A pose-independent method for 3D face landmark formalization

作者：Vezzetti Enrico^*; Moos Sandro; Marcolin Federica; Stola Vincenzo

来源：Computer Methods and Programs in Biomedicine, 2012, 108(3): 1078-1096.

DOI：10.1016/j.cmpb.2012.07.008

• Landmark extraction
• 3D face
• Differential Geometry
• PCA
• Procrustes Analysis

Recently, 3D landmark extraction has been widely researched and experimented in medical field, for both corrective and aesthetic purposes. Automation of these procedures on three-dimensional face renderings is something desirable for the specialists who work in this field. In this work we propose a new method for accurate landmark localization on facial scans. The method relies on geometrical descriptors, such as curvatures and Shape Index, for computing candidate and initial points, and on a statistical model based on Procrustes Analysis and Principal Component Analysis, which is fitted to candidate points, for extracting the final landmarks. The elaborated method is independent on face pose.

• 出版日期2012-12