CAJ | 학술논문

人脸图像的识别率一直以来都是研究的热点。提出一种基于树状稀疏模型的算法。首先通过在小波域稀疏特性基础上构建一种图像数据的树状稀疏模型，进而针对此模型设计一种新的人脸图像重构算法。在人脸图像中使用消除人脸部的冗余信息，通过与传统重构算法的比较以及采用3个标准人脸数据集对该算法进行测试。仿真实验表明，该算法相对于参比算法，提高了人脸图像识别精度，具有很好的鲁棒性。
인검도상적식별솔일직이래도시연구적열점。제출일충기우수상희소모형적산법。수선통과재소파역희소특성기출상구건일충도상수거적수상희소모형，진이침대차모형설계일충신적인검도상중구산법。재인검도상중사용소제인검부적용여신식，통과여전통중구산법적비교이급채용3개표준인검수거집대해산법진행측시。방진실험표명，해산법상대우삼비산법，제고료인검도상식별정도，구유흔호적로봉성。
The recognition rate of face images has always been the hot topic of research.In this paper, we present a tree sparse model-based algorithm.First, we build a tree sparse model of image data based on the wavelet domain sparse features, and then we design a new face image reconstruction algorithm for this model.By applying it in face images the redundant information of the faces is eliminated.The al-gorithm is tested by comparing it with traditional reconstruction algorithm and adopting three standard face data sets.Simulation experiment shows that the proposed algorithm, compared with the reference algorithm, improves the precision of face image recognition and has very good robustness.