Scene reconstruction method and device based on point cloud, equipment, storage medium and product
By using a uniform distribution sampling algorithm to divide and feature fusion of pyramid-shaped neighborhoods in point cloud reconstruction, the problem of low scene reconstruction accuracy in the existing technology is solved, significantly improving the fusion of features of object surface and edge area, and improving the weight Structural accuracy.
CN119180909APending Publication Date: 2024-12-24CHINA MOBILE ZIJIN INNOVATION INST CO LTD +3
View PDF 0 Cites 0 Cited by
Patent Information
- Application Number
- CN202411205757.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-29
- Publication Date
- 2024-12-24
AI Technical Summary
Technical Problem
In the prior art, the point cloud is downsampled layer by layer based on the farthest distance sampling algorithm, resulting in the key features of the object surface and edge areas being ignored, resulting in low scene reconstruction accuracy.
Method used
The point cloud data is divided into pyramid-shaped neighborhoods based on a uniform distribution sampling algorithm, and divided into more neighborhoods step by step, extracting and fusing target point features to build a more accurate scene model.
Benefits of technology
Through pyramid-shaped neighborhood division and feature fusion, the key features of the object surface and edge areas are more comprehensively integrated, improving the accuracy of scene reconstruction.
✦ Generated by Eureka AI based on patent content.
Smart Images
Figure CN119180909A_ABST
Abstract
The invention discloses a point cloud-based scene reconstruction method, device and equipment, a storage medium and a product, relates to the technical field of computers, and discloses a point cloud-based scene reconstruction method comprising the following steps: obtaining point cloud data; based on a preset uniform distribution sampling algorithm, performing pyramid-shaped neighborhood division on the point cloud data to obtain target point information of each hierarchy neighborhood; performing feature extraction on the target point information of the neighborhoods of each level to obtain target point features, and performing feature fusion on the target point features to obtain fused target point features; and constructing a target scene model corresponding to the point cloud data based on the fused target point features. According to the design of pyramid-shaped neighborhood division, rough division is carried out at a higher level, so that each neighborhood contains more points, then the points are refined downwards layer by layer, and finally fine division is realized at a lower level, so that key features of the surface and edge regions of an object are concerned, and the precision of scene reconstruction is improved.
Need to check novelty before this filing date? Find Prior Art