This paper introduces minimal solvers that jointly solve for affine-rectification and radial lens undistortion from the image of translated and reflected coplanar features. The proposed solvers use the invariant that the affine-rectified image of the meet of the joins of radially-distorted conjugately-translated point correspondences is on the line at infinity. The hidden-variable trick from algebraic geometry is used to reformulate and simplify the constraints so that the generated solvers are stable, small and fast. Multiple solvers are proposed to accommodate various local feature types and sampling strategies, and, remarkably, three of the proposed solvers can recover rectification and lens undistortion from only one radially-distorted conjugately-translated affine-covariant region correspondence. Synthetic and real-image experiments confirm that the proposed solvers demonstrate superior robustness to noise compared to the state of the art. Accurate rectifications on imagery taken with narrow to fisheye field-of-view lenses demonstrate the wide applicability of the proposed method. The method is fully automatic.