Research Of Line Characteristic Extraction Based On Beamlet

Line characteristic extraction is one of the basic theories of image procession, image description and pattern recognition. Beamlet is a multi-scale tool for line detecting, which can effectively extract line characteristic of image. In combination with the actual needs and the current hot research issues, this paper studies the correlative theory of line characteristic extraction and Beamlet system. The main work is as follows:(1) Combing the conventional edge detecting technology with imaginary part of Gabor function, a new edge detection algorithm based on Gabor filter group is proposed. The experiment results show that the algorithm can extract the edge of images effectively.(2) The conventional line characteristic extracting algorithm is studied. The paper analyzes the basic principles and the different characteristics of Radon transform and Hough transform through simulation.(3) The paper introduces the basic theory of Beamlet, and analyzes the basic theory of structure-less algorithm and BD-RDP algorithm in-depth, and then did the simulation.(4) To the question that the traditional fast algorithm which does not take full advantage of the similarity at different scale, this paper proposes a new fast structure algorithm based on global similarity. This algorithm is more suitable than the traditional fast algorithm for multi-scale analysis.(5) The problem of traditional Beamlet algorithms is that the position is uncertain and the short lines are difficult to be extracted. So there are false alarms and false dismissal in traditional Beamlet algorithms. After the analysis of the connection between Beam and Beamlet, the paper combines Beamlet and Beam to form a new recursive algorithm. The experiment results show that the algorithm can improve the phenomena of false alarms and false dismissal.