Fully automatic expression-invariant face correspondence

摘要: We consider the problem of computing accurate point-to-point correspondences among a set human face scans with varying expressions. Our fully automatic approach does not require any manually placed markers on scan. Instead, learns locations landmarks present in database and uses this knowledge to automatically predict these newly available The predicted are then used compute between template model To accurately fit expression scan, we use as blendshape model. algorithm was tested faces different ethnic groups strongly Experimental results show that obtained correspondence is both highly consistent for most 3D models.