In order to reduce the computational complexity of this algorithm, combined with the Fayyad boundary point principle, a new algorithm, which selects the best segmentation threshold of the continuous attribute values, was proposed. According to the principle that there always exist the boundary points at the optimal segmentation point of the continuous attribute values, the improved algorithm only selected the best segmentation threshold from the few points of boundary. The improved C4.5 classifier was established and trained, and then it was applied in the recognition of people and vehicle targets in video sequences. The results show that the computation of the improved C4.5 algorithm is reduced by nearly 20% and also greatly improves the efficiency of generating a decision tree, and at the same time, the classification accuracy is slightly increased.%结合Fayyad边界点原理提出一种新的连续值属性最佳分割阈值的选择算法.根据Fayyad连续值属性的最佳分割点总在边界点处的原理,只在连续属性分界点处的少数几个分割点中选择最佳分割阈值.构造并训练了改进C4.5分类器,将其应用于视频序列中的人车目标识别.实验结果表明:改进C4.5算法的计算量减少近20％,大大提高了决策树的生成效率,分类准确率也略有提高.
展开▼
