Retinal Blood Vessels Segmentation using Local Ternary Pattern

Abstract

Retinal vessels segmentation is an important task in medical image analysis. This paper presents a robust vessel segmentation algorithm. The proposed method initially, isolates the Green channel from the input image. On Green channel, Local Ternary Patterns (LTP) is applied to obtain the robust vessel features. To remove noise from the small vessel features, multiple morphological filters, such as spur, majority, opening, and fill operations are used and finally we get the prominent vessels pixel and segmented image. The proposed LTP based vessel segmentation algorithm outperforms recently published methods in terms of Accuracy in some cases on DRIVE and STARE datasets. The proposed method shows a noteworthy and precise addition for imminent computer-aided diagnosis applications.

Keywords

• Retinal Images, Segmentation, Vessels