Traditional visible light and infrared small field view positioning systems have some shortcomings, such as vulnerable to light conditions, poor environmental adaptability, and lack of visual information. In order to overcome these shortcomings, nonlinear epipolar constraint and target spatial positioning method based on ultra-wide field infrared binocular vision system was proposed. The generalized camera distortion model was used to describe the infrared ultra-wide camera imaging process, and the binocular system distortion model of non-parallel structure was establish. And then the theoretical derivation of the nonlinear epipolar constraint equation for homonymous image points was completed. A target location algorithm for ultra-wide field infrared binocular vision system was established from the point of view of space geometry. The experimental results show that the matching search of homonymous image points, included eight target points with the distance from 1.75~8.05 m in different scenes, is reduced from two-dimensional to one-dimensional under the conditions of epipolar constraint. The matching results were brought into the solution equation of three-dimensional spatial positioning and the distance error of positioning is between 0.9% and 7.0%. The higher positioning accuracy shows the correctness of the proposed method, and verifies the spatial positioning ability and advantage of the ultra-wide field infrared binocular vision system.