Non-invasive contrast normalisation and denosing technique for the retinal fundus image

Toufique Soomro, Junbin Gao

Research output: Contribution to journalArticle

Abstract

Diabetic retinopathy (DR) causes vision loss insufficiency due to impedimentrising from high sugar level conditions disturbing the retina. The progressionof DR occurs in the foveal avascular zone due to loss of tiny blood vessels of thecapillary network. Due to image acquisition process of the fundus camera, the colourretinal fundus image suffers from varying contrast and noise problems, varying contrastand noise problem in fundus image can be overcome. The technique has beenimplemented. The technique is based on the Retinex algorithm along with stationarywavelet transform. The technique has been applied on 36 high-resolution fundusimage database contain the 18 bad quality images and 18 good quality images. TheRETinex and stationary wavelet transform (RETSWT) is developed with the denoisingtechnique Based on stationary wavelet transform. RETSWT achieves an averagePSNR improvement of 2.39 dB the contrast improvement factor of 5.5 for good qualityimages while it achieves an average PSNR improvement of 2.20 dB for bad qualityimages with contrast improvement factor of 5.31. The RETSWT image enhancementmethod potentially reduces the need for the invasive fluorescein angiogram in DRassessment.
Original languageEnglish
Pages (from-to)265-279
Number of pages15
JournalAnnals of Data Science
Volume3
Issue number3
DOIs
Publication statusPublished - Sep 2016

Fingerprint Dive into the research topics of 'Non-invasive contrast normalisation and denosing technique for the retinal fundus image'. Together they form a unique fingerprint.

  • Cite this