基于RBF核函数的集成分类AdaBoost算法研究

摘要

The high complexity of motion estimation is always the bottleneck of the real-time video encoding application. To solve this problem , an enhanced motion search algorithm besed on the study of motion vector field adaptive search technique ( MVFAST) was proposed. The algorithm didn' t only use the length information from adjacent blocks'motion vectors but also exploited their separation degree to get a more accurate search strategy. Experimental results demonstrate that the proposed algorithm achieves a higher search speed at the same time offers a search quality superior to that provided by MVFAST and it can be used as a reference for the real time optimization on MPEG2 or H. 264.%基于径向基函数的神经网络、支持向量机已被广泛应用于模式分类.为了进一步提高分类的精度,将径向基函数应用于集成的AdaBoost算法,即以RBF神经网络和以RBF核函数的支持向量机分别作为AdaBoost的弱分类器,集成更高精度的强分类.通过对标准数据集的分类实验性能对比,证明了其算法解决分类问题有效性.