摘要
研究Gabor小波的构造及其特点,将其与AdaBoost算法相结合,优化原算法的性能.提取的Gabor特征对应到指定滤波器的指定采样点上,只需取这些指定采样点的一个领域内的图像与指定滤波器进行卷积,而不需要将整个图像与整个滤波器组进行卷积,可大大降低计算量提高运算速度.最后,通过算法对特征进行筛选,得到迭代误差率较低的特征来构建强分类器.
The paper has studied the structure and its characteristics of wavelet Gabor, by means of combining the Ada-Boost algorithm to optimize the performance of the original algorithm. Because the extracted Gabor characteristics are cor- responding to the designated sampling points, the proposed algorithm only need to get the convolution of specified filters and some images in some domains with these designated sample points, instead of getting the convolution with the entire images and the whole filter groups, which can greatly reduce convolution and improve the speed calculation. Finally, the algorithm can get the characteristics of lower iteration error rate to construct strong classifier through the choosing and culling.
出处
《华侨大学学报(自然科学版)》
CAS
北大核心
2011年第5期520-524,共5页
Journal of Huaqiao University(Natural Science)
基金
华侨大学科研基金资助项目(09HZRH)