Line Matching Based on Characteristic Ratio Invariants of Collinear Points

Abstract

This paper comes up with a method about line matching based on characteristic ratio invariants of collinear points, without resorting to any other constraints or priori knowledge. A novel concept called important point (IMP) is proposed for line matching: First, extracting lines using Hough transformation and obtaining the intersection points between the lines, Next, classifying the intersection points into IMPs and generating a two-dimensional characteristic ratio sequence of each IMP, Then, matching the IMPs using the two dimensional characteristic ratio sequence with NNDR, Finally, matching lines combining those correct matched IMPs with cross product symbol. Extensive experiments are conducted with this method. The results show that our proposed method is highly distinctive and effective for line matching under various kinds of transformations and outperform MSLD and B.Fan's method.