机构地区: 华南师范大学计算机学院
出 处: 《计算机工程》 2009年第19期175-177,共3页
摘 要: 针对传统全变分(TV)模型在测试过程中存在的问题,将弹性网引入TV模型中,采用二次多项式对TV模型所丢弃的人脸低频信息进行光照归一化处理,并提取图像的高频信息,在YaleB图像库中测试其性能,仿真实验结果表明,相对于TV模型,TV+二次多项式模型能够有效提高图像识别率。 Aiming at the problems in test process of traditional Total Variation(TV) model, the elastic network is introduced into TV model. The face low-frequency information lost by TV model is conducted with illumination normalization by means of quadratic polynomial. The high-frequency information of images is extracted. The performance is tested in YaleB image library. The simulation experimental results show that, compared with TV model, this new model with quadratic polynomial can promote image recognition rate effectively.
领 域: [自动化与计算机技术] [自动化与计算机技术]