Influencing Factors of Precision in an Active Vision System

摘要

Active vision technologies are important application based on machine vision. In the system, the encoded pattern with some styles and characters will be projected onto the surface of object. Its image will be modulated by object and captured by camera. Then, some image processing algorithms will be utilized to calculate the image coordinate of each interest point. Subsequently, the 3D Euclidean surface of object can be known. In the paper, pseudo-random principle is adopted to make the pattern of the projector in an active vision system. The pattern can make each interest point on the surface of object be recognized exclusively. It can also make following image processing simple and the active vision system convenient. The paper describes the influencing factors of precision in detailed, such as the pre-calibration of projector and the 3D Euclidean reconstruction algorithm. Accordingly, some improving measures have been presented.