An Algorithm for Detecting and Measuring Jagged Elliptical Regions from a Binary Image; Region Identification
Christos Liambas, Constantine Tsouros · 2008
Here is a short explanation of an algorithm, which introduces a new approach for detecting and measuring elliptical regions from a binary image. These regions appear in the form of highly irregular white shapes in a black and white image. The algorithm computes three ellipses for every shape in the image. The inscribed ellipse Eiwhich is the largest ellipse contained in the shape, the circumscribed ellipse Ecwhich is the smallest ellipse that contains the shape and the approximation ellipse Eawhich is the ellipse with area equal to the area of the shape. The goal is to fit the Eaellipse to the elliptical region (region of interest-ROI) by processing the minimum possible number of pixels, with maximum efficiency. Of course, the algorithm is capable of dealing with the individual case of circular regions. Some computational results are presented on a set of benchmark images from actual data. Finally, a comparison with the NASA's Goddard IDL program library takes place in order to confirm the performance of the proposed algorithm.