Paper Info
Title
Estimating 3D Shape from the Optical Flow of Photometric Stereo Images
Abstract
Photometric stereo (PS) images are obtained from a single camera under different illuminations, and have traditionally been employed for the estimation of surface gradient in computer vision. Recently, it has been suggested that such images can also be matched to yield relevant depth cues both in human and in artificial vision. Here, we analyse the optical flow information carried by pairs of PS images taken under slightly different illuminations: By modelling the displacement of the irradiance pattern over the imaged surfaces, due to the illumination change, as a rotation plus a small translational correcting factor, we are able to obtain good 3D surface reconstructions through a structure-from-motion least-squares approach. Our framework for shape estimation does not require knowledge of the reflectance map function, which is at the core of the traditional approach to photometric stereo.
Year
DOI
Venue
1998
10.1007/3-540-49795-1_22
IBERAMIA
Keywords
Field
DocType
photometric stereo,optical flow,surface gradient,structure-from-motion least-squares approach,shape estimation,imaged surface,computer vision,surface reconstruction,artificial vision,photometric stereo images,different illumination,ps image,least square,structure from motion
Computer vision,Surface reconstruction,Surface gradient,Computer science,Stereopsis,Photometry (optics),Image processing,Artificial intelligence,Optical flow,Photometric stereo,Computer stereo vision
Conference
ISBN
Citations 
PageRank 
3-540-64992-1
1
0.45
References 
Authors
4
1
Name
Order
Citations
PageRank
José R. A. Torreão15910.18