A fast extracting straight lines algorithm with parame tric control
BU Jia-jun,WANG Liang,CHEN Chun(College of Computer Science, Zhejiang University,Arcsoft Multimedia Research Center, Zhejiang University, Hangzhou 310027, China )
Presents an algorithm that extracts stra ight lines using global analysis and parametric control from intensity images. This algorithm solves the problem of efficiently extracting straight l ines in the domain of image understanding. First, pixels with similar gradient d irection were grouped into line support region, and then line representations we re obtained from these regions. Finally, the error range of every line was calcu lated. The a lgorithm appears to be more effective than previous techniques because: 1) gradient orientation (rather than gradient magnitude) was used as the initial organizing criterion prior to the extraction of straight lines, 2) the global context of the intensity variations associated with a straight line was d etermined prior to any local decisions on participating edge elements, 3) an approach was given to estimate the appropriateness of straight lines extracted with this algorithm, and 4) parametric control was used to reduce the time requ ired to process an image. The algorithm for extracting straight lines should have characteristics of high computing speed and high accuracy. The results sho w that the str aight lines extraction algorithm with global context analysis and parametric con trol is very effective and can be used in many applications .
【CateGory Index】： TP391.4