How do I use houghLinesP?

[zachrose]

I'm trying to figure some basic things about houghLinesP, both by poking around in a debugger and trying to read/understand the header files and Python/C++ APIs.

I can tell it's a method of a Matrix object. When I call it with no arguments, I get this error:

OpenCV Error: Assertion failed (src.type() == CV_8UC1) in equalizeHist, file /tmp/opencv-OOOq/opencv-2.4.8.2/modules/imgproc/src/histogram.cpp, line 3128
libc++abi.dylib: terminating with uncaught exception of type cv::Exception: /tmp/opencv-OOOq/opencv-2.4.8.2/modules/imgproc/src/histogram.cpp:3128: error: (-215) src.type() == CV_8UC1 in function equalizeHist

Abort trap: 6

This error is Greek to me, unfortunately. Should I be using this method differently, or with arguments? Is there a way to read the OpenCV C++/Python API and know what the equivalent JavaScript method signature would look like? (I'm guessing it's more like the Python interface, except that it's a method of the image? Are there default parameters for rho, theta, and threshold?)

[salmanulhaq]

houghLinesP() expects the image to be grayscale (single channel) but what you're feeding it is a color image. This should have been handled within the code but is not (I'm updating it). What you need to do is call convertGrayscale() before calling houghLinesP(). Here's a quick example:

cv.readImage('/home/salman/image.jpg', function(err, img){

    img.convertGrayscale()
    var lines = img.houghLinesP()
})

Hope this helps.

[peterbraden]

The error messages are pretty bad, sorry about that.

[danschultzer]

@salmanulhaq has the correct solution. I'll close this for now.

[pijaginw]

Hi, I'm also trying to use houghLinesP() in my code (my goal is to detect lines from a stream of images) but nothing seems to work. Does the code which @salmanulhaq wrote should work? Or do I need to pass some additional arguments? Any help would be appreciated. Here's my simple code:

const cv = require('opencv');
cv.readImage('./car1.jpg', function (err, im) {
     if (err) {
        throw err;
    }

    const width = im.width();
    const height = im.height();
    if (width < 1 || height < 1) {
        throw new Error('Image has no size');
    }
    im.convertGrayscale();
    var lines = im.houghLinesP();

    // for (var i=0; i < lines.size(); i++) {
        // can I access the lines this way?
        // and then draw them using im.line()?
    // }
    im.save('./car1_res.jpg');
});

[ghost]

This took me a while to figure out so just in case someone else comes looking for how to draw the lines:

var outImage = new cv.Matrix(height, width);
for (var i = 0; i < lines.length; i++) {
    var line = lines[i];
    var pt1 = [line[0], line[1]];
    var pt2 = [line[2], line[3]];
    outImage.line(pt1, pt2, RED);
}  
outImage.save('tmp/whatever.jpg')