推估影像中人臉角度及運用3D人頭模型轉換成正臉影像

通常監視攝影機會裝設在天花板等高處，由上往下拍攝，因此所拍得的人臉通常並非正臉影像，而是具有俯角及旋轉角度的人臉影像。因人臉並非平面，無法單純使用平行投影(Affine Transform)方式來轉換成正臉。惟人臉資料庫通常為正臉影像，因此增加了辨識上的困難。目前對於辨識不同角度變化的人臉，有人臉特徵點擷取、2D模形轉換及3D人頭模型轉換等方法。本研究採用藉3D人頭模型，將各種角度人臉轉換為正臉影像。

關鍵字

3D人頭模型；人臉姿勢推估；轉成正臉； 3D模型；姿勢不變；人臉偵測；人臉辨識

並列摘要

Surveillance cameras are usually mounted near ceiling and pointing downward at an angle. Therefore, human face images acquired normally are not frontal facial images, but are faces looking downward and possibly sideways. Also because human face is not planar, simple affine transformation applied to the entire face would not be able to convert such a face image into a front viewing image. However, the face images collected in databases are mostly frontal face images. This discrepancy poses a series problem to face recognition tasks. Currently, on pose invariant face recognition, there are methods based on pose invariant facial features, methods based on database with multiple 2D views of each person, and methods based on 3D head models. In this study, we are trying to convert a downward and sideway facing face into a frontal face image using a 3D head model.

並列關鍵字

3D model ； pose estimation ； front viewing facial image,face detection ； face recognition

參考文獻

[2] G. Yang and T. S. Huang, “Human Face Detection in Complex Background” , Pattern Recognition, vol. 27, no. 1, pp. 53-63, 1994.

[3] J. Yang and A. Waibel, “A Real-Time Face Tracker”, Proc. Third Workshop Applications of Computer Vision, pp. 142-147, 1996.

[4]. K.C. Yow and R. Cipolla, “Enhancing Human Face Detection Using Motion and Active Contours”, Proc. Third Asian Conf. Computer Vision, pp. 515-522, 1998.

[5] K. Sobottka and I. Pitas, “Face Localization and Feature Extraction Based on Shape and Color Information”, Proc. IEEE Int’l Conf. Image Processing, pp. 483-486, 1996.

[8] M. Turk and A. Pentland , “Eigenfaces for Recognition”, J. Cognitive Neuroscience, vol. 3, no. 1, pp. 71-86, 1991.

國際替代計量

推估影像中人臉角度及運用3D人頭模型轉換成正臉影像

未授權

主題瀏覽