改进的二维变换Gabor小波滤波器特征提取算法

Feature extraction algorithm based on improved 2D Gabor wavelet filter

针对在复杂背景下图像特征存在提取困难的问题,提出一种基于尺度不变性的改进二维Gabor变换小波滤波器特征点提取算法。增加在滤波过程中Gabor滤波器的通道数量,利用多通道特性的特点将图像中的特征点提取出来,增强其物理直观性;在提取特征点的过程中,增强算法在复杂背景干扰下的健壮性,即将尺度空间理论引入到Gabor滤波器,将提取的特征增加一个尺度不变量,使得算法在复杂背景图像中提取特征的能力加强。

For the feature acquisition of complex background image,an improved wavelet filtering feature extraction method was proposed.The channel number of Gabor filter was increased,and the feature of the image was then more physically intuitive.The scale space theory was introduced to the Gabor filter,and the features of the extracted features were added to a scale invariant,which enhanced the ability of the algorithm on extracting features in complex background images.