Abstract: Many diseases cause harm to the human eye. A diabetic retinopathy could be one of eye disease and it was the essential blindness reason. Diabetic retinopathy earliest marks are the swelling of eye blood vessels that lead to being lesions in the retina. In this research a new algorithm for detecting hard exudates in retina by automated way is proposed. It uses a new algorithm which is based on triple circular segmentation followed by filtering using Kalman filter. Morphological operation is also applied to enhance and to refine the Kalman filter results. The suggested method was tested with 60 images from different databases images and achieved performance was taken with 20 images that obtained sensitivity; 86%, specificity; 99.9%, accuracy; 99.5 % and positive predictive value; 88% which demonstrate the main considerable value of the proposed algorithm. These results are considered a superior performance as compared with the earlier current techniques and are strong to image quality variability.