摘要
中心频率位于高频段的Gabor滤波器具有较小的窗口,描述图像的局部特征,满足特征点表示的独立性要求;图像的高频信息随位置的不同有显著的变化。所以,在利用Gabor小波变换系数对人脸图像的特征点进行定位时,图像的高频信息更有利于特征点的定位。本文研究了不同频率的Gabor及其组合对特征点定位的影响,通过分析和试验得到了有利于人脸特征定位的Gabor频率组合。
Gabor filters whose center frequencies lie in high frequency band have smaller window size and describe me local features of image, this satisfy the need of independence for feature points representation. High frequency information of image varies distinctly with the change of position. Hence, when locating the feature points in the face image by Gabor wavelet transform coefficients, high frequency information are more favorable for feature points location. This paper discusses the influence of different frequencies and their combination on the feature point location. And the Gabor frequency combination suitable for facial feature location is obtained by analysis and experiments.
出处
《自动化技术与应用》
2008年第6期35-39,共5页
Techniques of Automation and Applications
基金
河南科技大学人才科学研究基金(200601)
