Topological Active Volumes
- PDF / 1,740,952 Bytes
- 9 Pages / 600 x 792 pts Page_size
- 72 Downloads / 185 Views
Topological Active Volumes N. Barreira Grupo de Visi´on Artificial y Reconocimiento de Patrones (VARPA), LFCIA, Departamento de Computaci´on, Universidade da Coru˜na, 15071 A Coru˜na, Spain Email: [email protected]
M. G. Penedo Grupo de Visi´on Artificial y Reconocimiento de Patrones (VARPA), LFCIA, Departamento de Computaci´on, Universidade da Coru˜na, 15071 A Coru˜na, Spain Email: [email protected] Received 29 December 2003; Revised 4 February 2005 The topological active volumes (TAVs) model is a general model for 3D image segmentation. It is based on deformable models and integrates features of region-based and boundary-based segmentation techniques. Besides segmentation, it can also be used for surface reconstruction and topological analysis of the inside of detected objects. The TAV structure is flexible and allows topological changes in order to improve the adjustment to object’s local characteristics, find several objects in the scene, and identify and delimit holes in detected structures. This paper describes the main features of the TAV model and shows its ability to segment volumes in an automated manner. Keywords and phrases: image segmentation, 3D reconstruction, active nets, active volumes.
1.
INTRODUCTION
The scene in a 3D image is usually made of an object or several objects. The segmentation task consists of isolating the points that belong to each object from the image and integrating them into a coherent and consistent model of the detected structures. Unfortunately, the detection of objects inside 3D data is difficult due to the complex topology of the objects and the cost of the operations in a 3D space. Furthermore, the available means of acquisition introduce noise in data, so direct segmentation is often inadequate and the use of an advanced segmentation technique is needed. There are many approaches for segmentation proposed in the literature. Nowadays, level set methods and deformable models have become two of the most promising techniques for segmentation, mapping, tracking, and modelling tasks. Level set methods were introduced by Osher and Sethian [1] and have found several applications in image processing, such as segmentation and reconstruction of complex shapes [2, 3] due to their ability to perform topological transformations. However, level sets’ formulation is more complex than deformable models’. Deformable models were introduced by Kass et al. [4] in 2D as explicit deformable contours and generalised to the 3D case by Terzopoulos et al. [5]. In recent years, many models have been developed for the treatment of 3D scenes [6, 7, 8]. They have tried to solve some of the limitations of classical deformable models, such
as the sensitivity to the initialisation or the parametric definition of the model, which restricts their topology to segment simple objects. These models have also adapted their operation to concrete domains by means of the definition of different energy terms or the use of several topologies. In particular, some models have increased their degree of flexibility
Data Loading...
