Range Measurement by Computer Vision Systems Based on Invariant Moments

We review range measurement methods based on various configurations of computer vision sensors. We discuss the mathematical model of a stereoscopic pair and the conditionality of its fundamental matrix, which is essential for correct recognition and range measurement. A lot of researchers study the measurement based on single video or photo cameras, they are often installed on small mobile flying vehicles. We focus on the methods based on invariant moments. It is demonstrated that range measurement can be performed by a single camera. A complex of experiments was conducted in order to study single camera range measurement based on invariant moments. The deviation of experimental values of distance measurement that we have done didn't exceed 9%. The error can be explained by the inaccuracy in measuring the distance from the camera lens to the object, unstable position of the camera, and low quality of the obtained image due to the camera restrictions.