Automatic 3D Face Feature Points Extraction with Spin Images

Abstract

Abstract. We present a novel 3D facial feature location method based on the Spin Images registration technique. Three feature points are localized: the nose tip and the inner corners of the right and left eye. The points are found directly in the 3D mesh, allowing a previous normalization before the depth map calculation. This method is applied after a preprocess stage where the candidate points are selected measuring curvatures on the surface and applying clustering techniques. The system is tested on a 3D Face Database called FRAV3D with 105 people and a widely variety of acquisition conditions in order to test the method in a non-controlled environment. The success location rate is 99.5% in the case of the nose tip and 98% in the case of eyes, in frontal conditions. This rate is similar even if the conditions change allowing small rotations. Results in more extremely acquisition conditions are shown too. A complete study of the influence of the mesh resolution over the spin images quality and therefore over the face feature location rate is presented. The causes of the errors are discussed in detail.

Unique Signatures of Histograms for Local Surface Description

“…σ 1 ) and resample the 3D meshes down to 1/8 of their original point density. For a fair comparison in this experiment, our implementation of SI -used throughout all the evaluation-normalizes each descriptor to the unit vector to make it more robust to density variations [3]. Finally, in Experiment 3 the dataset consists of scenes and models acquired in our lab by means of a 3D sensing technique known as Spacetime Stereo [21], [22].…”

“…These tasks find numerous applications in fields such as robotics, automation, biometric systems, reverse engineering, search in 3D object databases [1] [2] [3].…”

3D Face Landmarking Method under Pose and Expression Variations

“…For neutral frontal face, we obtain localized rate of 99.18% on GavabDB database and 100% on FRGC 1.0 databse. However, C. Conde [12] achieved localized rate of 98.65% on FRAV database and X. Dong [13] obtained the nose tip localized rate of 99.2% using 3DRLS and profile analysis on BJUT 3D database. Table 3 summarizes their results as well as ours.…”

Multi-attribute robust facial feature localization

“…However given the noisy nature of the 3D data, they require a preprocessing stage to reduce spikes and discontinuities. Recent 3D landmarking studies can be listed as [7], [9], [12]. In addition, multi-modal approaches that combine 2D and 3D data, have shown promising performance [4], [16].…”