@article{heinrich_2013, title={Efficient and robust model fitting with unknown noise scale}, volume={31}, number={10}, journal={Image and Vision Computing}, author={Heinrich, S. B.}, year={2013}, pages={735–747} }
 @article{heinrich_snyder_frahm_2011, title={Maximum likelihood autocalibration}, volume={29}, ISSN={["1872-8138"]}, DOI={10.1016/j.imavis.2011.07.003}, abstractNote={This paper addresses the problem of autocalibration, which is a critical step in existing uncalibrated structure from motion algorithms that utilize an initialization to avoid the local minima in metric bundle adjustment. Currently, all known direct (not non-linear) solutions to the uncalibrated structure from motion problem solve for a projective reconstruction that is related to metric by some unknown homography, and hence a necessary step in obtaining a metric reconstruction is the subsequent estimation of the rectifying homography, known as autocalibration. Although autocalibration is a well-studied problem, previous approaches have relied upon heuristic objective functions, and have a reputation for instability. We propose a maximum likelihood objective and show that it can be implemented robustly and efficiently and often provides substantially greater accuracy, especially when there are fewer views or greater noise.}, number={10}, journal={IMAGE AND VISION COMPUTING}, author={Heinrich, Stuart B. and Snyder, Wesley E. and Frahm, Jan-Michael}, year={2011}, month={Sep}, pages={653–665} }
 @article{krish_heinrich_snyder_cakir_khorram_2010, title={Global registration of overlapping images using accumulative image features}, volume={31}, ISSN={0167-8655}, url={http://dx.doi.org/10.1016/j.patrec.2009.09.016}, DOI={10.1016/j.patrec.2009.09.016}, abstractNote={This paper introduces a new feature-based image registration technique which registers images by finding rotation- and scale-invariant features and matching them using a novel feature matching algorithm based on an evidence accumulation process reminiscent of the generalized Hough transform. Once feature correspondence has been established, the transformation parameters are then estimated using non-linear least squares (NLLS) and the standard RANSAC (random sample consensus) algorithm. The technique is evaluated under similarity transforms – translation, rotation and scale (zoom) and also under illumination changes.}, number={2}, journal={Pattern Recognition Letters}, publisher={Elsevier BV}, author={Krish, Karthik and Heinrich, Stuart and Snyder, Wesley E. and Cakir, Halil and Khorram, Siamak}, year={2010}, month={Jan}, pages={112–118} }