An efficient and precise retinal vessel centerline extraction approach based on ridge detection is proposed. Firstly, the candidate retinal vessel centerline is obtained by using ridge detection. Then, illumination equalization and the multi-scale morphological enhancement are applied. Finally, the Otsu dual threshold segmentation method and Otsu single threshold segmentation method are respectively applied to the enhanced image for each scale in the retina ROI. The final vessel centerline is obtained with logic AND operation between the candidate centerline and the summation of segmentation results of each scale. The proposed method is evaluated by using database of 25 fluorescein retinal images and some color retinal images. The experiments demonstrate the method can extract the slim vessels and low contrast vessels well, while persevering vessels connectivity well. The extracted vessels are compared with the skeleton of the manual vessel image, and the average overlap rate is 83.5%. The proposed method outperforms the Hoover algorithm and conservatively hand-labeled method for the small vessels extraction and connectivity.