OpenCV实现保险丝颜色识别（附源码）

原创 Color Space OpenCV与AI深度学习

点击下方卡片，关注“OpenCV与AI深度学习”

视觉/图像重磅干货，第一时间送达!

导读

本文主要介绍Halcon转OpenCV实例--保险丝颜色识别(附源码)。

实例来源

实例来源于Halcon例程color_fuses.hdev--classify fuses by color

下面是Halcon实例代码和实现效果：

* color_fuses.hdev: classify fuses by colordev_update_window ('off')* ***** step: set up fuse properties and hue ranges* ****FuseColors := ['Orange','Red','Blue','Yellow','Green']FuseTypes := [5,10,15,20,30]* HueRanges: Orange 10-30, Red 0-10...HueRanges := [10,30,0,10,125,162,30,64,96,128]Count := 0dev_close_window ()dev_open_window (0, 0, 800, 600, 'black', WH)while (Count <= 4)    * ****    * step: acquire image    * ****    read_image (Image, 'color/color_fuses_0' + Count)    dev_display (Image)    set_tposition (WH, 12, 512)    write_string (WH, 'color/color_fuses0' + Count + '.png')    * ****    * step: extract saturated hues    * ****    decompose3 (Image, Red, Green, Blue)    trans_from_rgb (Red, Green, Blue, Hue, Saturation, Intensity, 'hsv')    threshold (Saturation, Saturated, 60, 255)    reduce_domain (Hue, Saturated, HueSaturated)    for Fuse := 0 to |FuseTypes| - 1 by 1        * ****        * step: classify specific fuse        * ****        threshold (HueSaturated, CurrentFuse, HueRanges[Fuse * 2], HueRanges[Fuse * 2 + 1])        connection (CurrentFuse, CurrentFuseConn)        fill_up (CurrentFuseConn, CurrentFuseFill)        select_shape (CurrentFuseFill, CurrentFuseSel, 'area', 'and', 6000, 20000)        area_center (CurrentFuseSel, FuseArea, Row1, Column1)        dev_set_color ('magenta')        for i := 0 to |FuseArea| - 1 by 1            set_tposition (WH, Row1[i], Column1[i])            write_string (WH, FuseColors[Fuse] + ' ' + FuseTypes[Fuse] + ' Ampere')        endfor        set_tposition (WH, 24 * (Fuse + 1), 12)        dev_set_color ('slate blue')        write_string (WH, FuseColors[Fuse] + ' Fuses: ' + |FuseArea|)    endfor    stop ()    Count := Count + 1endwhiledev_update_window ('on')

实现思路也比较简单，先将图像转到HSV颜色空间，然后分离S通道做阈值(60~255)，再分离H通道根据不同颜色的H范围来判定颜色。

OpenCV实现步骤与代码

测试图：

实现步骤：

【1】图像转到HSV颜色空间

【2】通道分离, 分离出H, S, V通道

【3】S通道做二值化(60~255)，然后通过轮廓查找提取每个保险丝的ROI

【4】对每个ROI做颜色判断：通过判断H通道特定范围内的像素数量

实现代码与测试效果：

#公众号：OpenCV与AI深度学习import numpy as npimport cv2
FuseColors = ['Orange','Red','Blue','Yellow','Green']
def check_color(ROI):  index = 0  #判断是否为红色  _,thresRL = cv2.threshold(ROI,0,255,cv2.THRESH_BINARY)  _,thresRH = cv2.threshold(ROI,10,255,cv2.THRESH_BINARY)  thresRed = thresRL - thresRH  numRed = cv2.countNonZero(thresRed)  #cv2.imshow('thresRed',thresRed)  #判断是否为橙色  _,thresOL = cv2.threshold(ROI,5,255,cv2.THRESH_BINARY)  _,thresOH = cv2.threshold(ROI,25,255,cv2.THRESH_BINARY)  thresOrange = thresOL - thresOH  numOrange = cv2.countNonZero(thresOrange)  #cv2.imshow('thresOrange',thresOrange)  #判断是否为蓝色  _,thresBL = cv2.threshold(ROI,90,255,cv2.THRESH_BINARY)  _,thresBH = cv2.threshold(ROI,110,255,cv2.THRESH_BINARY)  thresBlue = thresBL - thresBH  numBlue = cv2.countNonZero(thresBlue)  #cv2.imshow('thresBlue',thresBlue)  #判断是否为黄色  _,thresYL = cv2.threshold(ROI,25,255,cv2.THRESH_BINARY)  _,thresYH = cv2.threshold(ROI,65,255,cv2.THRESH_BINARY)  thresYellow = thresYL - thresYH  numYellow = cv2.countNonZero(thresYellow)  #cv2.imshow('thresYellow',thresYellow)  #判断是否为绿色  _,thresGL = cv2.threshold(ROI,65,255,cv2.THRESH_BINARY)  _,thresGH = cv2.threshold(ROI,90,255,cv2.THRESH_BINARY)  thresGreen = thresGL - thresGH  numGreen = cv2.countNonZero(thresGreen)  #cv2.imshow('thresGreen',thresGreen)
  max_val = max(numRed, numBlue,numYellow, numGreen,numOrange)  #print(max_val)   if max_val == numOrange:    index = 0  elif max_val == numRed:    index = 1  elif max_val == numBlue:    index = 2  elif max_val == numYellow:    index = 3  else:    index = 4
  return index    
img=cv2.imread("./color_fuses_01.png")cv2.imshow('src',img)rows,cols,channel = img.shapehsv_img=cv2.cvtColor(img,cv2.COLOR_BGR2HSV)hImg,sImg,vImg=cv2.split(hsv_img)
_,thres = cv2.threshold(sImg,60,255,cv2.THRESH_BINARY)cv2.imshow('thres',thres)cv2.imshow('hImg',hImg)cv2.imwrite('h.jpg',hImg)#cv2.waitKey()
contours,hierarchy = cv2.findContours(thres, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)for i in range(0,len(contours)):  rect = cv2.minAreaRect(contours[i])  box = cv2.boxPoints(rect)  box = np.int0(box)  width = rect[1][0]  height = rect[1][1]  if width < 100 or height < 100:    continue
  (x, y, w, h) = cv2.boundingRect(contours[i])  ROI = hImg[y:y+h,x:x+w]  index = check_color(ROI)  center = (int(rect[0][0]),int(rect[0][1]))  radius = (int)(max(width,height)/2+20)  if index == 0:    cv2.circle(img,center,radius,(0,128,255),3)    #img = cv2.drawContours(img,[box],0,(0,128,255),3)    cv2.putText(img,FuseColors[index],center,0,1.2,(255,255,0),2)      elif index == 1:    cv2.circle(img,center,radius,(0,0,255),3)    #img = cv2.drawContours(img,[box],0,(0,0,255),3)    cv2.putText(img,FuseColors[index],center,0,1.2,(0,255,0),2)    
  elif index == 2:    cv2.circle(img,center,radius,(255,255,0),3)    #img = cv2.drawContours(img,[box],0,(255,255,0),3)    cv2.putText(img,FuseColors[index],center,0,1.2,(255,0,255),2)  elif index == 3:    cv2.circle(img,center,radius,(0,255,255),3)    #img = cv2.drawContours(img,[box],0,(0,255,255),3)    cv2.putText(img,FuseColors[index],center,0,1.2,(0,255,128),2)  elif index == 4:    cv2.circle(img,center,radius,(0,255,0),3)    #img = cv2.drawContours(img,[box],0,(0,255,0),3)    cv2.putText(img,FuseColors[index],center,0,1.2,(0,0,255),2)  cv2.imshow('result',img)cv2.waitKey() cv2.destroyAllWindows()