estimated the 3D model of non-rigid objects from a video sequence recording human speech and used the probabilistic principal components analysis method to learn the 3D model of human face. used a universal model to reconstruct a 3D face model from a video sequence. It estimates the camera calibration parameters using the matching feature pixels in each frame image and then restores the depth information of the human 3D face geometry. The shape from motion-based method (SFM) uses the image sequence to estimate the 3D structure of the human face in a controlled motion. Hence, it is more suitable for static application scenes. Additionally, the active method also needs to take sufficient time to complete the scanning process, during which the face must be kept still for a few seconds. Moreover, sensor errors and the properties of the face surface mean that depth maps from depth sensors are often noisy. Unfortunately, the laser scanner device is harmful to human eyes and is too expensive for many applications. On one hand, the active method indirectly obtains high-quality 3D facial model using depth maps captured from sensors, such as a laser scanner device ( C y b e r w a r e T M) or depth camera ( M i c r o s o f t K i n e c t T M). These approaches can be divided into two categories: active and passive. In the recent decades, significant progress has been made in 3D face reconstruction and a number of methods have been proposed in the literature. Three-dimensional (3D) face reconstruction is one of the most fundamental and challenging problems in 3D vision systems, as it plays an important role in many fields such as face geometry measurement, face recognition, face component transfer and face replacement. A comparison with the state-of-the-art monocular bilinear model-based method shows that the proposed method has a significantly higher level of accuracy. Additionally, the proposed method uses only two or more uncalibrated images with an arbitrary baseline, estimating calibration and shape simultaneously. Meanwhile, it fully explores the implied 3D information of the multi-view images, which also enhances the robustness of the results. Furthermore, the texture constraint extracts a high-precision 3D facial shape where traditional methods fail because of their limited number of feature points or the mostly texture-less and texture-repetitive nature of the input images. The feature prior constraint is used as a shape prior to allowing us to estimate accurate 3D facial contours. It extends the traditional method using the monocular bilinear model to the multi-view-based bilinear model by incorporating the feature prior constraint and the texture constraint, which are learned from multi-view images. Here, we propose a novel facial reconstruction framework that accurately extracts the 3D shapes and poses of faces from images captured at multi-views. Furthermore, they are affected by the accuracy of the feature extraction method and occlusion. However, traditional methods often depend on monocular cues, which contain few feature pixels and only use their location information while ignoring a lot of textural information. Face reconstruction is a popular topic in 3D vision system.
0 Comments
Leave a Reply. |