Structure-from-motion method for multi-video sequences

A technology of video sequence and structure method, which is applied in image data processing, instrumentation, calculation, etc.

Active Publication Date: 2012-09-12
ZHEJIANG SENSETIME TECH DEV CO LTD
View PDF3 Cites 11 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to the large number of optimization variables, optimization efficiency and memory requirements are still the two bottlenecks of the motion inference structure.

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
  • Structure-from-motion method for multi-video sequences
  • Structure-from-motion method for multi-video sequences
  • Structure-from-motion method for multi-video sequences

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0146] Using a structural approach to motion inference for multiple video sequences, camera motion paths and scene point clouds are estimated for a single 18 video sequence capturing a large-scale scene. The input sequence contains a total of 27600 frames, see the snapshot figure 1 (a), the recovered camera path and scene point cloud see figure 1 (b). figure 1 In (c), the recovered camera path is overlaid on the satellite image of Google Earth to verify the solution accuracy. Including all calculations, the average processing time per frame is 0.6 seconds. All calculations do not involve hardware acceleration and run on a single thread.

[0147] The steps of the motion inference structure method for multiple video sequences are as follows:

[0148] 1) Based on the SIFT feature description quantity, use the continuous feature tracking algorithm and the discontinuous feature matching algorithm to match the SIFT feature points corresponding to the same scene points distribute...

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 structure-from-motion method for multi-video sequences, comprising the following steps of: 1) using a continuous characteristic tracking algorithm and a non-continuous characteristic matching algorithm on the basis of SIFI characteristic description values, and matching SIFT characteristic points which are corresponding to a same scene point and distributed in different images; 2) using a structure-from-motion algorithm on basis of matching of the SIFT characteristic points which are corresponding to the same scene point and distributed in different images, recovering corresponding sub-images of video sequences, and registering the corresponding sub-images of the video sequences in a unified coordinate system; and 3) using a segment-based progressive optimization algorithm to iteratively spread and eliminate errors existing in the corresponding sub-images of the video sequences. The structure-from-motion method for the multi-video sequences can efficiently match characteristic locuses distributed in non-adjacent sub-sequences, improve the solving quality of each sub-image, break through the memory and efficiency bottleneck of the traditional solving method for large-scale scenes, and globally and efficiently optimize the three-dimensional structure of the entire scene and camera variables in a limited memory environment.

Description

technical field [0001] The invention relates to a motion inference structure, in particular to a motion inference structure method for multiple video sequences. Background technique [0002] Motion inference structure technology refers to the automatic estimation of the three-dimensional position of the feature points in the scene and the camera motion parameters corresponding to each image from the image collection or video sequence. Finding the corresponding positions of the feature points in the image is crucial to the solution quality of the motion inference structure. Compared with image collections, video sequences contain richer geometric information and structural information of the scene. For video sequences, a common method is to use feature point tracking algorithms between every two adjacent frames, such as Lucas, B.D., Kanade, T.: An iterative image registration technique with an application to stereo vision. In: IJCAI, pp. 674-679(1981); Shi, J., Tomasi, C.: ...

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): G06T7/20G06T7/00
Inventor 章国锋鲍虎军刘浩敏
Owner ZHEJIANG SENSETIME TECH DEV CO LTD
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