Hierarchical Visual Localization and Measurement Method for Outdoor Large-Scale Environment

The current large-scene environment recognition technology is mainly used in the field of automatic driving, providing environmental information for self-driving cars. It has relatively high localization and navigation capabilities and can make efficient driving decisions. This paper proposes a set of hierarchical visual measurement methods based on binocular stereo vision system. Deep learning target detection algorithm is deployed to provide a visual solution for automatic loading and unloading of container ports which are typical large-scale environments and require a perceptual measurement system to locate and measure containers. Experimental results are presented to verify the feasibility and performance of our proposed method.