检测这些圆，先找轮廓后通过轮廓点拟合椭圆
import cv
import numpy as np
import matplotlib.pyplot as plt
import math
from Ransac_Process import RANSAC
def lj_img(img):
wlj, hlj = img.shape[], img.shape[0]
lj_dis = # 连接白色区域的判定距离
for ilj in range(wlj):
for jlj in range(hlj):
if img[jlj, ilj] ==: # 判断上下左右是否存在白色区域并连通
for im in range(, lj_dis):
for jm in range(, lj_dis):
if ilj - im >= and jlj - jm >= 0 and img[jlj - jm, ilj - im] == 255:
cv.line(img, (jlj, ilj), (jlj - jm, ilj - im), (255, 255, 255), thickness=1)
if ilj + im < wlj and jlj + jm < hlj and img[jlj + jm, ilj + im] ==:
cv.line(img, (jlj, ilj), (jlj + jm, ilj + im), (255, 255, 255), thickness=1)
return img
def cul_area(x_mask, y_mask, r_circle, mask):
mask_label = mask.copy()
num_area =
for xm in range(x_mask+r_circle-, x_mask+r_circle+10):
for ym in range(y_mask+r_circle-, y_mask+r_circle+10):
# print(mask[ym, xm])
if (pow((xm-x_mask),) + pow((ym-y_mask), 2) - pow(r_circle, 2)) == 0 and mask[ym, xm][0] == 255:
num_area +=
mask_label[ym, xm] = (, 0, 255)
cv.imwrite('./test2/mask_label.png', mask_label)
print(num_area)
return num_area
def mainFigure(img, point):
# params = cv.SimpleBlobDetector_Params() # 黑色斑点面积大小：1524--1581--1400--周围干扰面积: 1325--1695--1688--
# # Filter by Area. 设置斑点检测的参数
# params.filterByArea = True # 根据大小进行筛选
# params.minArea =e2
# params.maxArea =e4
# params.minDistBetweenBlobs = # 设置两个斑点间的最小距离 10*7.5
# # params.filterByColor = True # 跟据颜色进行检测
# params.filterByConvexity = False # 根据凸性进行检测
# params.minThreshold = # 二值化的起末阈值,只有灰度值大于当前阈值的值才会被当成特征值
# params.maxThreshold = * 2.5 # 75
# params.filterByColor = True # 检测颜色限制，黑色，255白色
# params.blobColor =
# params.filterByCircularity = True
# params.minCircularity =.3
point_center = []
# cv.imwrite('./test2/img_source.png', img)
img_hsv = cv.cvtColor(img, cv2.COLOR_BGR2HSV)
# cv.imwrite('./test2/img_hsv.png', img_hsv)
w, h = img.shape[], img.shape[0]
w_hsv, h_hsv = img_hsv.shape[], img_hsv.shape[0]
for i_hsv in range(w_hsv):
for j_hsv in range(h_hsv):
if img_hsv[j_hsv, i_hsv][] < 200 and img_hsv[j_hsv, i_hsv][1] < 130 and img_hsv[j_hsv, i_hsv][2] > 120:
# if hsv[j_hsv, i_hsv][] < 100 and hsv[j_hsv, i_hsv][1] < 200 and hsv[j_hsv, i_hsv][2] > 80:
img_hsv[j_hsv, i_hsv] =, 255, 255
else:
img_hsv[j_hsv, i_hsv] =, 0, 0
# cv.imwrite('./test2/img_hsvhb.png', img_hsv)
# cv.imshow("hsv", img_hsv)
# cv.waitKey()
# 灰度化处理图像
grayImage = cv.cvtColor(img_hsv, cv2.COLOR_BGR2GRAY)
# mask = np.zeros((grayImage.shape[], grayImage.shape[1]), np.uint8)
# mask = cv.cvtColor(mask, cv2.COLOR_GRAY2BGR)
# cv.imwrite('./mask.png', mask)
# 尝试寻找轮廓
contours, hierarchy = cv.findContours(grayImage, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
# 合并轮廓
if len(contours) >:
# print(contours)
# 去掉离图中心最远的圆
max_idex, dis_max =, 0
for c_i in range(len(contours)):
c = contours[c_i]
cx, cy, cw, ch = cv.boundingRect(c)
dis = math.sqrt(pow((cx + cw / - w / 2), 2) + pow((cy + ch / 2 - h / 2), 2))
if dis > dis_max:
dis_max = dis
max_idex = c_i
contours.pop(max_idex)
# print(contours)
if len(contours) >:
contours_merge = np.vstack([contours[], contours[1]])
for i in range(, len(contours)):
contours_merge = np.vstack([contours_merge, contours[i]])
cv.drawContours(img, contours_merge, -1, (0, 255, 255), 1)
cv.imwrite('./test2/img_res.png', img)
# cv.imshow("contours_merge", img)
# cv.waitKey()
else:
contours_merge = contours[]
else:
contours_merge = contours[]
# RANSAC拟合
points_data = np.reshape(contours_merge, (-, 2)) # ellipse edge points set
print("points_data", len(points_data))
#.Ransac fit ellipse param
Ransac = RANSAC(data=points_data, threshold=.5, P=.99, S=.5, N=20)
# Ransac = RANSAC(data=points_data, threshold=.05, P=.99, S=.618, N=25)
(X, Y), (LAxis, SAxis), Angle = Ransac.execute_ransac()
# print( (X, Y), (LAxis, SAxis))
# 拟合圆
cv.ellipse(img, ((X, Y), (LAxis, SAxis), Angle), (0, 0, 255), 1, cv2.LINE_AA) # 画圆
cv.circle(img, (int(X), int(Y)), 3, (0, 0, 255), -1) # 画圆心
point_center.append(int(X))
point_center.append(int(Y))
rrt = cv.fitEllipse(contours_merge) # x, y）代表椭圆中心点的位置（a, b）代表长短轴长度，应注意a、b为长短轴的直径，而非半径,angle 代表了中心旋转的角度
# print("rrt", rrt)
cv.ellipse(img, rrt, (255, 0, 0), 1, cv2.LINE_AA) # 画圆
x, y = rrt[]
cv.circle(img, (int(x), int(y)), 3, (255, 0, 0), -1) # 画圆心
point_center.append(int(x))
point_center.append(int(y))
# print("no",(x,y))
cv.imshow("fit circle", img)
cv.waitKey()
# cv.imwrite("./test2/fitcircle.png", img)
# # 尝试寻找轮廓
# contours, hierarchy = cv.findContours(grayImage, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
# # print('初次检测数量: ', len(contours))
# if len(contours) ==:
# cv.drawContours(mask, contours[0], -1, (255, 255, 255), 1)
# cv.imwrite('./mask.png', mask)
# x, y, w, h = cv.boundingRect(contours[0])
# cv.circle(img, (int(x+w/2), int(y+h/2)), 1, (0, 0, 255), -1)
# cv.rectangle(img, (x, y), (x + w + 1, y + h + 1), (0, 255, 255), 1)
# point_center.append(x + w / + point0[0])
# point_center.append(y + h / + point0[1])
# cv.imwrite('./center1.png', img)
# else:
# # 去除小面积杂点， 连接轮廓，求最小包围框
# kernel = np.ones((3, 3), dtype=np.uint8)
# kernel = np.ones((2, 2), dtype=np.uint8)
# grayImage = cv.dilate(grayImage, kernel1, 1) # 1:迭代次数，也就是执行几次膨胀操作
# grayImage = cv.erode(grayImage, kernel2, 1)
# cv.imwrite('./img_dilate_erode.png', grayImage)
# contours, hierarchy = cv.findContours(grayImage, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
# if len(contours) ==:
# cv.drawContours(mask, contours[0], -1, (255, 255, 255), 1)
# cv.imwrite('./mask.png', mask)
# x, y, w, h = cv.boundingRect(contours[0])
# cv.circle(img, (int(x + w / 2), int(y + h / 2)), 1, (0, 0, 255), -1)
# cv.rectangle(img, (x, y), (x + w + 1, y + h + 1), (0, 255, 255), 1)
# point_center.append(x + w / + point0[0])
# point_center.append(y + h / + point0[1])
# cv.imwrite('./center1.png', img)
# else:
# gray_circles = cv.HoughCircles(grayImage, cv2.HOUGH_GRADIENT, 4, 10000, param1=100, param2=81, minRadius=10, maxRadius=19)
# # cv.imwrite('./img_gray_circles.jpg', gray_circles)
# if len(gray_circles[]) > 0:
# print('霍夫圆个数:', len(gray_circles[]))
# for (x, y, r) in gray_circles[]:
# x = int(x)
# y = int(y)
# cv.circle(grayImage, (x, y), int(r), (255, 255, 255), -1)
# cv.imwrite('./img_hf.jpg', grayImage)
#
# detector = cv.SimpleBlobDetector_create(params)
# keypoints = list(detector.detect(grayImage))
# for poi in keypoints: # 回归到原大图坐标系
# x_poi, y_poi = poi.pt[], poi.pt[1]
# cv.circle(img, (int(x_poi), int(y_poi)), 20, (255, 255, 255), -1)
# point_center.append(poi.pt[] + point0[0])
# point_center.append(poi.pt[] + point0[1])
# cv.imwrite('./img_blob.png', img)
# else:
# for num_cont in range(len(contours)):
# cont = cv.contourArea(contours[num_cont])
# # if cont >:
# # contours.append(contours[num_cont])
# if cont <=:
# x, y, w, h = cv.boundingRect(contours[num_cont])
# cv.rectangle(grayImage, (x, y), (x + w, y + h), (0, 0, 0), -1)
# cv.imwrite('./img_weilj.png', grayImage)
# grayImage = lj_img(grayImage)
# cv.imwrite('./img_lj.png', grayImage)
# contours, hierarchy = cv.findContours(grayImage, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_NONE)
# # print('再次检测数量: ', len(contours))
#
# cv.drawContours(mask, contours[0], -1, (255, 255, 255), 1)
# cv.imwrite('./mask.png', mask)
# x, y, w, h = cv.boundingRect(contours[0])
# cv.circle(img, (int(x + w / 2), int(y + h / 2)), 1, (0, 0, 255), -1)
# cv.rectangle(img, (x, y), (x + w + 1, y + h + 1), (0, 255, 255), 1)
# point_center.append(x + w / + point0[0])
# point_center.append(y + h / + point0[1])
# cv.imwrite('./center1.png', img)
return point_center[], point_center[1]
if __name__ == "__main__":
for i in range(,6):
imageName = "s"
imageName += str(i)
path = './Images/danHoles/' + imageName + '.png'
print(path)
img = cv.imread(path)
point = [0, 0]
cir_x, cir_y = mainFigure(img, point)
# img = cv.imread('./Images/danHoles/s2.png')
# point = [0, 0]
# cir_x, cir_y = mainFigure(img, point)
Ransac_Process.py
import cv
import math
import random
import numpy as np
from numpy.linalg import inv, svd, det
import time
class RANSAC:
def __init__(self, data, threshold, P, S, N):
self.point_data = data # 椭圆轮廓点集
self.length = len(self.point_data) # 椭圆轮廓点集长度
self.error_threshold = threshold # 模型评估误差容忍阀值
self.N = N # 随机采样数
self.S = S # 设定的内点比例
self.P = P # 采得N点去计算的正确模型概率
self.max_inliers = self.length * self.S # 设定最大内点阀值
self.items =
self.count = # 内点计数器
self.best_model = ((, 0), (1e-6, 1e-6), 0) # 椭圆模型存储器
def random_sampling(self, n):
# 这个部分有修改的空间，这样循环次数太多了，可以看看别人改进的ransac拟合椭圆的论文
"""随机取n个数据点"""
all_point = self.point_data
select_point = np.asarray(random.sample(list(all_point), n))
return select_point
def GeometricConic(self, ellipse):
# 这个部分参考了GitHub中的一位大佬的，但是时间太久，忘记哪个人的了
"""计算椭圆方程系数"""
# Ax ^ + Bxy + Cy ^ 2 + Dx + Ey + F
(x, y0), (bb, aa), phi_b_deg = ellipse
a, b = aa /, bb / 2 # Semimajor and semiminor axes
phi_b_rad = phi_b_deg * np.pi /.0 # Convert phi_b from deg to rad
ax, ay = -np.sin(phi_b_rad), np.cos(phi_b_rad) # Major axis unit vector
# Useful intermediates
a = a * a
b = b * b
# Conic parameters
if a > 0 and b2 > 0:
A = ax * ax / a + ay * ay / b2
B = * ax * ay / a2 - 2 * ax * ay / b2
C = ay * ay / a + ax * ax / b2
D = (- * ax * ay * y0 - 2 * ax * ax * x0) / a2 + (2 * ax * ay * y0 - 2 * ay * ay * x0) / b2
E = (- * ax * ay * x0 - 2 * ay * ay * y0) / a2 + (2 * ax * ay * x0 - 2 * ax * ax * y0) / b2
F = ( * ax * ay * x0 * y0 + ax * ax * x0 * x0 + ay * ay * y0 * y0) / a2 + \
(- * ax * ay * x0 * y0 + ay * ay * x0 * x0 + ax * ax * y0 * y0) / b2 - 1
else:
# Tiny dummy circle - response to a or b2 == 0 overflow warnings
A, B, C, D, E, F = (, 0, 1, 0, 0, -1e-6)
# Compose conic parameter array
conic = np.array((A, B, C, D, E, F))
return conic
def eval_model(self, ellipse):
# 这个地方也有很大修改空间，判断是否内点的条件在很多改进的ransac论文中有说明，可以多看点论文
"""评估椭圆模型，统计内点个数"""
# this an ellipse ?
a, b, c, d, e, f = self.GeometricConic(ellipse)
E = * a * c - b * b
if E <=:
# print('this is not an ellipse')
return, 0
# which long axis ?
(x, y), (LAxis, SAxis), Angle = ellipse
LAxis, SAxis = LAxis /, SAxis / 2
if SAxis > LAxis:
temp = SAxis
SAxis = LAxis
LAxis = temp
# calculate focus
Axis = math.sqrt(LAxis * LAxis - SAxis * SAxis)
f_x = x - Axis * math.cos(Angle * math.pi / 180)
f_y = y - Axis * math.sin(Angle * math.pi / 180)
f_x = x + Axis * math.cos(Angle * math.pi / 180)
f_y = y + Axis * math.sin(Angle * math.pi / 180)
# identify inliers points
f, f2 = np.array([f1_x, f1_y]), np.array([f2_x, f2_y])
f_distance = np.square(self.point_data - f1)
f_distance = np.square(self.point_data - f2)
all_distance = np.sqrt(f_distance[:, 0] + f1_distance[:, 1]) + np.sqrt(f2_distance[:, 0] + f2_distance[:, 1])
Z = np.abs( * LAxis - all_distance)
delta = math.sqrt(np.sum((Z - np.mean(Z)) **) / len(Z))
# Update inliers set
inliers = np.nonzero(Z <.8 * delta)[0]
inlier_pnts = self.point_data[inliers]
return len(inlier_pnts), inlier_pnts
def execute_ransac(self):
Time_start = time.time()
while math.ceil(self.items):
# print(self.max_inliers)
#.select N points at random
select_points = self.random_sampling(self.N)
#.fitting N ellipse points
ellipse = cv.fitEllipse(select_points)
#.assess model and calculate inliers points
inliers_count, inliers_set = self.eval_model(ellipse)
#.number of new inliers points more than number of old inliers points ?
if inliers_count > self.count:
ellipse_ = cv.fitEllipse(inliers_set) # fitting ellipse for inliers points
self.count = inliers_count # Update inliers set
self.best_model = ellipse_ # Update best ellipse
# print("self.count", self.count)
#.number of inliers points reach the expected value
if self.count > self.max_inliers:
print('the number of inliers: ', self.count)
break
# Update items
# print(math.log( - pow(inliers_count / self.length, self.N)))
self.items = math.log( - self.P) / math.log(1 - pow(inliers_count / self.length, self.N))
return self.best_model
if __name__ == '__main__':
#.find ellipse edge line
contours, hierarchy = cv.findContours(grayImage, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_NONE)
#.Ransac fit ellipse param
points_data = np.reshape(contours, (-, 2)) # ellipse edge points set
Ransac = RANSAC(data=points_data, threshold=.5, P=.99, S=.618, N=10)
(X, Y), (LAxis, SAxis), Angle = Ransac.execute_ransac()
检测对象
检测结果
蓝色是直接椭圆拟合的结果
红色是Ransc之后的结果
