Apparatus which detects lines approximating an image by repeatedly

Abstract

Each of feature points (Xs, Ys) in X and Y orthogonal coordinate system of an image taken by a television camera is converted into a group of points (.rho., .theta.) of polar coordinate system. Memory includes addresses allotted to each one of polar coordinates (.rho., .theta.) for storing frequency data. Each time the feature point is converted, the frequency data at addresses corresponding to the converted points (.rho., .theta.) is incremented. A particular point (.rho.m, .theta.m) where the stored frequency data is at maximum is picked out, and a rectilinear line corresponding to the particular point (.rho.m, .theta.m) is determined to be a rectilinear line which approxiates a succession of feature points (Xs, Ys). In this manner, a method of detecting a rectilinear line appearing on a photographed image is provided. This processing operation is repeated in a plurality of runs for the same succession of feature points. However, during the conversion into the polar coordinates of an earlier run, the conversion takes place at a low density and covers an increased range of values of .rho. and .theta. while during a later run, the conversion takes place at a higher density and in a narrowed range which is centered about the particular point (.rho.m, .theta.m) which was obtained during the conversion into the polar coordinates of the previous run.