An automatic approach to Hemorrhages detection

Microaneurysms and hemorrhage detection is the first step of diabetic retinopathy (DR). The red lesions fundus images are used to analyse the hemorrhage. Occurance of blindness can be reducing by early detection of microaneurysms. In this paper we proposed robust approach for automatic red lesions detection in fundus images based on mathematical morphology, pixel classification and various filters. Geometric characteristics are well providing by the mathematical morphology and also provide image segmentation which is used for further processing. FP reduction is based on four parameters, Area, aspect ratio, mean and foveal filter. The proposed algorithm provides promising results to detect microaneurysms.