一种改进的Gabor滤波器特征抽取算法及其应用

特征抽取是模式识别中的一个关键问题。丈中提出一种改进的基于Gabor滤波器的特征抽取算法。该算法应用Gabor滤波器的多尺度特性与样本图像进行卷积，将得到的Gabor特征矢量，根据其邻近分量的离散程度进行加权处理。与传统方法相比，该算法可以有效增强离散程度相对较小的特征分量在分类中的作用，分类效果较好；同时充分利用样本图像的统计信息，具有一定的鲁棒性。将该算法应用于车辆检测系统中，数据表明其能有效降低车辆检测的错误率，增强系统的鲁棒性。

Improved Feature Extraction Algorithm Based on Gabor Filter and Its Application

Feature extraction is one of the key problems in pattern recognition field, An improved feature extraction algorithm was put forward based on Gabor filter, In this algorithm, features were extracted with the multi-scale recognition of Gabor filter and the extracted features were then weighted according to their neighboring features degree of dispersion, Comparing with the conventional methods, it can enhance the effect of the features whose degree of dispersion is relatively small and widely used the statistical information in the sample image, With the application in a vehicle detection system, the experimental data show visible improvements both in diminishing error rate and robustness.