Project information
- Category: Retinal Vessel Segmentation
- Data: DRIVE, STARE, CHASE_DB1
- Project date: November, 2017
- Project URL: https://doi.org/10.3390/sym9110276
Retinal Vessel Segmentation via Structure Tensor Coloring and Anisotropy Enhancement
Retinal vessel segmentation is essential for developing diagnostic systems for retinal diseases. This study proposes a fully automated segmentation method, enhancing vessels with a Frangi Filter, followed by Structure Tensor analysis, anisotropy enhancement, and Contrast Limited Adaptive Histogram Equalization. The final segmentation is achieved using the Otsu algorithm and tensor coloring, with post-processing for refinement. The method achieved high performance on the DRIVE, STARE, and CHASE_DB1 datasets, with mean accuracies of 0.9183, 0.9442, and 0.9236, respectively, and execution times of 6.104, 6.4525, and 18.8370 seconds.