查看文章

The paradigm of Fuzzy Morphology extends the concept of binary morphology to handle grayscale images. Fuzzy Morphology provides meaningful, local and simple operations that, when properly combined, form powerful transformations. We use this approach to segment out vessels in eye-fundus images, which can be used to diagnose medical conditions such as diabetic retinopathy. To automatically estimate the presence of such conditions, distinguishing vessels from other artifacts becomes a necessary initial step. To address the problem of segmenting curvilinear-like objects such as vessels, our methodology consists on applying the same structuring element rotated several times. We construct a vessel segmentation method and compare it with current state-of-the-art alternatives, showing the potential of our approach.