CODEWITHSUNDEEP
image-processingopencv
AruniRC
AruniRC

Reputation: 5148

How to extract thin lines as separate contours/connected components with OpenCV?

I've used the distance tranform thinning on an image. Now trying to extract each connected component separtely - if there are two thin lines, then it should detect three such separate lines and components.

/*finding contours*/
IplImage *cc_color; 
cc_color = cvCreateImage(cvGetSize(thin_img), IPL_DEPTH_8U, 3);


    CvMemStorage *mem;
mem = cvCreateMemStorage(0);


int count = 0;
char* ch = new char [2];

CvSeq *contours = 0;

  CvSeq *ptr;
/*finding contours of morphed image*/
cvFindContours(thin_img, mem, &contours, sizeof(CvContour), CV_RETR_CCOMP,       CV_CHAIN_APPROX_SIMPLE, cvPoint(0,0));

    /*all contours on one image - random coloring*/
for (ptr = contours; ptr != NULL; ptr = ptr->h_next) 
    {
            CvScalar ext_color = CV_RGB( rand()&255, rand()&255, rand()&255 ); /*randomly coloring different contours*/
            cvDrawContours(cc_color, ptr, ext_color, CV_RGB(0,0,0), -1, CV_FILLED, 8, cvPoint(0,0));          
    }

thin_image is the input. the output should have each line colored randomly as a different contour/component. However it's only detecting closed shapes as contours. How do i detect lines as components??

Output image: Input image

Input image: enter image description here

The red boxes indicate example parts that should be detected as components. But only closed-shapes are detected.

Upvotes: 3

Views: 3173

Answers (1)

etarion
etarion

Reputation: 17141

If you are looking for the lines/borders as components instead of the inner regions, you should invert the thin_image (black -> white and white -> black) before applying findContours.

Upvotes: 3

Related Questions