提出了一种由仿射变换关系到线性回归的3d人脸空间姿态估计方法。
A method for estimating 3d space face pose was proposed based on affine transformation and linear regression.
对非线性仿射系统，将状态估计问题转化为微分和高阶微分提取问题。
The problem of states estimate is converted into the one of extracting the differential and high orders differentials for nonlinear affine system.
本文对H．263中基于平移模型的重叠块运动估计补偿方法进行改进，提出了一种基于仿射模型的重叠块运动补偿方法，该方法对复杂运动图像质量有明显改善。
This paper proposes an overlapped block affine motion compensation method to improve the translational motion model used in H. 263, and it improved the quality of image with complicated motion.
这种方法是建立在轴仿射模型与传统仿射模型的等价关系上，通过主轴进行估计。
Then we use the DIRECT algorithm to estimate the axis affine parameters, and convert to the traditional affine parameters.
仿射变换下的点模式匹配和参数估计是计算机视觉中的重要研究课题，其关键是寻找对仿射变换不变的不变量。
Matching and parameter estimation of two point-patterns under an affine transform is a very important research topic in the field of computer vision.
再通过计算曲线的统计特性作为图像间待匹配特征，并由信号相关的度量方法粗估计出图像间全局仿射变换参数。
After calculating the statistical information of curves as matching features, we can evaluate the coarse affine transformation parameters using simple correlation measure.
再通过计算曲线的统计特性作为图像间待匹配特征，并由信号相关的度量方法粗估计出图像间全局仿射变换参数。
After calculating the statistical information of curves as matching features, we can evaluate the coarse affine transformation parameters using simple correlation measure.
应用推荐