Hough Transform

Definition

The Hough transform (HT) is a coordinate transformation introduced by Hough (Hough 1962). It is useful in computer vision as a method for retrieving shapes within digital images. It was first conceived for lines, circumferences, and simple polygons (Duda and Hart 1972) and later generalized to arbitrary shapes (Ballard 1981).

HT is best explained for the case of retrieval of lines. HT maps every meaningful point of an x–y image space into a line of an m–c parameters space. Meaningful features of an image for the purpose of HT are identified by retrieving points that have high gradient values, as these points could possibly belong to the contour of, e.g., a straight or curved line; computationally, they can be determined by edge detection operators. Figure 1 depicts the HT of points p and q of the image on the left into lines in the parameters space on the right. The various points of the HT line of a point p give the set of slope (m) and intercept (c) values of the bundle...