Multi-view Stereo Point Cloud Generation Method and Application Camera Based on Particle Swarm Optimization

Abstract

The invention belongs to the technical field of image processing, and discloses a method for generating a multi-view stereo point cloud based on particle swarm optimization and a camera for application thereof, including: initializing information, extracting SURF features of each image, performing stereo matching on the extracted features, and performing feature matching The epipolar constraints are imposed on the matching pair, and the initial 3D space patch corresponding to the feature matching pair is generated, and the parameters of the initial 3D space patch are optimized using the particle swarm optimization algorithm; the optimized patch is expanded in the spatial neighborhood to form a dense point cloud, Perform filtering operations on the resulting dense point cloud; perform patch expansion and filtering operations multiple times until the stopping condition is met. The present invention uses SURF features and epipolar constraints to effectively ensure the accuracy of stereo matching, and the particle swarm optimization algorithm can avoid falling into a local optimum when optimizing patch parameters to obtain a global optimum solution, further improving the quality of point cloud reconstruction.

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a method for generating a multi-view stereo point cloud based on particle swarm optimization and a camera for its application. Background technique [0002] Multi-view 3D reconstruction is to use the camera to shoot a certain scene and objects at different angles, and use the captured multi-view image sequence to restore the 3D structural model of the scene and the objects in the scene. In the 3D reconstruction of the scene model, there are two classic methods to obtain the 3D model of the scene or object: one is to directly use the optical principle to scan the scene or object with a specific device to restore the 3D model of the scene or object, which is called the active method. Such as structured light projection method and laser scanning method; the other is to use image sequence information to restore the three-dimensional model of the scene or object f...

AI Technical Summary

Problems solved by technology

[0003] To sum up, the problems existing in the existing technology are: the current equipment-based 3D stereo reconstruction method has expensive hardware equipment, cumbersome manual operation, and limited application occasions; the feature mismatch of the image-based 3D stereo reconstruction method makes the final generated The accuracy of the point cloud decreases, and the gradient algorithm is easy to fall into the local optimum when optimizing the parameters, and the global optimal solution cannot be obtained

Images