Dynamic environment information detection method based on semantic segmentation network and multi-view geometry

Abstract

The invention relates to a dynamic environment information detection method based on a semantic segmentation network and multi-view geometry, and the method comprises the following steps: calibratinga camera, and removing image distortion; acquiring and inputting an environment image; segmenting the input image through a semantic segmentation network to obtain masks of all objects and realize preliminary dynamic segmentation; ORB feature points being extracted from the input image, and then a descriptor being calculated; detecting and eliminating dynamic feature points by using a method of combining multi-view geometry and semantic information; matching the ORB feature points to obtain pose information of the robot; judging and inserting a key frame, and performing point cloud processingthrough a local mapping thread to obtain a sparse point cloud map; loop detection being used for optimizing the pose and correcting the drift error. According to the invention, the operation precisionand robustness of the RGBD SLAM system in a high-dynamic environment are improved. And meanwhile, by using the lightweight semantic segmentation network, the system storage and time expenditure can be reduced, and the real-time performance of the system is also considered while high precision is maintained.

Description

technical field [0001] The invention relates to the field of vision-based positioning and navigation in the autonomous inspection of UAVs, and more specifically, to a dynamic environment information detection method based on semantic segmentation network and multi-view geometry. Background technique [0002] In the process of UAV intelligent inspection, the UAV needs to determine the next operation independently according to the real-time information of the current environment. Therefore, real-time positioning and working environment mapping of UAVs are important links in the process of UAV intelligent inspection. Especially in the collaborative work of multi-UAVs arranged in a grid, the environment detected by each UAV is a dynamic scene (including moving objects that sometimes disappear), so it is necessary to Special algorithms need to be developed for dynamic scenes. [0003] Simultaneous Localization and Mapping (SLAM) is a method that can estimate the current positio...

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the problems of low detection accuracy and robustness in the above-mentioned prior art, the present invention provides a dynamic environment information detection method based on semantic segmentation network and multi-view geometry, improves detection accuracy and system robustness, and can also improve detection speed

Images