[Python image line coordinate extraction]

Problem Description:

When performing image processing, sometimes it is necessary to extract the coordinate points on the image, and then perform curve fitting, as shown in the figure below, but the extracted xy coordinates will have many repeated values, which will affect the curve fitting effect. Three methods are provided here, and you can supplement other methods by yourself.

warn: The image passed in by the function is the image after binarization, the pixels are only 0 and 255, how to binarize by Baidu.

1: Directly extract the coordinates of all black points. This method is obvious, and there will be many repeated x coordinates. Directly add the code:

def extract_line_position(image):       # 提取坐标，存在bug，会重复提取x的值
    list_y = []
    list_x = []    # 存储值为0的行号和列号
    
    for i in range(len(image)):
        for j in range(len(image[i])):
            if image[i][j] == 0:
                list_x.append(j)
                list_y.append(len(image) - i)
    return list_x, list_y

2 Directly extract the coordinates of the upper edge of the curve. This method will be distorted when the curve is thick or vertically downward, but it is suitable for the current scene. The code is as follows:

def get_line_position(image):
    list_x = []
    list_y = []
    # y_len = len(image)
    # print(y_len)
    for i in range(len(image[0])):    #  遍历列数
        for j in range(len(image)):   # 遍历行数
            if image[j][i] == 0:
                list_x.append(i)
                list_y.append(len(image)-j)
                break
    return list_x,list_y

3: Extract the median value of the upper and lower edges of the curve to obtain the position coordinates, code:

def get_lineMedium_position(image):
    image = np.delete(image, 0, axis=0)  # 删除第一行
    image = np.delete(image, 0, axis=1)  # 删除第一列
    list_x = []
    list_y = []
    start_index = 0
    end_index = 0
    for i in range(len(image[0])):  # 遍历列数
        for j in range(len(image)-1):  # 遍历行数
            if image[j][i] == 255 and image[j+1][i] == 0:
                start_index = j
                continue
            if image[j][i] == 0 and image[j+1][i] == 255:
                end_index = j
                y_position = (start_index + end_index)/2
                list_x.append(i)
                list_y.append(len(image) - y_position)
                start_index = 0
                end_index = 0
                break
    return list_x, list_y

The picture is not shown here, the effect is quite good, there is no need to refine the curve, saving time, but it should be used according to the actual situation.

Finally, a class is organized for everyone to use:

import numpy as np
import matplotlib.pyplot as plt


plt.rcParams['font.sans-serif'] = ['SimHei']  # 显示中文
plt.rcParams['axes.unicode_minus'] = False  # 正常显示负号

"""
December 16, 2022，LuoNicus
获取曲线分割后的图像点坐标类
用不同的方法获取坐标
只可提取像素 0 or 255，提取其它自行修改
正常提取，上边缘提取，中值提取，各有优缺，根据实际情况使用
"""

"""   图像像素示意
    [[255,255, 255, 0,   0]
     [0,   0,  255, 0,   0]
     [0,   0,  255, 0,   0]
     [0,   0,  255, 255,255]]     
"""

class Get_Line_Positon:
    def __init__(self, image):
        self.image = image
        self.list_x = []
        self.list_y = []
        print("已获取图像信息,准备提取二值图像位置坐标")

    def by_allline_point(self):  # 提取坐标，存在bug，会重复提取x的值
        # 细化算法存在bug，第一行和第一列为0黑色，不合理，因此首先去掉
        print("warn:常规算法提取，容易出现多个x为相同值")
        self.image = np.delete(self.image, 0, axis=0)  # 删除第一行
        self.image = np.delete(self.image, 0, axis=1)  # 删除第一列

        for i in range(len(self.image)):
            for j in range(len(self.image[i])):
                if self.image[i][j] == 0:
                    # print(mask[i][j],j,i)
                    self.list_x.append(j)
                    self.list_y.append(len(self.image) - i)
        return self.list_x, self.list_y

    def by_upline_point(self):  # 提取直线坐标，提取曲线的边缘坐标，避免出现x的重复值
        print("提取曲线上边缘的值")
        self.image = np.delete(self.image, 0, axis=0)  # 删除第一行
        self.image = np.delete(self.image, 0, axis=1)  # 删除第一列
        # y_len = len(self.image)
        # print(y_len)

        for i in range(len(self.image[0])):  # 遍历列数
            for j in range(len(self.image)):  # 遍历行数
                if self.image[j][i] == 0:
                    self.list_x.append(i)
                    self.list_y.append(len(self.image) - j)
                    break
        return self.list_x, self.list_y

    def by_lineMedium_point(self):  # 提取像素上下值的中位数坐标，该方法不用对函数进行细化
        print("提取曲线的中值")
        self.image = np.delete(self.image, 0, axis=0)  # 删除第一行
        self.image = np.delete(self.image, 0, axis=1)  # 删除第一列
        start_index = 0
        end_index = 0

        for i in range(len(self.image[0])):  # 遍历列数
            for j in range(len(self.image) - 1):  # 遍历行数
                if self.image[j][i] == 255 and self.image[j + 1][i] == 0:
                    start_index = j
                    continue
                if self.image[j][i] == 0 and self.image[j + 1][i] == 255:
                    end_index = j
                    y_position = (start_index + end_index) / 2
                    self.list_x.append(i)
                    self.list_y.append(len(self.image) - y_position)
                    start_index = 0
                    end_index = 0
                    break
        return self.list_x, self.list_y

Welcome to add! !