The　identification　and　visualization　of　submillimeter　microdefects　have　long　been　a　critical　focus　within　the　realm　of　ultrasonic　testing.　The　utilization　of　high-frequency　signals　is　imperative　due　to　the　small　acoustic　interaction　structure.　An　additional　benefit　of　using　lasers　is　their　non-contact　nature　and　ability　to　utilize　multiple　imaging　modes,　effectively　reducing　false　or　missed　detections　and　enhancing　credibility.　This　study　proposes　an　advanced　optimization　algorithm　for　imaging　and　quantitative　evaluation　of　submillimeter　microdefects　using　laser　ultrasonic　full　matrix　capture　(FMC)　and　total　focusing　method　(TFM).　The　analysis　of　acoustic　directionality　under　laser　excitation　and　detection　provides　a　foundation　for　mode　selection.　In　preprocessing,　the　multi-scale　principal　component　analysis　(MSPCA)　method　is　introduced　to　suppress　coherent　noise　and　blind　spots.　Moreover,　phase　coherence　methods　are　implemented　to　further　reduce　mode　artifacts,　background　noise,　and　residual　surface　acoustic　wave　(SAW).　Additionally,　multi-mode　imaging　and　mode　fusion　are　achieved　using　reflected　longitudinal　waves,　reflected　shear　waves,　converted　longitudinal　waves,　and　converted　shear　waves.　Experiments　were　carried　out　on　two　different　types　of　microdefects,　namely　side　drilled　holes　(SDHs)　and　blind　holes　(BHs),　resulting　in　high　resolution　TFM　imaging　with　a　minimum　BH　size　of　0.3　mm.　The　outcomes　demonstrate　the　efficacy　of　MSPCA　in　suppressing　coherent　noise,　with　further　enhancement　in　imaging　quality　after　phase　coherence　weighting.　In　addition,　the　-6　dB　threshold　of　multi-mode　images　also　provided　the　size　information.　This　research　is　poised　to　contribute　to　the　resolution　of　quality　assessment　challenges　in　powder　metallurgy,　welding,　and　additive　manufacturing　processes.