Line Extraction And Matching Based On Hough Transformation

The straight line is important texture feature of remote sensing image, line extraction and matching playing an important role in three-dimensional reconstruction field. Edge detection method was used to extract edge points and extract line information accurately through Hough transformation and applies to the line matching. Study of line matching algorithm in order to help complete three-dimensional reconstruction accurately. The following aspects were mainly studied:The basic theory of the edge detection was discussed, the advantages, disadvantages and their image type of several main methods and algorithm of edge detection were concluded and judged.The basic theory of Hough transformation was studied and the parameter space data that make full use of Hough transformation being forward, combines with zero clearing in the process of the Hough transformation peak value detecting, straight line and false line creation were erased by set threshold. Detecting the straight line ascertaining endpoint and the length.Against the deficiency of line extracts algorithmic, in order to improving noiseproof feature and abstracting speed, adaptive median filter, neighbourhood calculation were put forward and grey edge image was changed into bianry image. The gray feature of line hold-up area have fully utilized, the epipolar constraint was introduced to matching line. Resolve calculation area describe similar degree amounts problems due to the similar line which generate by light, field constraint and similarity scenery points.