Fundus　abnormalities　were　usually　diagnosed　by　clinical　ophthalmologists　using　the　equipment　like　ophthalmoscope,　which　was　influenced　by　complex　method,　subjective　fault　and　low　efficiency.　Fortunately,　fundus　can　be　shown　in　the　key　structures　of　the　fundus　digital　image.　Hence,　it　is　necessary　to　develop　a　computer-aided　fundus　image　processing　system　as　a　tool　to　assist　in　the　diagnosis　diseases.　Algorithms　were　proposed　including　the　pre-processing　such　as　contrast　enhancement　and　spatial　filtering,　binaryzation,　morphology,　boundary　extraction　and　skeleton　extraction.　This　paper　preliminary　summarized　the　key　technologies　of　segmentation　on　the　structures　of　optic　cup,　optic　disc　and　blood　vessel.　Experiments　were　performed　in　MATLAB.　Automatic　algorithms　were　evaluated　to　be　able　to　detect　the　key　structures　of　the　fundus　digital　image.　It　is　expected　to　be　applied　in　computer-aided　fundus　image　processing　system.　©　(2013)　Trans　Tech　Publications,　Switzerland.