Non-Lambertian Object Reconstruction Method Based on Group Sparse Photometric Stereo Vision

A photometric stereo, non-Lambertian technology, applied in 3D modeling, image data processing, instruments, etc., to achieve a wide range of applications, saving construction costs and data acquisition time.

Active Publication Date: 2017-06-23
ZHEJIANG UNIV
View PDF3 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to solve the problem that the existing photometric stereo vision technology requires a large number of light sources for the reconstruction of strong and high-gloss objects, and provides a method for realizing photometric stereo vision by special arrangement of light sources and applying the idea of ​​grouping and sparseness

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
  • Non-Lambertian Object Reconstruction Method Based on Group Sparse Photometric Stereo Vision
  • Non-Lambertian Object Reconstruction Method Based on Group Sparse Photometric Stereo Vision
  • Non-Lambertian Object Reconstruction Method Based on Group Sparse Photometric Stereo Vision

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0057] Such as image 3 As shown, the normal recovery accuracy achieved by the method of the present invention is compared with the existing method on the simulation data. The two methods are the method based on the Lambertian assumption and the method using Bayesian learning. The simulation data is selected from the 100 materials in the MERL database given by W.Matusik et al. [A data-driven reflectance model, ACM Transactions on Graphics22(3), 759–769(2003).] using the Cook Torrance model to simulate combine. Use the drawn image to restore the normal, compare it with the real normal, and calculate the angle error. The reconstruction errors are reordered according to the errors in the present invention from small to large. For most materials, the method of the present invention is better than the traditional method.

Embodiment 2

[0059] Such as Figure 4 As shown, the method of the present invention is used to realize 3D reconstruction of an actual scene and compared with the traditional method. Among them, (a) is a photo of the object to be reconstructed under natural conditions; (b) is a reconstruction result map based on the Lambertian assumption, (c) is a reconstruction result map based on the method of Bayesian learning: (d) is a reconstruction result map using the present invention The reconstruction result plot obtained by the method. From the original photo, it can be found that the highlight spot of the object is wide and has strong highlights. Therefore, the method based on the Lambertian hypothesis produces a large unreasonable bias at the highlights, and even if the Bayesian learning method is used to perform high-light There are still large discontinuities in the detection, but the method of the present invention obtains a relatively smooth reconstruction result, which is more consistent ...

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 group sparsity based photometric stereo method for realizing non-Lambert object reconstruction. The method comprises steps as follows: normalized intensity of each pixel of a non-Lambert object in different light source directions is extracted and is grouped, highlight detection is realized with a group sparsity method, and normal restoration and three-dimensional reconstruction are completed finally. According to the method, structural information of the light source direction is sufficiently utilized, and the reconstruction of the complex non-Lambert object can be realized through a smaller number of light sources, so that the reconstruction of objects with different reflection characteristics can be realized with lower cost in shorter collection time.

Description

technical field [0001] The invention relates to restoring the normal direction of the scene surface by using the photometric stereo vision technology, and can obtain ideal results under the condition that the number of light sources is relatively small and the high light of the object to be reconstructed is strong. Background technique [0002] Photometric stereo vision has the characteristics of non-contact, high precision, and better recovery of details on the surface of objects. Therefore, it has always been a relatively popular research direction and has great application potential. A typical photometric stereo vision system includes a camera and several light sources. Under ideal Lambertian reflection conditions, the intensity of the pixel in the image satisfy: [0003] [0004] Where ρ is the product of the linear ratio of the scene surface reflectance to the camera, n is the normal direction of the scene surface, s is the light source intensity, and l is the li...

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): G06T17/00
Inventor 沈会良韩天奇
Owner ZHEJIANG 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