Determining the orientation of the observed object in threedimensional space using stereo vision methods

Abstract

The second order extended Kalman filter and Markov nonlinear filter for data processing in interferometric systems P Ermolaev and M Volynsky Abstract. The task of matching image of an object with its template is central for many optoelectronic systems. Solution of the matching problem in three-dimensional space in contrast to the structural alignment in the image plane allows using a larger amount of information about the object for determining its orientation, which may increase the probability of correct matching. In the case of stereo vision methods for constructing a three-dimensional image of the object, it becomes possible to achieve invariance w.r.t. background and distance to the observed object. Only three of the orientation angle of the object relative to the camera are uncertain and require measurements. This paper proposes a method for determining the orientation angles of the observed object in three-dimensional space, which is based on the processing of stereo image sequences. Disparity map segmentation method that allows one to ensure the invariance of the background is presented. Quantitative estimates of the effectiveness of the proposed method are presented and discussed. IntroductionThe fundamental problem arising in the process of matching is the problem of prior uncertainty due to the variability of object and shooting conditions (angle and lighting), as well as a lack of information about the properties of object. There are two basic approaches to the determination of the angle of observation or combination of reference and current visual information that can be divided by the fact whether there is alignment in the image plane or in three-dimensional space. In the first case, the current image should be aligned with the template image formed for the corresponding angle, distance, lighting conditions and background [1]. In this approach, the number of templates for each object can be quite large, and the process of aligning with each of them to establish the angle could become resource consuming.In the second case, the alignment should be made for the reconstructed disparity map and threedimensional model of the object [2]. Data on the three-dimensional shape of an object is initially presented in a form invariant to lighting conditions. Invariance w.r.t. background is also easily reached by depth map segmentation as a background has zero disparity. Furthermore, invariance w.r.t. distance to the object which is extracted by stereovision methods is automatically achieved. These features of the second approach can provide a higher probability of correct matching. To determine the relative orientation of the observed object in the processing sequence of stereoscopic images one can use the methods of alignment the three-dimensional model of the object reconstructed on the basis of dense depth maps for images from a sequence of stereo pairs. In this paper, we propose a method that uses the approach described above to determine the orientation angles of the object in three-dimensional space.