如何使用OpenCV在路边画一条线？

Question

I use this code to do line detection with canny and hough,but the effect is not good.

I want to draw a line on the curb. The purpose of drawing a straight line is to caculate the position of the line in the image, because when I do self-driving, I want my ROS car to drive along the curb.My country is right side driving.

import cv2
import numpy as np

img = cv2.imread(r"E:\test_opencv\images\testcanny.jpg")
blur_img = cv2.GaussianBlur(img, (3, 3), 0)
edges = cv2.Canny(blur_img, 250, 450, apertureSize=3)
lines = cv2.HoughLines(edges, 1, np.pi / 180, 118)
minLineLength = 800
maxLineGap = 15
threshold=80
lines = cv2.HoughLinesP(edges, 1, np.pi / 180, threshold, minLineLength, maxLineGap)
for i in range(len(lines)):
    for x1, y1, x2, y2 in lines[i]:
        cv2.line(img, (x1, y1), (x2, y2), (0, 255, 0), 4)
cv2.imshow('canny', edges)
cv2.imshow('Result', img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Original picture:

Canny picture:

Result picture:

Only some short green lines on the result picture.

In fact,I just want the red part of the curb to become a green line.

Just like this:

How to fix the code?

Answer 1

Using classic Hough transform ( HoughLines ) gives a better result other than the the probabilistic Hough transform ( HoughLinesP ). I ran this code on your exact image and got these results. The threshold value for the first image is 100 and for the next image is 200

The code:

import cv2
import numpy as np

img = cv2.imread("D:\\1.jpg")
blur_img = cv2.GaussianBlur(img, (3, 3), 0)
edges = cv2.Canny(blur_img, 250, 450, apertureSize=3)

rho = 1  #Distance resolution of the accumulator in pixels.
theta = np.pi/180 # Angle resolution of the accumulator in radians.
threshold  = 200 #Accumulator threshold parameter. Only those lines are returned that get enough votes ( >\texttt{threshold} ).

lines = cv2.HoughLines(edges, rho, theta, threshold)

for i in range(len(lines)):
    for r,th in lines[i]:
        a = np.cos(th)
        b = np.sin(th)
        x0 = a*r
        y0 = b*r
        x1 = int(x0 + 1000*(-b))
        y1 = int(y0 + 1000*(a))
        x2 = int(x0 - 1000*(-b))
        y2 = int(y0 - 1000*(a))

        cv2.line(img,(x1,y1),(x2,y2),(0,255,0),2)

cv2.imshow('canny', edges)
cv2.imshow('Result', img)
cv2.waitKey(0)
cv2.destroyAllWindows()

And further information on Hough transform:

https://opencv-python-tutroals.readthedocs.io/en/latest/py_tutorials/py_imgproc/py_houghlines/py_houghlines.html

Answer 2

For this picture, I use this code to make a better effect.

import cv2
import numpy as np

img = cv2.imread(r"E:\test_opencv\images\testcanny.jpg")
blur_img = cv2.GaussianBlur(img, (3, 3), 0)
gray = cv2.cvtColor(blur_img,cv2.COLOR_BGR2GRAY)
edges = cv2.Canny(gray, 250, 350, apertureSize=3)
rho = 1  #Distance resolution of the accumulator in pixels.
theta = np.pi/180 # Angle resolution of the accumulator in radians.
threshold  = 300 #Accumulator threshold parameter. Only those lines are returned that get enough votes ( >\texttt{threshold} ).
lines = cv2.HoughLines(edges, rho, theta, threshold)
if lines is not None:
    for i in range(len(lines)):
        for r,th in lines[i]:
            a = np.cos(th)
            b = np.sin(th)
            x0 = a*r
            y0 = b*r
            x1 = int(x0 + 1000*(-b))
            y1 = int(y0 + 1000*(a))
            x2 = int(x0 - 1000*(-b))
            y2 = int(y0 - 1000*(a))
            cv2.line(img,(x1,y1),(x2,y2),(0,255,0),2)
cv2.imshow('canny', edges)
cv2.imshow('Result', img)
cv2.waitKey(0)
cv2.destroyAllWindows()

Result picture: