二维Gabor滤波器的快速实现

研究表明2D-Gabor滤波器能有效地应用于边缘检测、纹理分析、图象编码及增强等场合. 通常在实际应用中,为了取得良好的效果,2D-Gabor滤波器所构成的模板必须选得足够大. 我们根据2D-Gabor函数的特点设计了实现2D-Gabor滤波的两种方法.一是利用它的分离 性将2D-Gabor卷积分解成四个一维卷积;二是建立2D-Gabor卷积与高斯卷积之间的关系, 然后利用有限均匀递推滤波器的级联近似合成2D-Gabor滤波.文中详细讨论了这两种算法 的运算量及其特点.

An Efficient Algorithm for the Implementation of A 2D Gabor Filtering

Researches have been done to expose the efficiency of 2D Gabor filtering in edge detection,texture analysis, image coding and restoration, and motion measurement. Two algorithms for 2D Gabor filtering using the characteristics of 2D Gabor function have been developed. One algorithm uses four 1D Gabor filterings sequentially instead of its original 2D filtering. The other is to establish the relationship between the 2D Gabor filtering and the Gaussian filtering, and then to synthesize it by uniform cascaded filters. The performance of the algorithms is discussed in detail.