Title:	EVALUATION OF TWO RECONSTRUCTION SYSTEMS BASED ON STEREOVISION
Author(s):	Amira Soudani, Ezzeddine Zagrouba
ISBN:	978-972-8939-48-9
Editors:	Yingcai Xiao
Year:	2011
Edition:	Single
Keywords:	Computer vision, Stereovision, Calibration, Matching, Rectification, 3D Reconstruction.
Type:	Short Paper
First Page:	308
Last Page:	312
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	This paper presents an evaluation and comparison of two 3D reconstruction approaches. The first one is an existing sparse reconstruction which is based on the triangulation of extracted points of interest using calibration parameters. The results obtained by this approach present the outline of the object. The second one is improved dense reconstruction approach based on a generated disparity map. For this approach, we improve the stereo matching algorithm. In fact, the initial disparity map is computed using a weighted dissimilarity measure with application of adaptive correlation windows. Then, planes are modeled by a set of planar surface patches. First, an initial set of planes is estimated using only reliable pixels. Then, those planes will be merged to obtain a final representation of the scene. Thus, each pixel and region will be assigned to a disparity layer according to a cost function which is subject of minimization. The experimentations and evaluations show that the second approach gives efficient results concerning the recovered informations.
	Go Back