The wavelet transform divides image into different but interrelated multi-resolution and multi-level sub-bands, so quantization coding strategy of image wavelet coefficient has great flexibility. To address to energy aggregation and correlated fe

基于多特征的AdaBoost行人检测算法, 提出一种融合灰 度和轮廓信息的新的多特征综合表示方法. 该方法通过统计样本的权重直方图建立分类模 型, 并用多个直方图的乘积表示样本在多特征下对应的联合概率分布-AdaBoost pedestrian detection algorithm based on multi-feature, presents a fusion of grayscale and contour information of the new multi-feature

滚动轴承多部位多类别故障诊断属于大规模或者较复杂分类问题，利用智能方法诊断时需 要设计结构合理的神经网络才能实现高精度诊断 -Rolling multi-site multi-class fault diagnosis are more complex or large-scale classification problems, the need for the rational design of neural network structure in order to achieve