Analysis for center deviation of circular target under perspective projection. Issue 7 (12th August 2019)

• Record Type: Journal Article
• Authors: Sun, Yu
• Is Part Of: Engineering computations. Volume 36:Issue 7(2019)
• Page Start: 2403
• Page End: 2413
• Publication Date: 2019-08-12
• Subjects: Feature extraction -- Centre deviation -- Circle -- Perspective projection -- Projective geometry; Computer-aided engineering -- Periodicals; Computer graphics -- Periodicals; 620.00285
• DOI: 10.1108/EC-09-2018-0431
• Languages: English
• ISSNs: 0264-4401

Abstract

Abstract : Purpose: Accurate feature localization is a fundamental problem in computer vision and visual measurement. In a perspective projection model of the camera, the projected center of a spatial circle and the center of the projection ellipse are not identical. This paper aims to show how to locate the real projection center precisely in the perspective projection of a space circle target. Design/methodology/approach: By analyzing the center deviation caused by projection transformation, a novel method is presented to precisely locate the real projection center of a space circle using projective geometry. Solution distribution of the center deviation is analyzed, and the quadratic equation for determining the deviation is derived by locating vanishing points. Finally, the actual projected center of the circular target is achieved by solving the deviation quadratic equations. Findings: The procedures of the author's method are simple and easy to implement. Experimental data calculated that maximum deviation occurs at approximately between 3π/10 and 2π/5 of the angle between the projection surface and the space target plane. The absolute reduction in error is about 0.03 pixels; hence, it is very significant for a high-accuracy solution of the position of the space circle target by minimizing systematic measurement error of the perspective projection. Originality/value: The center deviation caused by the space circle projection transformation is analyzed, and the detailed algorithm steps to locate the real projection center precisely are described.