bvbdort's profile - activity

Isn't rawSize should be 3 * DEFAULT_FRAME_Y * DEFAULT_FRAME_X; ?

What error did you get when you execute this code ? Are using Windows or Ubuntu ?

Can you post original RGB image, may be you could directly find the rectangle and can align easily.

yes this is from laser data , when i use the angle of segment for calculating distance between lines its giving wrong result.

thanks for suggestions, Since i needsegment start and endpoint for furthur calculations increasing maxLineGap is not an option. Also for fitline() i dont know in advance which segments are from sameline.

Hi,

I am using HoughLinesP to calculate lines from image. But the line segments from same line have different angles.

the segments start and end points and angle (in degrees) and length.

id = 0 start(x,y) = (1331 , 790 )   end(x,y) = (1366 , 794 ) Angle = 6.52311 len = 35.2278
id = 2 start(x,y) = (1508 , 807 )   end(x,y) = (1543 , 810 ) Angle = 4.90158 len = 35.1283
id = 4 start(x,y) = (1291 , 786 )   end(x,y) = (1328 , 789 ) Angle = 4.63781 len = 37.1214
id = 8 start(x,y) = (1391 , 795 )   end(x,y) = (1423 , 798 ) Angle = 5.35854 len = 32.1403
id = 9 start(x,y) = (1224 , 779 )   end(x,y) = (1261 , 782 ) Angle = 4.63781 len = 37.1214

To calculate angle i used atan2(y2-y1,x2-x1)*180/3.14

I am expecting same angle since the segments are part of same line.

Is there a way to improve the angle calculation or improving line detection ?

thanks.

matching part was missing, mDMatch you are passing empty.

i just put complete code lines, let me know it its working and gives you required result.

how did you check the device id of your camera ?

change

trunctaed.copyTo(target);   // erases all content  of target and keeps truncated image only.

to

trunctaed.copyTo(target(cv::Rect(0, 0, 30, 30))); // replace specified rect of target with truncate image.
Highgui.imwrite(PATH6, target); //img_3  result is in target.

above change gives.

Update :

    Mat temp = Highgui.imread(PATH5);//img_1
    Mat trunctaed = temp.submat(new Rect(0, 0, 30, 30));
    Mat target = Highgui.imread(PATH6);//img_2
    trunctaed.copyTo(target(cv::Rect(0, 0, 30, 30))); 
    Highgui.imwrite(PATH6, target); //img_3

@wuling add your link as answer.

try removing abs for sin and cos calculation.

thanks for the answer. Yes this result was better. But using 5° is not good for my case. Since i was using line angle for further calculation. Also i can see that the line 38 from a edge because of line. I am using it with 1° and you answer is helpful.

i tried with 1° and 5° but the result was same.

i tried it and didnt solve, here the issue i guess due to width of the edge

In c++ use

  cvtColor(img_in, img_out, CV_BGR2YUV);

In python

 img_out = cv2.cvtColor(img_in, cv2.COLOR_BGR2YUV)

Hi,

I am using HoughLinesP to extract lines from grayscale image.

Below is my psudo code.

threshold(img_tmp,bin_img,30,255,cv::THRESH_BINARY);
Canny(bin_img, canny_img, 50, 255, 3);
vector<Vec4i> tmp_lines;
HoughLinesP(canny_img,tmp_lines, 1, CV_PI/360,10, 10, 1);

As seen in below image single edge (i.e single line in image) is returning more than one line. 16,22 and 0,28 and 10,11 and 23,18,24 lines correspond to single line.

How can i get single line for single edge, any suggestions ?

Input image:

Extracted lines

Zoomed image

thanks.

You can do it by generating random color using rand() function in c++

    for ( ) // inside loop
    {
       // create random color variable
       Scalar clr(rand() % 255,rand() % 255,rand() % 255);  

       // draw line with random color            
       cv::Line(shape, cv::Point(z,a), cv::Point(o,n), clr,1,1); 
    }

Did you try Hausdorff distance ?

Hi all ,

I am using ROS Mapstitch package based on Opencv to find transformation bewtween maps ( grayscale images of same size and resolution) to combine them.

The implementation is based on opencvs estimateRigidTransform function, ORB feature extraction and matching the distance of pairwise feature candidates

For my input files the result from estimateRigidTransform is not correct.

Can somebody has any suggestion or know anyother existing methods to improve the result.

input image1

input image 2

keypoint matching

I can see Keypoints are matching to wrong location. In Ideal case upperwall keypoint should match to upperwall.

result

Thanks.

Add below header, it solves

#include "opencv2/video/tracking.hpp"