Research On Acceleration Of Ellipse Detection Based On Generalized Pascal's Mapping

Ellipse detection is a basic task in computer vision,which provides an effective support scheme for image analysis,and has a wide range of applications in many practical scenarios.For instance,ellipse detection can participate in the quality inspection of industrial devices,and efficiently identify traffic signs in intelligent transportation,assist in medical diagnosis in the field of medical imaging,or help shape analysis of different objects in the fields of biology and agriculture.Therefore,fast ellipse detection running on limited resources is crucial for various real-time computer vision systems.In the ellipse detection process,it is challenging for ellipse fitting a huge number of fragment candidates(edges or arcs),while pruning or selecting these candidate objects will also consumes a lot of computing resources.Therefore,in this study,we introduce a generalized Pascal mapping theorem(GPM)to recursively degenerate points on a planar curve of n th degree to 3(n-1) points on a curve of (n-1)th degree.That is to say,the introduced GPM enables time-consuming arc grouping on a line,a curve of the lowest degree,that preserves intrinsic geometric properties of ellipse.Therefore,based on the necessity of theorem,we designed an efficient GPM acceleration module for elliptical arc grouping,which finds valid elliptical arc candidate combination by determining whether the mapped three points lie on one line,invoking minimal computation.This article further embeds the GPM acceleration module into two representative advanced arc-based ellipse detection methods,which reduces their average running time of detection by 17% and 44% in five widely used data sets,respectively.This results in a new algorithm with faster detection speed,while ensuring that its detection accuracy is comparable or even higher.The factor for its success is that the calculation on the three collinear points is not only less complicated,but also the calculation accuracy is higher than the calculation accuracy of the points on the six ellipses,because the former is relatively less sensitive to point changes and position errors.