Abstract:
Research Work: This Article focuses the perspective of retinal blood vessel classification and detection in digitized retinal images. The color-based channels like green and red channels extraction processes lead to the enhanced identification of retinal blood vessel through the performance measures using Receiver Operating Characteristics (ROC). Processes with Methodologies: This proposed system uses the digital images of the databases diaretdb0 and diaretdb1. The retinal vascular structures of these two databases are very precise and perfect. The resized image further involved with two different progressions of color-channel (Red and Green channels particularly) conversion. These extracted features are enhanced with CLAHE (Contrast Limited Adaptive Histogram Equalization) technique. The morphological operations performed for optic disk, noise and border removal. The difference between its’ MATLAB based erosion and dilation processes on basis of Canny-Edge Detection technique with Threshold value. These kind of performance measure of the system learned through Receiver Operating Characteristics (ROC) with Sensitivity (Se) and Specificity (Sp) using True Positive (P), True Negative (P1), False Positive (N) and False Negative (N1) to be used for determining the ‘Authorization’ or ‘Unauthorization’. The “Count of Pixels” indicates area of blood vessels in respect with Red and Green channels based features are calculated for gaining accuracy of the retinal vascular structure of an image. Result: The performance measure with ROC is applied for retinal blood vessel identification. This kind of systemic processes may be a new perspective in classification, detection and identification of retinal blood vessels.
Page(s):
205-212
DOI:
DOI not available
Published:
Journal: Journal of Theoretical and Applied Information Technology, Volume: 59, Issue: 1, Year: 2014