GPU-accelerated image segmentation based on level sets and multiple texture features
- PDF / 3,648,224 Bytes
- 23 Pages / 439.642 x 666.49 pts Page_size
- 7 Downloads / 193 Views
GPU-accelerated image segmentation based on level sets and multiple texture features Daniel Reska1
· Marek Kretowski1
Received: 20 April 2020 / Revised: 9 September 2020 / Accepted: 16 September 2020 / © The Author(s) 2020
Abstract In this paper, we present a fast multi-stage image segmentation method that incorporates texture analysis into a level set-based active contour framework. This approach allows integrating multiple feature extraction methods and is not tied to any specific texture descriptors. Prior knowledge of the image patterns is also not required. The method starts with an initial feature extraction and selection, then performs a fast level set-based evolution process and ends with a final refinement stage that integrates a region-based model. The presented implementation employs a set of features based on Grey Level Co-occurrence Matrices, Gabor filters and structure tensors. The high performance of feature extraction and contour evolution stages is achieved with GPU acceleration. The method is validated on synthetic and natural images and confronted with results of the most similar among the accessible algorithms. Keywords Image segmentation · Active contour model · Level set method · Texture analysis · GPU acceleration.
1 Introduction Image segmentation is one of the most fundamental problems in computer vision. Deformable models [34] are a successful class of segmentation algorithms based on the idea of a deforming shape that adapts to the desired image region. The fundamental form of the deformable model-based segmentation method was proposed by Kass et al. [25] as an active contour model (ACM), also known as a “snake”. The snake model is a parametric curve with an evolution process controlled by a set of external and internal energies. External energies attract the shape to the desired image area and move it towards the boundaries of the segmented region, while the internal forces control the contour smoothness. This method Daniel Reska
[email protected] Marek Kretowski [email protected] 1
Faculty of Computer Science, Bialystok University of Technology, Wiejska 45A, 15-351 Bialystok, Poland
Multimedia Tools and Applications
overcomes many problems, like image noise and boundary irregularities, and makes it easy to extend with new types of image features and constraints. These advantages made the classical ACM very influential and widely improved, e.g. with the introduction of expansion forces [10], edge-based vector field energies [58], and region-based image energies [44]. The original ACM also had some drawbacks, like difficulties in topological adaptability (requiring additional mechanisms [30]) and sensitivity to initialisation. A major improvement came with the incorporation of the level set framework, proposed by Osher and Sethian [37, 47]. Instead of the explicitly defined parametric contour, the level set approach proposed an evolving surface, where the contour is implicitly represented as a set of surface points that have their height equal to zero (the zero level set).
Data Loading...
