An Effective Method for Extracting Elliptical Targets from Images

High-accuracy automatic extraction of image target points is the foundation of photogrammetric 3D reconstruction. An adaptive-thresholding-based contour extraction method is proposed to effectively extract elliptical target points from images. A template-matching method is used to automatically determine the region of interest containing the target point in the image; within this region, the image edges are processed to extract the elliptical target contour, and least-squares ellipse fitting is used to determine the center coordinates of the elliptical target point, achieving sub-pixel accuracy. During ellipse fitting of the target point, gross-error pixels in the fit are removed to improve the accuracy of center-coordinate determination. Screening conditions are set according to the ellipse features to ensure the uniqueness of the detected ellipse within the region of interest.

Download paper here