Speed up the computation of randomized algorithms for detecting lines, circles, and ellipses using novel tuning- and LUT-based voting platform (Q2383740)

Shape analysis is a fundamental problem in image processing. Lines, circles, and ellipses are the most important features since they often occur in images. Algorithms to detect lines, circles, and ellipses are divided into two categories, namely the deterministic and the nondeterministic ones. Most of the existing deterministic algorithms are based on the Hough transform. Based on the randomized approach, many nondeterministic algorithms have been developed to speed up the computing time. This paper focuses on a new computational scheme to speed up the computational requirements in several existing randomized algorithms for detecting lines, circles, and ellipses. The new computational scheme is based on Bresenham's drawing paradigm and a novel LookUp Table (LUT) based voting platform. Moreover, this paper provides a detailed time complexity analysis for each proposed improved algorithm. Under the same robustness, experimental results show that the execution-time improvement ratios of the authors' three proposed algorithms are about 28, respectively, and these improvement ratios are very close to theoretic analyses.