A Methodology for Hierarchical Image Segmentation Evaluation
This paper proposes a method to evaluate hierarchical image segmentation procedures, in order to enable comparisons between different hierarchical algorithms and of these with other (non-hierarchical) segmentation techniques (as well as with edge detector
- PDF / 1,078,772 Bytes
- 13 Pages / 439.37 x 666.142 pts Page_size
- 23 Downloads / 231 Views
Faculty of Mathematics, Complutense University of Madrid, 28040 Madrid, Spain {jtrodrig,cguada,jayage,javier_montero}@ucm.es Faculty of Statistics, Complutense University of Madrid, 28040 Madrid, Spain [email protected]
Abstract. This paper proposes a method to evaluate hierarchical image segmentation procedures, in order to enable comparisons between different hierarchical algorithms and of these with other (non-hierarchical) segmentation techniques (as well as with edge detectors) to be made. The proposed method builds up on the edge-based segmentation evaluation approach by considering a set of reference human segmentations as a sample drawn from the population of different levels of detail that may be used in segmenting an image. Our main point is that, since a hierarchical sequence of segmentations approximates such population, those segmentations in the sequence that best capture each human segmentation level of detail should provide the basis for the evaluation of the hierarchical sequence as a whole. A small computational experiment is carried out to show the feasibility of our approach. Keywords: Image segmentation Hierarchical Edge-based image segmentation evaluation
network
clustering
1 Introduction Image segmentation, understood as the identification of connected and homogenous regions of an image, is an essential tool in today applications of image processing. There is a wide variety of approaches to segmentation as well as of techniques to perform it [2], in the same way that also different approaches exist regarding how to evaluate the performance of segmentation procedures (see for instance [3, 6, 9, 10, 12, 13]). A technique based on segmentation is that of hierarchical image segmentation [1], whose aim is to produce a consistent sequence of segmentations identifying the objects in an image with different levels of detail. Hierarchical segmentation is a relevant extension of segmentation since various applications (see for instance [5]) require different detail levels to be simultaneously available, in such a way that object identification is consistent through the different detail levels. However, contrarily to non-hierarchical segmentation, there are relatively few hierarchical segmentation procedures, and these usually find the problem that there are not clearly accepted approaches on how to evaluate a hierarchical segmentation algorithm. © Springer International Publishing Switzerland 2016 J.P. Carvalho et al. (Eds.): IPMU 2016, Part I, CCIS 610, pp. 635–647, 2016. DOI: 10.1007/978-3-319-40596-4_53
636
J. Tinguaro Rodríguez et al.
The aim of this paper is to propose a method to enable evaluation of hierarchical segmentation procedures by an edge-based segmentation evaluation approach (see [3, 10]), possibly today’s most widely accepted and extended segmentation evaluation methodology. The main argument behind the proposed method is that a hierarchical sequence of segmentations captures or approximates the different possible levels of detail humans may use or that may be needed in an
Data Loading...
