A Distance Measurement Method Based on Dual-parallel-line Projection for A Non-parameter Shooting System

Most images or videos are taken using mobile phones or other daily necessities. Unlike a professional device, those ready-to-use everyday items do not have a binocular or multi-ocular visual system. Thus, it is almost impossible to obtain the whole set of the internal parameters (the characteristics of a device, for example, a focal length) and the external parameters (the coordinates and rotational angles of a world coordinate system). For this reason, how to extract a distance is a difficult task and only few methods have been proposed to solve this problem. This paper presents a dual-parallel-line projection model to simply calculate the distance of a straight line for a target in a image. The model does not require any internal or external parameters and only needs very limited geometric information of an image. An image is processed using only a monocular vision system and an image perspective transformation. First, a monocular vision model is established to extract mathematical relationships between a photographer, target points, and target straight lines in a image. Then, a perspective transformation is carried out to change the angle of view on the plane to an angle of depression. An object with known length in the figure is calibrated to calculate the length of a target on the plane. Finally, auxiliary lines related to an object with known geometric information are added to the orignal image, and the results of the perspective transformation for those lines are claculated. The geometrical distance of a straight line between the photographer and the target is calculated using the mathematical relationship of the monocular vision model. Examples illustrate the validity of the method.