Ship target 6D pose estimation method based on point cloud data

A point cloud data, pose estimation technology, applied in the field of deep learning and neural network, can solve time-consuming problems, achieve the effect of enhancing perception accuracy and improving real-time performance

Active Publication Date: 2021-02-23
HARBIN ENG UNIV
View PDF19 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Traditional 3D pose estimation algorithms generally can only detect a single target and are time-consuming

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Ship target 6D pose estimation method based on point cloud data
  • Ship target 6D pose estimation method based on point cloud data
  • Ship target 6D pose estimation method based on point cloud data

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0050] 1. Make a data set containing the target category, the three-dimensional coordinates of the target, the three-dimensional size of the target, and the three-dimensional pose of the target, that is, the three angles of the ship target's corresponding yaw angle, roll angle, and pitch angle for training.

[0051] 2. Build a neural network.

[0052] The backbone network uses PointNet++ for point-by-point point cloud feature extraction. The feature extraction of point sets consists of three parts, namely Sampling layer, Grouping layer, and Pointnet layer. The sampling algorithm of the sampling layer uses the iterative farthest point sampling method iterative farthest point sampling (FPS). A series of points are selected in the input point cloud, thereby defining the center of the local region. Then build a local neighborhood, find points within a given distance, then use a fully connected layer for feature extraction, and finally perform a pooling operation to obtain advanc...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a ship target 6D pose estimation method based on point cloud data, and the method comprises the steps: 1, obtaining a ship point cloud data set of an offshore scene, and enabling a data set label to comprise a target type, the three-dimensional coordinates of a target, the three-dimensional size of the target, and the three-dimensional pose of the target; 2, constructing aneural network, and performing point-by-point point cloud feature extraction by adopting Point Net + + to obtain point-by-point high-dimensional features; 3, generating a 3D bounding box proposal frombottom to top, a real segmentation mask is generated based on the 3D bounding box, foreground points are segmented, and meanwhile the bounding box proposal with angle information is generated from the segmentation points and used for RCNN input; and 4, performing proposal optimization based on the proposal obtained in the step 3, the foreground segmentation features and the spatial features so asto output final classification, a 3D frame and an attitude angle. According to the method, the pose estimation effect of the three-dimensional target is achieved in an end-to-end learning mode, and the real-time performance of pose estimation is improved.

Description

technical field [0001] The invention relates to a method for estimating a 6D pose of a ship target based on point cloud data. The invention relates to the field of point clouds, the pose estimation of a ship target in a sea scene, deep learning and the fields of neural networks. Background technique [0002] Pose estimation plays a very important role in the field of computer vision. It has great applications in using visual sensors to estimate robot pose for control, robot navigation, augmented reality, and others. The pose estimation problem of the target is to determine the spatial position of the object in 3D space and the angle at which the object rotates around the coordinate axis, where the rotation around the Z axis is the yaw angle (Yaw), and the rotation around the Y axis is the pitch angle (Pitch). The rotation around the X axis is the roll angle (Roll). In recent years, 6D pose estimation methods can be divided into four categories, namely, methods based on mod...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/73G06K9/62G06N3/08
CPCG06T7/73G06N3/08G06F18/214
Inventor 苏丽宋浩
Owner HARBIN ENG UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap