摘要
针对模式识别的主要研究问题,在此采用距离测度、贝叶斯、决策树、神经网络等分类算法,通过两组不同图像数据实验分析,比较分类结果的正确率。为了满足实时性要求,提出了从计算复杂度和运算时间上衡量各种分类算法的策略,并综合比较几种算法的特点以及应用范围,针对不足之处,提出了图像分类的改进思想。
Aiming at main phoblems of pattern recognition,adopting several classification algorithms based on distance,Bayes,decision tree,ANN,and so on.Comparing and analysing the classification performance with correct rate by experiment in two different datasets.For real-time system,a strategy is proposed that compared all of the algorithms with computational complexity and operation time.Comparing the features and application of several algorithms,and some ideas about how to improve the technology is given.
出处
《现代电子技术》
2009年第18期59-61,64,共4页
Modern Electronics Technique
基金
西安邮电学院中青年教师科研基金资助项目(109-0410)
