• Aucun résultat trouvé

New discrepancy measures for segmentationevaluation.

N/A
N/A
Protected

Academic year: 2021

Partager "New discrepancy measures for segmentationevaluation."

Copied!
1
0
0

Texte intégral

(1)

IEEE International Conference on Image Processing (ICIP) 2003

New discrepancy measures for segmentation evaluation.

A. B. Goumeidane, M. Khamadja, B. Belaroussi, H. Benoit-Cattin, C. Odet

Abstract : In this paper, we propose new evaluation measures for scene segmentation results, which are based on computing the difference between a region extracted from a segmentation map and the corresponding one on an ideal segmentation. The proposed measures take into account separately both under and over detected pixels. It also associates in its computation the compactness of the region under investigation.

Keywords : Image segmentation evaluation, Discrepancy methods., Difference measures, Refence segmentation

Centre de Recherche en Technologies Industrielles - CRTI - www.crti.dz

Powered by TCPDF (www.tcpdf.org)

Références

Documents relatifs

In this paper, we propose new evaluation measures /or scene segmentation results, which are based on computing the difference between a region extracted from

This approach consists of six measures that aim, together, to compute in the sense of segmentation evaluation, the discrepancy between a region extracted from a segmentation map and

This method consists of an association of measures that have the purpose to compute, in the sense of segmentation evaluation, the difference between two regions, one extracted from

This method consists of an association of measures that have the purpose to compute, in the sense of segmentation evaluation, the difference between two

The potential security zones that are progressively identified are kept or not depending on the constraints analysis per- formed on IICS elements that are involved in the new

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des

From top to bottom: (1) Input se- quence (with groundtruth overlaid); (2) Segmentation results from regular 3D MRFs with appearance-based observation model (544 pixels

Furthermore during the linking process it is unnecessary to create code elements with only one sub-code element because if the algorithm did not find any connected element on the