Three-dimensional scene fusion method and device based on monocular estimation

Abstract

The embodiment of the invention provides a three-dimensional scene fusion method and device based on monocular estimation. The method comprises the steps: inputting an obtained first image into a target monocular depth estimation network, and obtaining a target depth map, the target monocular depth estimation network being obtained through the training of an initial monocular depth estimation network; obtaining depth information of a target object in the target depth map according to the target depth map and the target semantic segmentation map; according to the depth information of the targetobject and the parameter information of the equipment for collecting the first image, the position information of the target object in a preset static three-dimensional scene is obtained, and a mapping relation exists between a coordinate system of the static three-dimensional scene and a world coordinate system where the target object is located. The problem that in the prior art, due to the fact that a target depth estimation method is not ideal in implementation, the fusion mode of a monitoring object and a static three-dimensional scene model is not ideal is solved.

Description

technical field [0001] The present invention relates to the technical field of computer vision, in particular to a method and device for 3D scene fusion based on monocular estimation. Background technique [0002] Assuming that a static 3D scene model constructed from a real scene is known, we can monitor the corresponding moving objects in the real scene, such as people, cars, etc., through the camera in real time, and fuse these objects with the static 3D scene model to provide a A more intuitive and technological monitoring screen. Among them, the depth estimation of the target is a key issue. If the depth information from the target to the camera can be effectively determined, the position of the target in the 3D scene can be determined. Common depth estimation methods are: [0003] 1. Binocular distance measurement. The binocular distance measurement method needs to determine the target distance according to the baseline of the binocular camera. The range is limited b...

AI Technical Summary

Problems solved by technology

[0008] The embodiment of the present invention provides a 3D scene fusion method and device based on monocular estimation, to at least solve the problem in the related art that the method of fusion between the monitoring object and the static 3D scene model is not ideal due to the unsatisfactory implementation of the target depth estimation method

Images