Diabetic Retinopathy is a disease that strikes the retina of the eye in patients who have diabetes mellitus. Medical examination against sufferers of Diabetic Retinopathy is done with observation directly by eye surgeons. In this case, eye retinal images are taken using the camera the fundus. Retinal fundus Photographs resulted from fundus cameras usually are not able to give a clear picture against the retinal blood vessels. This makes it difficult for doctor to analyze images of the retina. It takes a relatively long time to find out the results of the examination. Overcoming these weaknesses, a system was built using a computational model to change the retina image pixel retina into a feature of the retina. So it can help the doctor to decide medical actions quickly and precisely. In this research a system that can detect and classify the diabetic retinopathy was created, using local binary pattern method to extract the characteristics and learning vector quantization method for the classification process. Local binary pattern will generate an image of a uniform which has the most image information. The image will be a characteristics vector as input to the method of classification learning vector quantization. The results of the testing show that the number of levels of extraction of characteristics affect the results of the classification, in this case the best accuracy results is 85%.Keywords— Diabetic Retinopathy, Fundus Image, Local Binary Pattern, Pre-Processing, Learning Vector Quantization.