Gabor核函数窗的设置研究

Research on Design of Gabor Kernels’ Windows

Gabor滤波器广泛应用于模式识别等领域,但是在应用傅立叶变换进行Gabor快速卷积过程中,对Gabor核函数窗的设置研究较少。首先从理论上分析Gabor核函数窗的性质及其对特征提取的影响;其次从对称性、范围两个方面在YaleB人脸库上进行实验,研究Gabor核函数窗的不同设置对目标识别率的影响;最后给出相应的结论。

Gabor filters have been widely used in many applications such as pattern recognition. Unfortunately, the design of Gabor kernels＇ windows is seldom studied in the course of Gabor fast convolution with FFT. It analysed the characteristic of Gabor kernels＇ windows and the influence to feature extraction. Experiments were conducted on YaleB face image database to study the influence of Gabor kernels＇ windows design to recognition rates in terms of symmetry and scope. The conclusions were presented