Retinal Fundus Detection Using Skew Symmetric Matrix

The retina is the light sensitive tissue, lining the back of our eye. Light rays are focused onto the retina through our cornea, pupil and lens. The retina converts the light rays into impulses that travel through optic nerve to our brain, where they are interpreted as the images. The task of manually segmenting fundus from retina images is generally time-consuming and difficult. In most settings, the task is done by marking the fundus regions slice-by-slice, which limits the human rater’s view and generates distorted images. Manual segmentation is also typically done largely based on a single image with intensity enhancement provided by an injected contrast agent. In the current research the fundus is detected and extracted in retinal image. Fundus is distinguished from normal tissues by their image intensity, threshold-based or region growing techniques. The fundus in this approach is detected with the help of geometric features. Skew symmetric matrix is used to avoid any angular orientation. In this approach the accuracy on fundus is quite promising. Accuracy of fundus detection is improved according to the area and the acceptance rate .In this approach ,once the image is loaded, it is filtered and normalized. Then superpixels are generated using linear iterative clustering approach and the features are generated. From the available set of features, some of the features are selected using sequential forward selection approach .Classifier is constructed in order to determine different classes in a test image. Proposed work is two class problem in which algorithm is applied that consists of skew symmetric matrix .Experimental results show substantial improvements in the accuracy and the performance of fundus detection as well as in false acceptance rate and false rejection rate.