DocumentCode
1507715
Title
Recovering surface normal of specular object by Hough transform method
Author
Liu, Richard ; Han, Jinguang
Author_Institution
Sch. of Electron. & Inf. Eng., Xi´an Jiaotong Univ., Xi´an, China
Volume
4
Issue
2
fYear
2010
fDate
6/1/2010 12:00:00 AM
Firstpage
129
Lastpage
137
Abstract
In order to recover surface normal and albedo of specular objects under the condition of large shadows and highlights by photometric stereo, a novel robust method utilising Hough transform is proposed. The photometric image sequence is taken by a fixed camera when the light source is rotated around the optical axis of the camera. Since the bidirectional reflectance distribution function of many surfaces can be approximated as a combination of a diffuse component and a specular component, the diffuse part of the intensity at a fixed point on the surface obeys a sine curve when the light source rotates. The surface normal and albedo information are hidden in the sine curve and can be revealed by means of Hough transform. The bilateral symmetry of the sine curve is used to narrow the searching range of the initial phase and to speed up the Hough transform. Experiments performed on synthetic and real objects with large shadows and strong specular reflectance show the better performance of the proposed method.
Keywords
Hough transforms; image sequences; stereo image processing; Hough transform method; bidirectional reflectance distribution function; photometric image sequence; photometric stereo; sine curve; specular object; surface normal recovering;
fLanguage
English
Journal_Title
Computer Vision, IET
Publisher
iet
ISSN
1751-9632
Type
jour
Filename
5475473
Link To Document