提出一种改进的分区域差值算法,能有效的缩短一般人脸模型的轮廓变换时间。在图像拼接算法中,使用新的加权因子,使重叠区域能平稳的过渡,并且不出现亮度跃变现象。采用新的全视角人脸纹理生成流程,使人脸纹理更加逼真。实验证明,这些改进使三维人脸重建在时间和真实感方面均有较明显的提高。%This paper presents an improved sub-regional difference algorithm can effectively shorten the conversion time of the outline of the general face model.Image stitching algorithm,using the new weighting factor,so that the overlap region for a smooth transition,and the jump in brightness.Full view face texture generation process,make more realistic face texture.The experiments show that these improvements make the 3D face reconstruction in terms of time and reality have significantly improved.
展开▼
