车牌字符分割

import cv2 import numpy as np from PIL import Image from matplotlib import pyplot as plt def preprocess(gray): # # 直方图均衡化 #equ = cv2.equalizeHist(gray) # 高斯平滑 gaussian = cv2.GaussianBlur(gray, (3, 3), 0, 0, cv2.BORDER_DEFAULT) # 中值滤波 median = cv2.medianBlur(gaussian, 5) # Sobel算子，X方向求梯度 sobel = cv2.Sobel(median, cv2.CV_8U, 1, 0, ksize=3) # 二值化 ret, binary = cv2.threshold(sobel, 170, 255, cv2.THRESH_BINARY) #cv2.imshow("binary",binary) # 膨胀和腐蚀操作的核函数 element1 = cv2.getStructuringElement(cv2.MORPH_RECT, (9, 1)) element2 = cv2.getStructuringElement(cv2.MORPH_RECT, (8, 6)) # 膨胀2次，让轮廓突出 dilation = cv2.dilate(binary, element2, iterations=2) # 腐蚀一次，去掉细节 erosion = cv2.erode(dilation, element1, iterations=1) # 再3次膨胀，让轮廓明显一些 dilation2 = cv2.dilate(erosion, element2, iterations=3) cv2.imshow('dilation2', dilation2) cv2.waitKey(0) return dilation2 def findPlateNumberRegion(img): region = [] # 查找轮廓 binary,contours, hierarchy = cv2.findContours(img, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE) # 筛选面积小的 for i in range(len(contours)): cnt = contours[i] # 计算该轮廓的面积 area = cv2.contourArea(cnt) # 面积小的都筛选掉 if (area < 2000): continue # 轮廓近似，作用很小 epsilon = 0.001 * cv2.arcLength(cnt, True) approx = cv2.approxPolyDP(cnt, epsilon, True) # 找到最小的矩形，该矩形可能有方向 rect = cv2.minAreaRect(cnt) print("rect is: ") print(rect) # box是四个点的坐标 box = cv2.boxPoints(rect) box = np.int0(box) # 计算高和宽 height = abs(box[0][1] - box[2][1]) width = abs(box[0][0] - box[2][0]) # 车牌正常情况下长高比在2.7-5之间 ratio = float(width) / float(height) print(ratio) if (ratio > 5 or ratio < 2): continue region.append(box) return region def detect(img): # 转化成灰度图 gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 形态学变换的预处理 dilation = preprocess(gray) # 查找车牌区域 region = findPlateNumberRegion(dilation) # 用绿线画出这些找到的轮廓 for box in region: cv2.drawContours(img, [box], 0, (0, 255, 0), 2) ys = [box[0, 1], box[1, 1], box[2, 1], box[3, 1]] xs = [box[0, 0], box[1, 0], box[2, 0], box[3, 0]] ys_sorted_index = np.argsort(ys) xs_sorted_index = np.argsort(xs) x1 = box[xs_sorted_index[0], 0] x2 = box[xs_sorted_index[3], 0] y1 = box[ys_sorted_index[0], 1] y2 = box[ys_sorted_index[3], 1] img_org2 = img.copy() img_plate = img_org2[y1:y2, x1:x2] cv2.imshow('dingwei', img_plate) cv2.imwrite('dingwei.jpg', img_plate) cv2.namedWindow('img', cv2.WINDOW_NORMAL) cv2.imshow('img', img) # 带轮廓的图片 cv2.imwrite('contours.png', img) cv2.waitKey(0) cv2.destroyAllWindows() '''在imagePath中放入你要测试的图片''' imagePath = '1.jpg' im3 = Image.open(imagePath) im3.show(imagePath) img = cv2.imread(imagePath) detect(img) #进一步切割 img1 = Image.open("dingwei.jpg") #车牌大小统一成440*140 new_img = img1.resize((440,140),Image.BILINEAR) #车牌再次定位截取统一成([30,390],[20,390]) region = new_img.crop((30, 23, 416, 110)) region.save("dingwei1.jpg") #对车脾进行倾斜矫正 '''imga = cv2.imread("dingwei1.jpg", 0) imge = cv2.GaussianBlur(imga,(3,3),0) edges = cv2.Canny(imge, 50, 150, apertureSize = 3) lines = cv2.HoughLines(edges,1,np.pi/180,118) for l in lines[0]: theta = l[1] jiao = -(1.57079-theta)*180/3.14159 pil_im = Image.open('dingwei1.jpg') pil_im = pil_im.rotate(jiao) pil_im.save('300.jpg')''' im = Image.open('dingwei1.jpg') im.show('dingwei1.jpg') #灰度化 im = im.convert('L') im.save('001.png') #自己写的二值化 pixload = im.load() for x in range(im.size[0]): for y in range(im.size[1]): if pixload[x,y]<125: pixload[x,y] = 255 else: pixload[x,y] = 0 #去掉车牌的上下的圆孔 im.save('003.png') #im.show('003.png') pixload2 = im.load() print(im.size) for x in range(im.size[1]): count =0 for y in range(im.size[0]): if pixload2[y,x]==0: count=count+1 if count<50: for y in range(im.size[0]): pixload2[y, x] =255 im.save('004.png') #im.show('004.png') pixload3 = im.load() #去掉车牌上的圆点 for x in range(im.size[0]): count =0 for y in range(im.size[1]): if pixload3[x,y]==0: count=count+1 if count<0: for y in range(im.size[1]): pixload2[x, y] =255 im.save('005.png') #im.show('005.png') img = cv2.imread("005.png") # 读取图片 img_gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # 转换了灰度化 cv2.imshow('gray', img_gray) # 显示图片 cv2.waitKey(0) # 2、将灰度图像二值化，设定阈值是100 img_thre = img_gray cv2.threshold(img_gray, 140, 255, cv2.THRESH_BINARY_INV, img_thre) cv2.imshow('threshold', img_thre) cv2.waitKey(0) # 3、保存黑白图片 cv2.imwrite('132.png', img_thre) # 4、分割字符 white = [] # 记录每一列的白色像素总和 black = [] # ..........黑色....... height = img_thre.shape[0] width = img_thre.shape[1] white_max = 0 black_max = 0 # 计算每一列的黑白色像素总和 for i in range(width): s = 0 # 这一列白色总数 t = 0 # 这一列黑色总数 for j in range(height): if img_thre[j][i] == 255: s += 1 if img_thre[j][i] == 0: t += 1 white_max = max(white_max, s) black_max = max(black_max, t) white.append(s) black.append(t) #print(s) #print(t) arg = False # False表示白底黑字；True表示黑底白字 if black_max > white_max: arg = True # 分割图像 def find_end(start_): end_ = start_+1 for m in range(start_+1, width-1): if (black[m] if arg else white[m]) > (0.95 * black_max if arg else 0.95 * white_max): # 0.95这个参数请多调整，对应下面的0.05 end_ = m break return end_ n = 1 start = 1 end = 2 aaa =15 bb = 0 while n < width-2: n += 1 if (white[n] if arg else black[n]) > (0.05 * white_max if arg else 0.05 * black_max): # 上面这些判断用来辨别是白底黑字还是黑底白字 # 0.05这个参数请多调整，对应上面的0.95 start = n end = find_end(start) n = end if end-start > 9: aaa += 1 bb = bb+1 cj = img_thre[1:height, start:end] cv2.imwrite(str(aaa)+'.jpg',cj) # cv2.imshow('caijian', cj) cv2.waitKey(0) #显示分割后的图像 aaa = aaa-bb b = [] for i in range(bb): aaa += 1 img = cv2.imread(str(aaa)+'.jpg',0) b.append(img) for c in range(bb): print(c) c += 1 plt.subplot(1,bb,c) plt.imshow(b[c-1],'gray') plt.show()