MATLAB Answers

Prespectrive transform for Lane detection project

16 ビュー (過去 30 日間)
Eduardo Lacerda
Eduardo Lacerda 2020 年 1 月 4 日
コメント済み: Eduardo Lacerda 2020 年 1 月 26 日
hello , any ideas on how to get the transformation matrix for this projection :
my goal is to obtain this :
question1.PNG
from this :
question.PNG
i have tried this :
im = imread('test1.jpg');
initial_points=[580,475; 740,475; 300,650; 1050,650]
final_points= [250,0 ; 1100,0; 0,720; 1200,720]
tform = fitgeotrans(final_points,initial_points,'projective');
out = imwarp(im,tform);
and allways get something like this :
question3.PNG
with open.cv in python i can do it with :
#4-points on the initial iamge
initial_points=np.float32([(580,475),(740,475),(300,650),(1050,650)])
#same 4 points in the porjected image
final_points=np.float32([(250,0),(1100,0),(250,720),(1100,720)])
Transformation_Matrix = cv2.getPerspectiveTransform(initial_points, final_points)
warped = cv2.warpPerspective(image, Transformation_Matrix,(1280,720)) #equivalent to imwarp
and i get this :
Test_gray.jpg

  2 件のコメント

Qu Cao
Qu Cao 2020 年 1 月 7 日
Can you attach 'test1.jpg' here?
Eduardo Lacerda
Eduardo Lacerda 2020 年 1 月 7 日
here you have

サインイン to comment.

回答 (2 件)

Eduardo Lacerda
Eduardo Lacerda 2020 年 1 月 7 日
here it is
thanks!

  0 件のコメント

サインイン to comment.


Prabhan Purwar
Prabhan Purwar 2020 年 1 月 23 日
Hi,
Following code may help
clc
close all
clear
oim = imread('test1.jpg');
im=oim(475:650,300:1050,:);
initial_points=[580,475; 740,475; 300,650; 1050,650];
final_points=[575,0; 720,0; 310,720; 1000,720];
initial_points(:,1) = initial_points(:,1)-475;
initial_points(:,2) = initial_points(:,2)-300;
tform = fitgeotrans(final_points,initial_points,'projective');
invtform = invert(tform);
out = imwarp(im,invtform);
subplot(1,2,1)
imshow(out);
subplot(1,2,2)
imshow(im);
Output:
pic.jpg
For more information refer to the following link:

  1 件のコメント

Eduardo Lacerda
Eduardo Lacerda 2020 年 1 月 26 日
thanks a lot for you help ,any ideas on how can i change your code to get a complete bird eye view of my lane like i present in the 1st images ? i also have to keep the original image size
thanks again

サインイン to comment.

サインイン してこの質問に回答します。


Translated by