Sports field registration is the process of determining the area of a sports field in view of a video or image. This process underpins many technologies and opportunities for sports broadcasting, including game state reconstruction and virtual reality applications, leading to copious research in recent decades. In this paper, we describe an approach to sports field registration designed for Rugby League, and applicable to similar niche sports. Sports field registration is a field in which data is very expensive to annotate leading to a lack of methods that apply to sports without a large commercial influence. Our approach utilises simple Hough line detection techniques and builds upon their shortcomings to improve reliability. We use vote-based line aggregation - a novel approach to line-based sports field registration. Vote-based line aggregation takes as input the detected lines across every frame of a video and the homographies between these frames to generate a single set of field lines from which to generate a homography for registration. We show our method is significantly and consistently more effective than a single-frame Hough line detection method for registration and demonstrate how our approach solves regularly occurring issues in the field such as player occlusion and false positives. We also propose the extension of our method to further improve state-of-the-art deep learning segmentation methods.