Diagnosing diseases through retinal vasculature has long been a focus of medical research. However, segmenting vessels from retinal images remains challenging due to variations in image intensity and retinal vessel thickness. Existing enhancement filters often provide non-uniform responses across vessels of different radii and are less effective at vessel edges, bifurcations, and areas with vascular pathologies. Non-vascular structures often degrade performance, and some low-contrast small vessels are hard to detect after several down-sampling operations. To solve these problems, this study explores the use of multiscale second-order local structures to develop a vessel enhancement filter based on the eigenvalues of the non-local and non-singular fractional Hessian matrix. The suggested method’s effectiveness is assessed using the publicly accessible dataset HRF and some authentic images from SMS Medical Hospital, Jaipur, India. Our findings indicate that the suggested method outperforms most listed techniques and achieves this with improved computational efficiency.