Research on the Visual Recognition Algorithm for Loading Robot Vision System

Abstract

In order to study the questions about the recognition of relative position for the operating object and calculation of the object volume in the loading robot binocular stereo vision system. Based on the characteristics of the operating object, proposed the method that use of its vertices only to match the corresponding point of the camera imaging for location identification, also raised the approximation algorithm that firstly, cut up the whole, followed by calculate the volume of the various parts, then carry out the sum of the each segmentation volume. Experiments and analysis showed that the distance of camera and the object greater affect the visual system, less impact on the intensity of light；In the target object segmentation， the number of partition k=11 is better. Both methods produced the small errors for the visual recognition of the system, it can meet actual needs.