非负矩阵分解在交通标志标线特征识别的双约束算法

IDENTIFICATION OF THE TRAFFIC SIGNS MARKING GRAPHICS ON NON NEGATIVE MATRIX FACTORIZATION

对一类非负矩阵分解算法进行改进，在原算法的基础上增加了对基矩阵稀疏度控制参数μ。该算法有利地提高了处理交通标志标线图形类非负矩阵分解效率，通过双约束参数对目标函数的稀松度进行了有效控制，在保证交通标志标线分解有效性的基础上，提高了非负矩阵分解的可控性和有效性。

By adding the control parameters of sparse matrix,fast algorithm for a class of sparse non negative matrix factorization is improved on the basis of the original algorithm. The algorithm is advantageous to improve the actorization efficiency in processing of traffic signs marking graphics class of non negative matrix,and the sparse of the objective function has been effectively through the double constraint parameters.