igorgee / houghtransform

Naive implementation of the Hough Transform

Hough Transform

The Hough Transform is a feature extraction technique used in image analysis, computer vision, and digital image processing. The purpose of the technique is to find imperfect instances of objects within a certain class of shapes by a voting procedure. This voting procedure is carried out in a parameter space, from which object candidates are obtained as local maxima in a so-called accumulator space that is explicitly constructed by the algorithm for computing the Hough Transform.

H1

This program converts an image into an edge-map image. The method by which the transformation happens is with the Laplacian of Gaussian Filter.

H2

The next program is used to threshold out the weak edges as a result from the previous program.

H3

This is where the application of the Hough Transform occurs. This program will create a visual representation of the Hough Space, as well as a data format representing that Hough Space.

H4

Now we are ready to use out generated Hough space in the original image. This program will draw straight lines on the original image where it detects lines from the Hough space.

Program Executions

./h1 <input gray-level image> <output gray-level image>

./h2 <input gray-level image> <input gray-level threshold> <output binary image>

./h3 <input binary edge image> <output gray-level Hough image> <output Hough-voting-array>

• threshold used for sample images: 3

./h4 <input original gray-level image> <input Hough-voting-array> <input Hough threshold value> <output gray-level line image>

• threshold used for sample images: 90