In　the　Czochralski　method　for　crystal　growth,　the　uniformity　of　diameter　growth　is　critical　for　ensuring　crystal　quality.　Currently,　this　process　predominantly　depends　on　manual　visual　inspection,　which　is　inherently　subjective　and　lacks　precision.　To　address　these　challenges,　this　paper　proposes　an　online　detection　method　for　assessing　the　irregular　cross-sectional　dimensions　of　crystal　rods　through　image　monitoring.　Initially,　real-time　monitoring　data　of　crystal　growth　is　processed　to　generate　a　sequence　of　images,　followed　by　preprocessing　steps　that　include　denoising　and　contrast　enhancement.　Subsequently,　the　edges　of　the　crystal　rod　are　identified　using　weighted　pixel　values　and　gradients,　with　the　similarity　of　radius　sequences　recorded　in　various　directions　serving　as　the　fitness　value　for　potential　center　points.　The　whale　optimization　algorithm　is　then　utilized　to　identify　the　center　point　of　the　cross-section.　Finally,　the　cross-sectional　shape　is　delineated,　and　by　incorporating　camera　calibration,　pixel　dimensions　are　converted　to　physical　dimensions,　thereby　facilitating　the　online　detection　of　the　crystal　rod＇s　cross-sectional　size.　Experimental　results　obtained　from　neodymium-doped　yttrium　vanadate　(Nd:YVO4)　crystals　indicate　that　this　method　achieves　an　average　absolute　error　of　0.38　mm　and　0.24　mm　in　measuring　the　long　and　short　axes　of　the　cross-section,　respectively,　with　an　average　detection　time　of　0.81　seconds.　This　approach　has　the　potential　to　assist　or　even　replace　the　labor-intensive　manual　monitoring　of　crystal　growth.　©　2025　IEEE.