摘要
结合局部Fisher判别、张量子空间学习和零空间分析等技术的优点,提出了一种基于零空间分析的张量局部Fisher判别算法,其特点包括:i)引入类间判别信息,对局部Fisher判别技术进行调整,提升了算法识别性能并且降低了计算时间复杂度;ii)通过张量型降维思想对输入样本进行双边投影变换而非单边投影,获得了更高的信息压缩率;iii)随着训练样本量的变化,可采用基于零空间分析的求解方法和传统的直接迭代更新计算方法。通过ORL、Yale和ExYaleB 3个人脸数据库验证了所提算法的性能。
The tensor local fisher discriminant algorithm with null space analysis or NSTLFDA for short was proposed which incorporates the merits of three techniques, i. e. , tensor based methods, local Fisher discriminant analysis, and null space analysis. The main features of our implementation include： （i） local Fisher discriminant analysis is improved by inter-class discriminant information for better recognition performance and reduces time complexity, ii） the tensor based method employs two-sided transformations rather than single-sided one, and yields higher compression ratio, iii） while TLFDA directly uses an iterative procedure to calculate the optimal solution of two transformation matrices, the NSTLFDA method takes the advantages of null space information when the training samples number is less than the di- mensionality of the vector samples. The effectiveness of our new method was demonstrated by the ORL, Yale, and ExY- aleB face databases.
出处
《计算机科学》
CSCD
北大核心
2013年第5期11-18,37,共9页
Computer Science
基金
国家自然科学基金(61070043)
浙江省自然科学基金(LQ12F03011)
浙江工业大学校自然科学基金(2011XY020)资助
