Monocular optical flow-based potential obstacle detection method

Abstract

The invention belongs to the unmanned aerial vehicle technical field and relates to a monocular optical flow-based potential obstacle detection method. According to the method, an unmanned aerial vehicle is adopted to carry a sensor; an optical flow detection environment is simplified through a global adaptive iterative threshold and a frame difference method; the accuracy and real-time performance of target detection and tracking are improved through a pyramid model optical flow algorithm; a relative distance between the unmanned aerial vehicle and a detection angular point is estimated through optical flow FOE (Focus Of Expansion) and optical flow TTC (Time To Contact) calculation formulas. Compared with an existing visual ranging technology, the method of the invention can assist in reducing the number of cameras and saving cost. Compared with a traditional optical flow depth information detection algorithm, the method of the invention can assist in improving ranging accuracy and the environmental perception capacity of the unmanned aerial vehicle in an autonomous line inspection process, and can be widely applied to electric unmanned aerial vehicle line inspection.

Description

technical field [0001] The invention belongs to the technical field of unmanned aerial vehicles, and specifically relates to a potential obstacle detection method based on monocular visual flow. Background technique [0002] In the process of detecting the forward depth information of the monocular eye flow, the detection environment of the traditional optical flow algorithm needs to meet the three prerequisites of constant brightness, consistent space and continuous time of the detection environment, while the random environment in the process of UAV power inspection Therefore, the traditional optical flow algorithm often cannot meet the requirements of real-time detection of UAVs in the autonomous line inspection environment. [0003] In addition, the optical flow target detection and tracking algorithm is mainly to detect and track the full-frame cruising environment of the UAV. Complex environment detection will be accompanied by a huge amount of calculation and lengthy ...

AI Technical Summary

Problems solved by technology

[0002] In the process of detecting the forward depth information of the monocular eye flow, the detection environment of the traditional optical flow algorithm needs to meet the three prerequisites of constant brightness, consistent space and continuous time of the detection environment, while the random environment in the process of UAV power inspection Therefore, the traditional optical flow algorithm often cannot meet the requirements of real-time detection of drones in the autonomous line inspection environment.

[0003] In addition, the optical flow target detection and tracking algorithm is mainly to detect and track the full-frame cruising environment of the UAV. Complex environment detection will be accompanied by a huge amount of calculation and lengthy calculation time, which will reduce the detection accuracy of the algorithm in the environment, and is not suitable for Exploration of obstacles in the actual field environment

Images