Hough Transform In Geometry Detection

• Main Authors: Wu, Chujen, 吳巨仁
• Other Authors: Jeng, Jyhhorng
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/57534300515392133704

碩士 === 義守大學 === 資訊工程學系 === 100 === For image detection, shape recognition and analysis are important. In geometric images, straight lines and circles are detected most commonly, and edge information is important in shape recognition. There are many edge detection methods. Due to noise interference, information may produce broken or discontinuous features after the edge detection operations. These missing parts may be important information needed for studies. The missing parts can be compensated by Hough Transform. In image processing, Hough Transform is one of the methods to recognize geometric shapes. It’s an algorithm which would not be affected by image rotation and zooming, easy to transform images fast, and can link breakage or non-contiguous line segments together. Hough Transform finds the parameters from the split points in the image, and maps the split points from Euclidean space to parameter space by one-to-many. We find the image parameters with characteristics of the collinear by accumulating the number of all parameter positions using accumulator. The edge information would affect the accuracy at the final selection if the texture of the edge information is too complex. This work focuses on the strengthening and improvement of the straight lines. We increase the accuracy of the straight lines by regional sampling the results which were produced using Hough Transform in straight line detection.