Partial Matching of Planar Polylines Under Similarity Transformations

Given two planar polylines T and P with n and m edges, respectively, we present an Om2 n2  time, Omn space algorithm to nd portions of the text" T which are similar in shape to the pattern" P. In the common case of a simple pattern, such as a line segment or corner, m = O1 and our algorithm requires On2  time and On space. We use the well-known arclength versus cumulative turning angle graph to judge how well a scaled, rotated, and translated version of the pattern matches a piece of the text. Our match scoring function balances the length of a match against the mean squared error in the match; given two matches with the same mean squared error length, the longer lower mean squared error match will have a higher score. The match score is a function of the pattern scale, orientation, and position within the text, and our algorithm seeks to nd local maxima of the scoring function. An analytic formula for the highest scoring pattern orientation in terms of scale a...

• Type: Conference
• Year: 1997
• Where: SODA
• Authors: Scott D. Cohen, Leonidas J. Guibas
• Keywords: Algorithms, Match Scoring Function, Mean Squared Error, Scoring Function, SODA 1997