摘要
传统的SURF方法对动画图像角度特征匹配准确度较低,计算开销大,稳健性不好。引入Hes.sian矩阵尺度极值检测技术,提出一种改进的基于加速鲁棒特征匹配算法,实现对动画图像的角度特征优化匹配。首先基于摄像机的成像原理利用灰度直方图二进制均衡算法对图像进行增强处理,采用Hessian矩阵检测出图像每个尺度中的极值点,把图像角度极值点聚焦在图像的仿射闭合区域。提取仿射闭合区域的图像角度特征,使用SURF双向匹配算法实现角度特征优化匹配。仿真实验表明,改进方法能使图像的角度特征匹配准确度大幅提升,特征匹配准确率提高显著,有较好的鲁棒性和实时性。
According to the traditional SURF method, the accuracy is low, the computing cost is big, and the robustness is not good. The Hessian matrix scale extreme detection technology was introduced, an improved speeded-up robust feature （SURF） algorithm was proposed for realizing the animation image angle feature optimal matching. Firstly, on the basis of camera imaging principle, the gray histogram equalization algorithm of binary was used for image enhancement processing. Hessian matrix was used to detect the extreme point in each image. The image extreme points were focused in the affine closed region. The angle features were extracted, and SURF bidirectional matching algorithm was used to extract the fea-tures for the optimal matching. Simulation results show that the improved method can make the accuracy of image feature matching be improved greatly with good robustness and real-time performance.
出处
《科技通报》
北大核心
2014年第8期86-88,共3页
Bulletin of Science and Technology
基金
四川省教育厅青年基金课题(11zb173)
