In　this　paper,　a　novel　crack　and　void　identification　method　is　proposed　based　on　the　scaled　boundary　finite　element　method　(SBFEM)　with　the　pixel-based　quadtree　mesh　and　the　parameter　level　set　method　(PLSM),　in　which　the　particle　swarm　optimization　method　(PSO)　is　applied　to　optimize　the　objective　function.　It　fully　exploits　the　advantages　of　fast　image-based　quadtree　mesh　generation　and　efficient　assembly　of　global　stiffness　matrix　to　accomplish　a　fast　analysis　of　the　forward　problem　and　accelerate　the　efficiency　of　the　inversion.　Furthermore,　multiple　level　set　functions　are　introduced　to　represent　cracks　and　voids　simultaneously.　The　application　of　PSO　dexterously　avoids　the　difficulty　of　solving　sensitivity.　Numerical　examples　show　that　the　proposed　method　can　identify　the　position　and　shape　of　cracks　and　voids　accurately　by　discussing　different　influencing　factors.　In　addition,　the　method　has　shown　certain　anti-noise　performance.