identify_circles_and_get_average_radius

2023-12-24

algorithm

识别图片中的圆形并求所有圆形的平均半径

思路： 1）对图片进行预处理； 2）利用opencv，借助霍夫梯度法识别图像中的圆形； 3）在原图像中标记出识别到的圆形，并计算识别到的圆的平均半径。

实现过程： 待识别图像：

一、图像预处理

思路： 对图像进行epf边缘滤波，减小圆形识别的干扰；将图像转换成灰度图像。

具体实现： epf边缘滤波：

temp = cv.pyrMeanShiftFiltering(img, 2, 30)

灰度处理：

cimg = cv.cvtColor(temp, cv.COLOR_RGB2GRAY)

二、圆形识别

思路： 利用opencv的霍夫梯度法识别图像中的圆形。

具体实现：

circles = cv.HoughCircles(cimg,
                              cv.HOUGH_GRADIENT,
                              1,
                              20,
                              param1=50,
                              param2=30,
                              minRadius=20,
                              maxRadius=50)
circles = np.uint16(np.around(circles))
for i in circles[0, :]:
    cv.circle(img, (i[0], i[1]), i[2], (0, 255, 0), 2)  # 圆
    cv.circle(img, (i[0], i[1]), 2, (255, 255, 0), 2)  # 圆心

识别结果：

三、平均半径计算

思路： 根据比例尺计算出每个像素对应的、长度，再依次得到每个识别到的圆的半径，最后得到平均半径。

具体实现： 比例尺的计算：

    # ruler计算
    ruler_img = cv.imread(图像中比例尺的单独截图路径)
    ruler_size = np.shape(ruler_img)
    length = 0
    for i in range(len(ruler_size)):
        if length < ruler_size[i]:
            length = ruler_size[i]
    l_cons = 78
    delta_pixle = l_cons / length

平均半径的计算：

    # 平均半径计算
    cnt = 0
    sum_r = 0
    x = []
    y = []
    radius = []
    for circle in circles:
        cnt += 1
        sum_r += circle[2]
        x.append(circle[0])
        y.append(circle[1])
        radius.append(circle[2])
    ava_radius = sum_r / cnt
    ava_radius *= delta_pixle
    # 圆心坐标
    for i in range(len(x)):
        x[i] *= delta_pixle
    for i in range(len(y)):
        y[i] *= delta_pixle

完整代码： 为方便代码的修改，把各个单独的功能的代码放在了独立的文件中，文件结构如下图： 文件结构：

./main.py

import sys

sys.path.append(
    "/home/taot/Python/Useful_tools/circle_analyse_1/assistant_func")
import read_preprocess as rp
import circle_detection as cd
import cal_ava_radius as car
import cv2 as cv

pic = rp.pre_pro("/home/taot/Python/Useful_tools/circle_analyse_1/data.jpg")
img = pic[0]
cimg = pic[1]

ans = cd.cir_dec(img, cimg)
res = ans[0]
cv.imwrite("/home/taot/Python/Useful_tools/circle_analyse_1/res.png", res)

circles = ans[1][0]
ava_radius = car.ava_r(circles)
print("The avarage radius is: {}微米".format(ava_radius))

./read_preprocess.py

import cv2 as cv
import numpy as np


def pre_pro(pic_name):
    img = cv.imread(pic_name)
    cv.imshow("input img", img)
    cv.waitKey(0)
    cv.destroyAllWindows()

    # epf边缘滤波
    temp = cv.pyrMeanShiftFiltering(img, 2, 30)
    cv.imshow("epf", temp)
    cv.waitKey(0)
    cv.destroyAllWindows()
    # 灰度处理
    cimg = cv.cvtColor(temp, cv.COLOR_RGB2GRAY)
    cv.imshow("gray", cimg)
    cv.waitKey(0)
    cv.destroyAllWindows()

    return (img, cimg)

./circle_detection.py

import sys

sys.path.append(
    "/home/taot/Python/Useful_tools/circle_analyse_1/assistant_func")
import read_preprocess as rp
import cv2 as cv
import numpy as np


def cir_dec(img, cimg):
    circles = cv.HoughCircles(cimg,
                              cv.HOUGH_GRADIENT,
                              1,
                              20,
                              param1=50,
                              param2=30,
                              minRadius=20,
                              maxRadius=50)
    circles = np.uint16(np.around(circles))
    for i in circles[0, :]:
        cv.circle(img, (i[0], i[1]), i[2], (0, 255, 0), 2)  # 圆
        cv.circle(img, (i[0], i[1]), 2, (255, 255, 0), 2)  # 圆心
    cv.imshow("res", img)
    cv.waitKey(0)
    cv.destroyAllWindows()
    return (img, circles)

./cal_ava_radius.py

import numpy as np
import cv2 as cv


def ava_r(circles):
    # ruler计算
    ruler_img = cv.imread(
        "/home/taot/Python/Useful_tools/circle_analyse_1/ruler.png")
    ruler_size = np.shape(ruler_img)
    length = 0
    for i in range(len(ruler_size)):
        if length < ruler_size[i]:
            length = ruler_size[i]
    l_cons = 78
    delta_pixle = l_cons / length

    # 平均半径计算
    cnt = 0
    sum_r = 0
    x = []
    y = []
    radius = []
    for circle in circles:
        cnt += 1
        sum_r += circle[2]
        x.append(circle[0])
        y.append(circle[1])
        radius.append(circle[2])
    ava_radius = sum_r / cnt
    ava_radius *= delta_pixle
    for i in range(len(x)):
        x[i] *= delta_pixle
    for i in range(len(y)):
        y[i] *= delta_pixle
    np.savetxt("/home/taot/Python/Useful_tools/circle_analyse_1/x.txt", x)
    np.savetxt("/home/taot/Python/Useful_tools/circle_analyse_1/y.txt", y)
    np.savetxt("/home/taot/Python/Useful_tools/circle_analyse_1/radius.txt",
               radius)
    return ava_radius