High-resolution 3D surface strain magnitude using 2D camera and low-resolution depth sensor

摘要: Measures the 3-D surface strain impacted on face during facial expressions.Describes an automatic approach for calibrating Kinect with external camera.Method shows high correlation between maps calculated from two views.Method robust to multiple depth resolutions.Over 100 subjects and 600 expressions used testing. Generating 2-D of provides a useful feature that captures bio-mechanics skin tissue, has had wide application in several research areas. However, most applications have been restricted collecting data single pose. Moreover, methods strictly use images motion estimation can potentially suppress large strains because projective distortions caused by curvature face. This paper proposes method allows using low-resolution sensor. The algorithm consists automatically aligning rough approximation resolution camera image. We provide experimental results demonstrate robustness dataset collected Microsoft synchronized cameras, as well 101 publicly available expression video database (BU4DFE).