2017-03-12 92 views
0

我試圖區分長方形,梯形和半圓形。所以我所做的是在形狀周圍畫一個輪廓,然後畫一個旋轉的矩形。之後,我找到了輪廓和旋轉矩形的面積並取其比率。使用這個比率,我會確定形狀,因爲它對於前面提到的三種形狀是不同的。無法獲得適當的輪廓

(如果任何人有更魯棒的方法這三個區分這將受到讚賞。)

編輯的問題。我無法在圖像周圍畫出適當的輪廓。 下面是輸入和輸出圖像:

Input Image

Output Image

這裏是我的代碼提前

import cv2 
import numpy as np 

img = cv2.imread('h4.JPG') 
cv2.imshow('Input',img) 
#img = cv2.resize(img, None, fx=0.2,fy=0.2) 
img = cv2.GaussianBlur(img, (11,11), 0) 
img = cv2.fastNlMeansDenoisingColored(img,None,10,10,7,21) 
im = img.copy() 

imgray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY) 
ret,thresh = cv2.threshold(imgray,0,255,cv2.THRESH_BINARY) 
contours, hierarchy = cv2.findContours(thresh,cv2.RETR_TREE,cv2.CHAIN_APPROX_SIMPLE) 

max = 0 

for c in contours: 
    area = cv2.contourArea(c) 
    print area 
    if(np.any(max <= area)): 
     max = c 


A, B, C = cv2.minAreaRect(c) 
rotrect = cv2.minAreaRect(c) 
box = cv2.cv.BoxPoints(rotrect) 
box = np.int0(box) 
cv2.drawContours(im, contours, 0, (0,255,0), 2) 
cv2.drawContours(im, [box], 0, (0,0,255), 2) 

areaS = cv2.contourArea(contours[0]) 
areaR = B[0]*B[1] 

Ratio = areaS/areaR 

print "Shape Area: ",areaS 
print "Shape Rect: ",areaR 
print "Ratio: ",Ratio 

cv2.imshow('Output',im) 

if cv2.waitKey() and 0xff == 27: 
    cv2.destroyAllWindows() 

感謝。

+0

使用較高的閾值:'RET,脫粒= cv2.threshold(imgray,127255,cv2.THRESH_BINARY)' 。您_denoising_創建了非零像素,這些像素被'findContours'視爲_foreground_ – Miki

+0

Thanks @Miki。它有幫助。但可以請告訴我爲什麼你特別選擇了127作爲閾值 – StupidGuy

+0

某種程度上足夠高以去除低值(幾乎是黑色),但不能太高以去除高值(幾乎是白色)。 127在中間;)。從30到220的任何事情都可能工作得很好 – Miki

回答

0

我已經在註釋部分發布了由Miki提供的解決方案的代碼。

CODE:

im = cv2.imread('Figure.jpg', 1) 
gray_img = cv2.cvtColor(im,cv2.COLOR_BGR2GRAY) 
im1 = im.copy()      #---copy of the original image---- 

ret, thresh = cv2.threshold(gray_img, 127, 255, 0) 
blur_img = cv2.GaussianBlur(thresh, (11,11), 0) 

#---Finding and drawing contours--- 
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) 
cv2.drawContours(im1, contours, -1, (0, 255, 0), 3) 

#----Drawing a rotated rectangle---- 
cnt = contours 
rect = cv2.minAreaRect(cnt[0]) #---I used cnt[0] since there is only one contour, if there are more you can assign this within a for loop---- 
box = cv2.boxPoints(rect) 
box = np.int0(box) 
im = cv2.drawContours(im1, [box], 0, (0,0,255), 2) 

cv2.imshow("Final_Image.jpg", im1) 

結果:

enter image description here