Please use this identifier to cite or link to this item: http://dspace.univ-temouchent.edu.dz/handle/123456789/2206
Full metadata record
DC FieldValueLanguage
dc.contributor.authorAbbes, Saliha-
dc.contributor.authorBelkhrredj, Nessrine-
dc.date.accessioned2024-02-13T13:41:52Z-
dc.date.available2024-02-13T13:41:52Z-
dc.date.issued2015-
dc.identifier.citationhttps://theses.univ-temouchent.edu.dz/opac_css/doc_num.php?explnum_id=1912en_US
dc.identifier.urihttp://dspace.univ-temouchent.edu.dz/handle/123456789/2206-
dc.description.abstractDiabetic retinopathy (DR) is the leading cause of blindness and visual impairment in adults. Early detection of these diseases by regular screening is particularly important to prevent vision loss. The implementation methods based on early detection of clinical signs of DR allow to greatly improve the diagnosis of these diseases. We propose in this paper an image analysis system incorporating fundus image processing techniques for the detection of lesions associated with these diseases namely: exudates, cotton wool spots. The methods of detecting retinal pathologies proposed in this paper use clustering algorithms such as k-means and fuzzy c-means- (FCM), en suite, we propose a hybrid method that used both methods of classification. The algorithms developed in the context of this work are tested on a set of images. The evaluation of the proposed methods is performed by a comparison of results between k-means and FCM.en_US
dc.subjectDiabetic retinopathy (DR), Unsupervised Classification, Clustering, K-means, Fc-means (FCM).en_US
dc.titleClassification non supervisée : application des images rétinienne couleursen_US
Appears in Collections:Electronique

Files in This Item:
File Description SizeFormat 
Classification non supervisée : application des images rétinienne couleurs.pdf196,75 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.