【摘要】 提出了一种基于分支定界算法的白细胞图像特征选择方法 ,该方法可有效降低特征空间的维数 ,加速分类器的速度。为验证方法的有效性 ,分别用 10 6个原始特征和从 10 6个特征中选择 3 5个最优特征进行分类实验。结果表明 ,两种情况的分类效果无明显改变 ,有效缩短了分类器的分类时间更多还原

【Abstract】 A feature selection method for leukocyte classification based on branch & bound algorithm is proposed. It can be used to select the effective features from a feature space and speed up classification. To confirm its validity, 106 original features and 35 effective features from the 106 ones are applied for classification respectively. The results show that they hare almost equal classification accuracy and the technique described can obviously reduce time-cost for classification.更多还原