Machine　vision　is　now　an　excellent　tool　for　testing　a　wide　range　of　industrial　products　such　as　textiles,　printed　circuit　boards,　electronic　component　tags,　integrated　circuits　(ICs)　and　machine　tools.　In　this　paper,　we　propose　an　optical　lens　detection　system　based　on　machine　vision,　which　can　be　widely　used　in　the　defect　detection　of　camera　lenses,　eyeglass　lenses　and　other　related　optical　components.　Firstly,　the　hardware　composition　of　the　entire　system　and　the　flow　of　the　control　system　were　introduced.　Secondly,　through　the　image　preprocessing　algorithm　such　as　image　denoising　and　contrast　enhancement,　combined　with　the　algorithm　of　Canny　operator　and　morphological　processing,　the　reasonable　image　processing　results　were　obtained.　Finally,　a　decision　tree　classifier　based　on　C4.5　algorithm　was　established　according　to　the　shape　feature　of　the　defect,　which　realized　the　automatic　recognition　and　classification　of　surface　defects　on　optical　components.　The　experimental　results　show　the　effectiveness　of　this　method　in　the　detection　of　optical　component　defects,　which　provides　a　theoretical　basis　for　the　functional　analysis　of　optical　components　in　the　future.　Meanwhile,　it　can　be　extended　to　defect　detection　in　many　fields　such　as　circuit　boards　and　integrated　circuits.　©　2019　IEEE.