3] for i in range(2): plt.subplot(1, 2, i+1) plt.imshow(res[i]), plt.title(titles[i]) plt.xticks([]), plt.yticks([]) plt.show()
效果如下:
1.3 繪製圓形
在OpenCV中,繪製直線使用的函式為 circle() ,其函式原型如下:
def circle(img, center, radius, color, thickness=None, lineType=None, shift=None): # real signature unknown; restored from __doc__ """ circle(img, center, radius, color[, thickness[, lineType[, shift]]]) -> img . @brief Draws a circle. . . The function cv::circle draws a simple or filled circle with a given center and radius. . @param img Image where the circle is drawn. . @param center Center of the circle. . @param radius Radius of the circle. . @param color Circle color. . @param thickness Thickness of the circle outline, if positive. Negative values, like #FILLED, . mean that a filled circle is to be drawn. . @param lineType Type of the circle boundary. See #LineTypes . @param shift Number of fractional bits in the coordinates of the center and in the radius value. """ pass
繪製圓形也簡單,只需要確定圓心與半徑即可。
實踐程式碼如下:
import cv2 import numpy as np import matplotlib.pyplot as plt # 生成一個空灰度影像 img1 = np.zeros((400, 400), np.uint8) img1 = cv2.circle(img1, (150, 150), 100, 255, 5) # 生成一個空彩色影像 img3 = np.zeros((400, 400, 3), np.uint8) img3 = cv2.circle(img3, (150, 150), 100, (0, 255, 0), 5) titles = ['gray circle image', 'color circle image'] res = [img1, img3] for i in range(2): plt.subplot(1, 2, i+1) plt.imshow(res[i]), plt.title(titles[i]) plt.xticks([]), plt.yticks([]) plt.show()
效果如下:
1.4 繪製橢圓
在OpenCV中,繪製直線使用的函式為 ellipse() ,其函式原型如下:
def ellipse(img, center, axes, angle, startAngle, endAngle, color, thickness=None, lineType=None, shift=None): # real signature unknown; restored from __doc__ """ ellipse(img, center, axes, angle, startAngle, endAngle, color[, thickness[, lineType[, shift]]]) -> img . @brief Draws a simple or thick elliptic arc or fills an ellipse sector. . . The function cv::ellipse with more parameters draws an ellipse outline, a filled ellipse, an elliptic . arc, or a filled ellipse sector. The drawing code uses general parametric form. . A piecewise-linear curve is used to approximate the elliptic arc . boundary. If you need more control of the ellipse rendering, you can retrieve the curve using . #ellipse2Poly and then render it with #polylines or fill it with #fillPoly. If you use the first . variant of the function and want to draw the whole ellipse, not an arc, pass `startAngle=0` and . `endAngle=360`. If `startAngle` is greater than `endAngle`, they are swapped. The figure below explains . the meaning of the parameters to draw the blue arc. . . ![Parameters of Elliptic Arc](pics/ellipse.svg) . . @param img Image. . @param center Center of the ellipse. . @param axes Half of the size of the ellipse main axes. . @param angle Ellipse rotation angle in degrees. . @param startAngle Starting angle of the elliptic arc in degrees. . @param endAngle Ending angle of the elliptic arc in degrees. . @param color Ellipse color. . @param thickness Thickness of the ellipse arc outline, if positive. Otherwise, this indicates that . a filled ellipse sector is to be drawn. . @param lineType Type of the ellipse boundary. See #LineTypes . @param shift Number of fractional bits in the coordinates of the center and values of axes. ellipse(img, box, color[, thickness[, lineType]]) -> img . @overload . @param img Image. . @param box Alternative ellipse representation via RotatedRect. This means that the function draws . an ellipse inscribed in the rotated rectangle. . @param color Ellipse color. . @param thickness Thickness of the ellipse arc outline, if positive. Otherwise, this indicates that . a filled ellipse sector is to be drawn. . @param lineType Type of the ellipse boundary. See #LineTypes """ pass
這裡解釋一下引數:
- img:影像
- center:橢圓圓心座標
- axes:軸的長度
- angle:偏轉的角度
- start_angle:圓弧起始角的角度
- end_angle:圓弧終結角的角度
- color:線條的顏色
- thickness:線條的粗細程度
- line_type:線條的型別,詳情見CVLINE的描述
- shift:圓心座標點的數軸的精度
影像化如下:
實踐程式碼如下:
import cv2 import numpy as np import matplotlib.pyplot as plt # 生成一個空灰度影像 img_origin1 = np.zeros((400, 400), np.uint8) img_origin11 = img_origin1.copy() # 引數依次是:影像,橢圓圓心座標,軸的長度,偏轉的角度, 圓弧起始角的角度,圓弧終結角的角度,線條的顏色,線條的粗細程度,線條的型別 img1 = cv2.ellipse(img_origin1, (150, 150), (150, 100), 30, 10, 190, 250) img11 = cv2.ellipse(img_origin11, (150, 150), (150, 100), 30, 10, 190, 250, -1) # 生成一個空彩色影像 img_origin3 = np.zeros((400, 400, 3), np.uint8) img_origin33 = img_origin3.copy() # 注意最後一個引數 -1,表示對影像進行填充,預設是不填充的,如果去掉,只有橢圓輪廓了 img3 = cv2.ellipse(img_origin3, (150, 150), (150, 100), 30, 0, 180, 250) img33 = cv2.ellipse(img_origin33, (150, 150), (150, 100), 30, 0, 180, 250, -1) titles = ['gray ellipse image', 'color ellipse image', 'gray ellipse padding', 'color ellipse padding'] res = [img1, img3, img11, img33] for i in range(4): plt.subplot(2, 2, i+1) plt.imshow(res[i]), plt.title(titles[i]) plt.xticks([]), plt.yticks([]) plt.show()
效果如下:
2,影像輪廓
影像輪廓可以簡單認為成將連續的點(連著邊界)連在一起的曲線,具有相同的顏色或者灰度。輪廓在形狀分析和物體的檢測和識別中很有用。
- 為了更加準確,要使用二值化影像。在尋找輪廓之前,要進行閾值化處理,或者Canny邊界檢測。
- 查詢輪廓的函式會修改原始影像。如果你在找到輪廓之後還想使用原始影像的話,你應該將原始影像儲存到其他變數中。
- 在OpenCV中,查詢輪廓就像在黑色背景中超白色物體。你應該記住要找的物體應該是白色而背景應該是黑色。
2.1 cv2.findContours()函式
那麼如何在一個二值化影像中查詢輪廓呢?這裡推薦使用函式cv2.findContours():
函式cv2.findContours()函式的原型為:
cv2.findContours(image, mode, method[, contours[, hierarchy[, offset ]]])
注意:opencv2返回兩個值:contours:hierarchy。而opencv3會返回三個值,分別是img(影像), countours(輪廓,是一個列表,裡面存貯著影像中所有的輪廓,每一個輪廓都是一個numpy陣列,包含物件邊界點(x, y)的座標), hierarchy(輪廓的層析結構)。
函式引數:
第一個引數是尋找輪廓的影像,即輸入影像;
第二個參數列示輪廓的檢索模式,有四種(本文介紹的都是新的cv2介面):
- cv2.RETR_EXTERNAL: 表示只檢測外輪廓
- cv2.RETR_LIST: 表示檢測所有輪廓,檢測的輪廓不建立等級關係,並將其儲存到一條連結串列當中
- cv2.RETR_CCOMP :表示檢測所有的輪廓,並將他們組織為兩層:頂層是各部分的外部邊界,第二次是空洞的邊界
- cv2.RETR_TREE: 表示檢測所有輪廓,並重構巢狀輪廓的整個層次,建立一個等級樹結構的輪廓
第三個引數method為輪廓的近似辦法
- cv2.CHAIN_APPROX_NONE:以Freeman鏈碼的方式輸出輪廓,所有其他方法輸出多邊形(頂點的序列)。儲存所有的輪廓點,相鄰的兩個點的畫素位置差不超過1,即max(abs(x1-x2),abs(y2-y1))==1
- cv2.CHAIN_APPROX_SIMPLE:壓縮水平方向,垂直方向,對角線方向的元素,只保留該方向的終點座標,例如一個矩形輪廓只需4個點來儲存輪廓資訊
- cv2.CHAIN_APPROX_TC89_L1,CV_CHAIN_APPROX_TC89_KCOS使用teh-Chinl chain 近似演算法
這裡輪廓的近似的兩個方法我們可以從下面圖看更加明顯:
一個是輸出所有輪廓(即所有頂點的序列),另一個函式只保留他們的終點部分。
函式返回值:
一般情況下,cv2.findContours()函式返回兩個值,一個是輪廓本身,還有一個是每條輪廓對應的屬性。當然特殊情況下返回三個值。即第一個是影像本身。
contour返回值
cv2.findContours()函式首先返回一個 list,list中每個元素都是影像中的一個輪廓,用numpy中的ndarray表示。這個概念非常重要,通過下面程式碼檢視:
print (type(contours)) print (type(contours[0])) print (len(contours)) ''' 結果如下: <class 'list'> <class 'numpy.ndarray'> 2 '''
這裡我們使用 contour.jpg 這幅影像舉個例子,圖如下:
通過上述圖,我們會看到本例中有兩條輪廓,一個是五角星的,一個是矩形的。每個輪廓是一個 ndarray,每個 ndarray是輪廓上的點的集合,並且列印出list的長度為2。
由於我們知道返回的輪廓有兩個,因此可以通過:
cv2.drawContours(img,contours[0],0,(0,0,255),3) cv2.drawContours(img,contours[1],0,(0,0,255),3)
分別繪製兩個輪廓,同時通過:
print(len(contours[0])) print(len(contours[1])) ''' 結果如下: 4 368 '''
輸出兩個輪廓中儲存的點的個數,可以看出,第一個輪廓中只有四個元素,這是因為輪廓中並不是儲存輪廓上所有的點,而是隻儲存可以用直線描述輪廓的點的個數,比如一個“正立”的矩形,只需要四個頂點就能描述輪廓了。而第二個輪廓卻有368個元素,因為它是不規整的影像。
hiarachy返回值
此外,該函式還可返回一個可選的hiararchy結果,這是一個ndarray,其中的元素個數和輪廓個數相同,每個輪廓contours[i]對應4個hierarchy元素hierarchy[i][0] ~hierarchy[i][3],分別表示後一個輪廓、前一個輪廓、父輪廓、內嵌輪廓的索引編號,如果沒有對應項,則該值為負數。
print (type(hierarchy)) print (hierarchy.ndim) print (hierarchy[0].ndim) print (hierarchy.shape) ''' 結果如下: <class 'numpy.ndarray'> 3 2 (1, 2, 4) '''
可以看出,hierachy本身包含兩個ndarray,每個 ndarray對應一個輪廓,每個輪廓有四個屬性。
完整程式碼如下:
import cv2 img = cv2.imread('contour.jpg') gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) ret, thresh = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) print (type(contours)) print (type(contours[0])) print (len(contours)) ''' 結果如下: <class 'list'> <class 'numpy.ndarray'> 2 ''' print(len(contours[0])) print(len(contours[1])) ''' 結果如下: 4 368 ''' print (type(hierarchy)) print (hierarchy.ndim) print (hierarchy[0].ndim) print (hierarchy.shape) ''' 結果如下: <class 'numpy.ndarray'> 3 2 (1, 2, 4) ''' # cv2.imshow('thresh', thresh) # cv2.waitKey(0) # cv2.destroyWindow('thresh')
2.2 cv2.drawContours()
OpenCV中通過 cv2.drawContours在影像上繪製輪廓。
下面看一下cv2.drawContours()函式:
cv2.drawContours(image, contours, contourIdx, color[, thickness[, lineType[, hierarchy[, maxLevel[, offset ]]]]])
引數:
- 第一個引數是指明在哪幅影像上繪製輪廓;
- 第二個引數是輪廓本身,在Python中是一個list。
- 第三個引數指定繪製輪廓list中的哪條輪廓,如果是-1,則繪製其中的所有輪廓。後面的引數很簡單。其中thickness表明輪廓線的寬度,如果是-1(cv2.FILLED),則為填充模式。繪製引數將在以後獨立詳細介紹。
下面看一個例項,在一幅影像上繪製所有的輪廓:
#_*_coding:utf-8_*_ import cv2 import numpy as np img_path = 'contour.jpg' img = cv2.imread(img_path) img1 = img.copy() img2 = img.copy() img3 = img.copy() imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(imgray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy= cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) # 繪製獨立輪廓,如第四個輪廓 img1 = cv2.drawContours(img1, contours, -1, (0, 255, 0), 3) # 如果指定繪製幾個輪廓(確保數量在輪廓總數裡面),就會只繪製指定數量的輪廓 img2 = cv2.drawContours(img2, contours, 1, (0, 255, 0), 3) img3 = cv2.drawContours(img3, contours, 0, (0, 255, 0), 3) res = np.hstack((img, img1, img2)) cv2.imshow('img', img3) cv2.waitKey(0) cv2.destroyAllWindows()
需要注意的是 cv2.findContours()函式接受的引數是二值圖,即黑白的(不是灰度圖),所以讀取的影像先要轉化成灰度圖,再轉化成二值圖,後面兩行程式碼分別是檢測輪廓,繪製輪廓。
比如原圖如下:
檢測到的所有輪廓圖如下(當指定繪製輪廓引數為 -1 ,預設繪製所有的輪廓):
當指定繪製輪廓的引數為 0的時候,則會找到索引為0的影像的輪廓如下:
同理,當指定繪製輪廓的引數為 1的時候,則會找到索引為1的影像的輪廓如下:
注意:findcontours函式會“原地”修改輸入的影像,所以我們需要copy影像,不然原圖會變。。。。
2.3 cv2.boundingrect()函式
矩形邊框(Bounding Rectangle)是說,用一個最小的矩形,把找到的形狀包起來。還有一個帶旋轉的矩形,面積會更小。
首先介紹下cv2.boundingRect(img)這個函式,原始碼如下:
def boundingRect(array): # real signature unknown; restored from __doc__ """ boundingRect(array) -> retval . @brief Calculates the up-right bounding rectangle of a point set or non-zero pixels of gray-scale image. . . The function calculates and returns the minimal up-right bounding rectangle for the specified point set or . non-zero pixels of gray-scale image. . . @param array Input gray-scale image or 2D point set, stored in std::vector or Mat. """ pass
解釋一下引數的意義:img是一個二值圖,也就是它的引數;返回四個值,分別是x,y,w,h( x,y是矩陣左上點的座標,w,h是矩陣的寬和高);
用下面函式解釋更加形象:
x, y, w, h = cv2.boudingrect(cnt) # 獲得外接矩形 引數說明:x,y, w, h 分別表示外接矩形的x軸和y軸的座標,以及矩形的寬和高, cnt表示輸入的輪廓值
得到矩陣的座標後,然後利用cv2.rectangle(img, (x,y), (x+w,y+h), (0,255,0), 2)畫出矩行,我們前面有講這個函式,這裡不再贅述。
下面舉個例子來看看如何找出不規則影像的外接矩陣,並畫出其矩陣,首先圖如下:
我們的目的是找出這個不規則影像的外接矩陣,並展示出來,程式碼如下:
#_*_coding:utf-8_*_ import cv2 import numpy as np img_path = 'contour2.png' img = cv2.imread(img_path) img1 = img.copy() img2 = img.copy() imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(imgray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy= cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE) print('輪廓的總數為', len(contours)) # 輪廓的總數為 2 cnt = contours[0] x, y, w, h = cv2.boundingRect(cnt) img1 = cv2.rectangle(img1, (x,y), (x+w,y+h), (0, 255, 0), 2) cv2.imshow('img', img1) cv2.waitKey(0) cv2.destroyAllWindows()
效果如下:
2.4 cv2.contourArea()
opencv中使用cv2.contourArea()來計算輪廓的面積。
首先介紹下cv2.contourArea(cnt, True)這個函式,原始碼如下:
def contourArea(contour, oriented=None): # real signature unknown; restored from __doc__ """ contourArea(contour[, oriented]) -> retval . @brief Calculates a contour area. . . The function computes a contour area. Similarly to moments , the area is computed using the Green . formula. Thus, the returned area and the number of non-zero pixels, if you draw the contour using . #drawContours or #fillPoly , can be different. Also, the function will most certainly give a wrong . results for contours with self-intersections. . . Example: . @code . vector<Point> contour; . contour.push_back(Point2f(0, 0)); . contour.push_back(Point2f(10, 0)); . contour.push_back(Point2f(10, 10)); . contour.push_back(Point2f(5, 4)); . . double area0 = contourArea(contour); . vector<Point> approx; . approxPolyDP(contour, approx, 5, true); . double area1 = contourArea(approx); . . cout << "area0 =" << area0 << endl << . "area1 =" << area1 << endl << . "approx poly vertices" << approx.size() << endl; . @endcode . @param contour Input vector of 2D points (contour vertices), stored in std::vector or Mat. . @param oriented Oriented area flag. If it is true, the function returns a signed area value, . depending on the contour orientation (clockwise or counter-clockwise). Using this feature you can . determine orientation of a contour by taking the sign of an area. By default, the parameter is . false, which means that the absolute value is returned. """ pass
引數含義如下:
- contour:表示某輸入單個輪廓,為array
- oriented:表示某個方向上輪廓的面積值,這裡指順時針或者逆時針。若為True,該函式返回一個帶符號的面積值,正負值取決於輪廓的方向(順時針還是逆時針),若為False,表示以絕對值返回
面積的值與輸入點的順序有關,因為求的是按照點的順序連線構成的圖形的面積。
下面實踐一下:
#_*_coding:utf-8_*_ import cv2 import numpy as np img_path = 'contour2.png' img = cv2.imread(img_path) imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(imgray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy= cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE) cnt = contours[0] # 求輪廓的面積 area = cv2.contourArea(cnt) print(img.shape) # (306, 453, 3) print(area) # 57436.5 # 也可以看輪廓面積與邊界矩形比 x, y, w, h = cv2.boundingRect(cnt) rect_area = w*h extent = float(area) / rect_area print('輪廓面積與邊界矩形比為', extent) # 輪廓面積與邊界矩形比為 0.7800798598378357
2.5 cv2.arcLength()
opencv中使用cv2.arcLength()來計算輪廓的周長。
首先介紹下cv2.arcLength(cnt, True)這個函式,原始碼如下:
def arcLength(curve, closed): # real signature unknown; restored from __doc__ """ arcLength(curve, closed) -> retval . @brief Calculates a contour perimeter or a curve length. . . The function computes a curve length or a closed contour perimeter. . . @param curve Input vector of 2D points, stored in std::vector or Mat. . @param closed Flag indicating whether the curve is closed or not. """ pass
引數含義如下:
- curve:輸入的二維點集(輪廓頂點),可以是 vector或者Mat型別
- closed:用於指示曲線是否封閉
下面舉個例子:
#_*_coding:utf-8_*_ import cv2 import numpy as np img_path = 'contour2.png' img = cv2.imread(img_path) imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(imgray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy= cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE) cnt = contours[0] # 求輪廓的周長 arcLength = cv2.arcLength(cnt, True) print(img.shape) # (306, 453, 3) print(arcLength) # 1265.9625457525253
2.6 cv2.approxPolyDP()
cv2.approxPolyDP()函式是輪廓近似函式,是opencv中對指定的點集進行多邊形逼近的函式,其逼近的精度可通過引數設定。我們首先看一張圖:
對於左邊這張圖,我們可以近似為中間和右邊的這張圖,具體如何近似呢?我們先不說,下面接著學。
下面看看cv2.approxPolyDP()函式的原始碼:
def approxPolyDP(curve, epsilon, closed, approxCurve=None): # real signature unknown; restored from __doc__ """ approxPolyDP(curve, epsilon, closed[, approxCurve]) -> approxCurve . @brief Approximates a polygonal curve(s) with the specified precision. . . The function cv::approxPolyDP approximates a curve or a polygon with another curve/polygon with less . vertices so that the distance between them is less or equal to the specified precision. It uses the . Douglas-Peucker algorithm <http://en.wikipedia.org/wiki/Ramer-Douglas-Peucker_algorithm> . . @param curve Input vector of a 2D point stored in std::vector or Mat . @param approxCurve Result of the approximation. The type should match the type of the input curve. . @param epsilon Parameter specifying the approximation accuracy. This is the maximum distance . between the original curve and its approximation. . @param closed If true, the approximated curve is closed (its first and last vertices are . connected). Otherwise, it is not closed. """ pass
其引數含義:
- curve:表示輸入的點集
- epslion:指定的精度,也即原始曲線與近似曲線之間的最大距離,不過這個值我們一般按照周長的大小進行比較
- close:若為True,則說明近似曲線為閉合的;反之,若為False,則斷開
該函式採用的是道格拉斯—普克演算法(Douglas-Peucker)來實現。該演算法也以Douglas-Peucker 演算法和迭代終點擬合演算法為名。是將曲線近似表示為一系列點,並減少點的數量的一種演算法。該演算法的原始型別分別由烏爾斯-拉默(Urs Ramer)於 1972年以及大衛-道格拉斯(David Douglas)和托馬斯普克(Thomas Peucker)於 1973年提出,並在之後的數十年中由其他學者完善。
經典的Douglas-Peucker 演算法描述如下:
- 1,在曲線首位兩點A, B之間連線一條直線AB,該直線為曲線的弦
- 2,得到曲線上離該直線段距離最大的點C,計算其與AB之間的距離d
- 3,比較該距離與預先給定的閾值 threshold 的大小,如果小於 threshold,則該直線段作為曲線的近似,該段曲線處理完畢
- 4,如果距離大於閾值,則用C將曲線分為兩段AC和BC,並分別對兩段取新進行1~3處理
- 5,當所有曲線都處理完畢後,依次連線各個分割點形成的折線,即可以作為曲線的近似
示意圖如下:
示例如下:
#_*_coding:utf-8_*_ import cv2 import numpy as np img_path = 'contour2.png' img = cv2.imread(img_path) img1 = img.copy() img2 = img.copy() imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(imgray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy= cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) cnt = contours[0] # 繪製獨立輪廓,如第四個輪廓 img1 = cv2.drawContours(img1, [cnt], -1, (0, 255, 0), 3) epsilon = 0.1*cv2.arcLength(cnt, True) approx = cv2.approxPolyDP(cnt, epsilon, True) img2 = cv2.drawContours(img2, [approx], -1, (0, 255, 0), 3) res = np.hstack((img, img1, img2)) cv2.imshow('img', res) cv2.waitKey(0) cv2.destroyAllWindows()
效果如下:
2.7 cv2.minEnclosingCircle()
在opencv中也可以實現輪廓的外接圓,它是函式cv2.minEnclosingCircle()。
下面我們看一下cv2.minEnclosingCircle()的原始碼:
def minEnclosingCircle(points): # real signature unknown; restored from __doc__ """ minEnclosingCircle(points) -> center, radius . @brief Finds a circle of the minimum area enclosing a 2D point set. . . The function finds the minimal enclosing circle of a 2D point set using an iterative algorithm. . . @param points Input vector of 2D points, stored in std::vector\<\> or Mat . @param center Output center of the circle. . @param radius Output radius of the circle. """ pass
引數意思也很明瞭,這裡不再贅述。
實踐程式碼如下:
#_*_coding:utf-8_*_ import cv2 import numpy as np img_path = 'contour2.png' img = cv2.imread(img_path) img1 = img.copy() imgray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(imgray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy= cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE) cnt = contours[0] # 求輪廓的外接圓 (x, y), radius = cv2.minEnclosingCircle(cnt) center = (int(x), int(y)) radius = int(radius) img1 = cv2.circle(img1, center, radius, (0, 255, 0), 2) res = np.hstack((img, img1)) cv2.imshow('img', res) cv2.waitKey(0) cv2.destroyAllWindows()
效果如下:
2.8 cv2.fillConvexPoly()與cv2.fillPoly()填充多邊形
opencv中沒有旋轉矩形,也沒有填充矩陣,但是它可以使用填充多邊形函式 fillPoly()來填充。上面兩個函式的區別就在於 fillConvexPoly() 畫了一個凸多邊形,這個函式要快得多,不過需要指定凸多邊形的座標。而fillPoly()則不僅可以填充凸多邊形,任何單調多邊形都可以填充。
cv2.fillConvexPoly()函式可以用來填充凸多邊形,只需要提供凸多邊形的頂點即可。
下面看看cv2.fillConvexPoly()函式的原始碼:
def fillConvexPoly(img, points, color, lineType=None, shift=None): # real signature unknown; restored from __doc__ """ fillConvexPoly(img, points, color[, lineType[, shift]]) -> img . @brief Fills a convex polygon. . . The function cv::fillConvexPoly draws a filled convex polygon. This function is much faster than the . function #fillPoly . It can fill not only convex polygons but any monotonic polygon without . self-intersections, that is, a polygon whose contour intersects every horizontal line (scan line) . twice at the most (though, its top-most and/or the bottom edge could be horizontal). . . @param img Image. . @param points Polygon vertices. . @param color Polygon color. . @param lineType Type of the polygon boundaries. See #LineTypes . @param shift Number of fractional bits in the vertex coordinates. """ pass
示例如下:
#_*_coding:utf-8_*_ import cv2 import numpy as np img = np.zeros((500, 500, 3), np.uint8) triangle = np.array([[50, 50], [50, 400], [400, 450]]) cv2.fillConvexPoly(img, triangle, (0, 255, 0)) cv2.imshow('image', img) cv2.waitKey(0) cv2.destroyAllWindows()
我們使用綠色填充,效果如下:
cv2.fillPoly()函式可以用來填充任意形狀的圖型.可以用來繪製多邊形,工作中也經常使用非常多個邊來近似的畫一條曲線.cv2.fillPoly()函式可以一次填充多個圖型。
下面看看cv2.fillPoly()函式的原始碼:
def fillPoly(img, pts, color, lineType=None, shift=None, offset=None): # real signature unknown; restored from __doc__ """ fillPoly(img, pts, color[, lineType[, shift[, offset]]]) -> img . @brief Fills the area bounded by one or more polygons. . . The function cv::fillPoly fills an area bounded by several polygonal contours. The function can fill . complex areas, for example, areas with holes, contours with self-intersections (some of their . parts), and so forth. . . @param img Image. . @param pts Array of polygons where each polygon is represented as an array of points. . @param color Polygon color. . @param lineType Type of the polygon boundaries. See #LineTypes . @param shift Number of fractional bits in the vertex coordinates. . @param offset Optional offset of all points of the contours. """ pass
效果如下:
#_*_coding:utf-8_*_ import cv2 import numpy as np img = np.zeros((500, 500, 3), np.uint8) area1 = np.array([[50, 50], [50, 400], [100, 450]]) area2 = np.array([[300, 300],[450, 300], [450, 450], [300, 450]]) cv2.fillPoly(img, [area1, area2], (255, 0, 0)) cv2.imshow('image', img) cv2.waitKey(0) cv2.destroyAllWindows()
效果如下:
3,輪廓處理實戰
下面舉一個實際的例子來鞏固一下學習的知識點。
問題是這樣的,假設我相對這張圖的左邊面積做處理,我希望將其填充為白色(任何想要的顏色)。
也就是黑色圈外的顏色填充為白色,希望能完全利用上面學到的函式。
下面依次分析,首先對影像進行K-Means聚類,效果如下:
然後檢測輪廓,這裡儘量將所有的輪廓檢測出來,如下:
然後對需要的輪廓進行填充,結果如下:
上圖為最終的效果,程式碼如下:
import cv2 import numpy as np import matplotlib.pyplot as plt def show_image(img): cv2.imshow('image', img) cv2.waitKey(0) cv2.destroyAllWindows() def image_processing(filename): img = cv2.imread(filename) img = cv2.resize(img, dsize=(100, 100)) data = img.reshape((-1, 3)) data = np.float32(data) # 定義中心(tyep, max_iter, epsilon) criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 10, 1.0) # 設定標籤 flags = cv2.KMEANS_RANDOM_CENTERS # K-means 聚類,聚整合2類 compactness, labels2, centers2 = cv2.kmeans(data, 2, None, criteria, 10, flags) # 2 類 影像轉換回 uint8 二維型別 centers2 = np.uint8(centers2) res2 = centers2[labels2.flatten()] dst2 = res2.reshape(img.shape) gray = cv2.cvtColor(dst2, cv2.COLOR_BGR2GRAY) _, thresh = cv2.threshold(gray, 127, 255, cv2.THRESH_BINARY) contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) # 第一個引數是指明在哪副影像上繪製輪廓,第二個引數是輪廓本身,在Python中是list # 第三個引數指定繪製輪廓list中那條輪廓,如果是-1,則繪製其中的所有輪廓。。 # dst3 = cv2.drawContours(img, contours, -1, (0, 255, 0), 3) # show_image(dst3) for ind, contour in enumerate(contours): print('總共有幾個輪廓:%s' % len(contours)) # 其中x,y,w,h分佈表示外接矩陣的x軸和y軸的座標,以及矩陣的寬和高,contour表示輸入的輪廓值 x, y, w, h = cv2.boundingRect(contour) print(x, y, w, h) if w > 80 or h > 80: print(contours[ind]) print(type(contours[ind]), contours[ind].shape) # cv2.fillConvexPoly()函式可以用來填充凸多邊形,只需要提供凸多邊形的頂點即可。 cv2.fillConvexPoly(img, contours[ind], (255, 255, 255)) show_image(img) # # 用來正常顯示中文標籤 # plt.rcParams['font.sans-serif'] = ['SimHei'] # # # 顯示圖形 # titles = [u'原圖', u'聚類影像 K=2'] # images = [img, dst2] # for i in range(len(images)): # plt.subplot(1, 2, i + 1), plt.imshow(images[i], 'gray') # plt.title(titles[i]) # plt.xticks([]), plt.yticks([]) # plt.show() if __name__ == '__main__': filename1 = 'test.png' image_processing(filename)
openCV Contours詳解:https://www.pianshen.com/article/5989350739/
參考文獻:https://blog.csdn.net/hjxu2016/article/details/77833336
https://blog.csdn.net/sunny2038/article/details/12889059#(寫的好)
https://www.cnblogs.com/Ph-one/p/12082692.html
部落格園函式:https://www.cnblogs.com/Undo-self-blog/p/8438808.html#top
https://blog.csdn.net/on2way/article/details/46793911