Detection of retinal hemorrhage in color fundus image using splat feature segmentation

Diabetes is a worldwide disease which is one of the main reason for blindness in the older age of any human community world-wide. Advanced level of diabetes leads to retinal hemorrhage. There is no efficient algorithm to detect the presence of hemorrhage. We have surveyed many algorithms and also recognized their efficiency. A new algorithm is proposed to detect the presence of hemorrhage with maximum efficiency and accuracy. The algorithm works by partitioning the image into differentiated segments covering the entire retinal image. These segments are denoted by splats. Each splat here establishes a set of information which helps us to extract the appropriate boundary. The pixels are grouped by the similarity of the color, intensity and spatial location. Retinal ischemia and weak blood vessels are the main reasons for the occurrence of haemorrhage. The new algorithm is based on the segmentation which is grouped by the colour, intensity and the spatial of the entire image.