Unlock instant, AI-driven research and patent intelligence for your innovation.

A Template-Based 3D Point Cloud Target Detection and Pose Estimation Method

A three-dimensional point cloud and attitude estimation technology, applied in the field of industrial parts detection, can solve the problems of reducing the overall system accuracy, increasing the difficulty of ICP matching, and unable to generate a good initial attitude value, and achieving the effect of high stability and mobility

Active Publication Date: 2020-10-16
视研智能科技(广州)有限公司
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The traditional neural network-based target detection algorithm cannot generate a good initial attitude value, which increases the difficulty of ICP matching and reduces the overall system accuracy

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
  • A Template-Based 3D Point Cloud Target Detection and Pose Estimation Method
  • A Template-Based 3D Point Cloud Target Detection and Pose Estimation Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0040] This embodiment provides a template-based 3D point cloud object detection and pose estimation method, such as figure 1 As shown, the method includes the following steps:

[0041] S1: Construct and train the neural network model; such as figure 2 Shown, the training of described neural network model comprises the following steps:

[0042] S1.1: Mark the 3D coordinates and pose of the target object on the 3D point cloud data as training samples.

[0043] S1.2: Take the plane perpendicular to the height direction of the point cloud as the projection plane, project the point cloud obtained by scanning the target vertically onto the projection plane, and grid it to obtain a depth map.

[0044] S1.3: Use the plane perpendicular to the height direction of the point cloud as the projection plane, rotate the 3D point cloud template to different poses and project to the projection plane, and obtain a set of depth maps of the point cloud templates in different poses.

[0045] ...

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 relates to a template-based three-dimensional point cloud target detection and attitude estimation method, the method comprising the following steps: S1: constructing and training a neural network model; S2: inputting the three-dimensional point cloud for target detection and attitude estimation into the S1 performs target detection and pose estimation in the trained neural network model; S3: output 3D point cloud target detection and pose estimation results. The invention solves the difficult problem of synchronous target detection and attitude estimation, and solves the problem of poor transferability of deep neural network. The trained model has extremely high stability and transferability, and can be applied to various template-based target detection and pose estimation tasks other than training samples.

Description

technical field [0001] The invention relates to the field of industrial parts detection, and more specifically, to a method for detecting and estimating a three-dimensional point cloud object based on a template. Background technique [0002] In the out-of-order grasping application of industrial parts, template-based object detection and pose estimation technology is one of its core technologies. Common template matching algorithms include LINEMOD, PPF (Point-Pair Feature), etc., which can be divided into two steps: 2D or 3D feature extraction and feature matching. Commonly used point cloud features include PPF, image gradient features, etc. Since the features used are primary visual features, the extracted features are easily disturbed by various factors such as lighting conditions, shadows, missing point clouds, and occlusions, resulting in a large number of gross errors in feature matching. Therefore, the existing mainstream algorithms eliminate a large number of match...

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 Patents(China)
IPC IPC(8): G06T7/00G06T7/50G06T7/73G06K9/62G06N3/04G06N3/08
CPCG06T7/001G06T7/50G06T7/73G06N3/08G06T2207/10028G06T2207/20081G06T2207/20084G06T2207/20076G06T2207/30108G06N3/044G06N3/045G06F18/241
Inventor 王磊吴伟龙周建品李争
Owner 视研智能科技(广州)有限公司