Comparative Study on Retinal Blood Vessel Detection

In this work, an algorithm has been developed for automatic detection of blood vessels in normal and abnormal retinal images. The algorithm comprises of four steps: second derivative Gaussian function filtering, local entropy thresholding, median filtering and length filtering. The algorithm so developed has been tested on twenty ocular fundus images. The results of this algorithm are compared with those obtained from other methods. Other methods which are considered in this work for comparison are algorithm as described in [15], imageJ and Hand labeled ground truth segmentation. The comparative results reveal that the algorithm developed in this work is efficient with respect to its accuracy, speed, simplicity and PSNR values. Statistical two-way analysis of variance (ANOVA) is also performed on the comparative results.