基于径向小波变换的图像特征提取算法
Wavelet-Radial Descriptor Based Algorithm for Image Feature Extraction
-
摘要: 提出了一种基于径向小波变换的图像特征提取算法,该方法使用径向对称函数来获得图像的边缘信息。不同于其他的边缘检测算法,本方法既不使用方向滤波器,也不使用图像的极坐标形式,而且适合于文本和非文本图像,容易实现。算法中的参数在数学上能够严格证明,且通过计算不断调整,生成不变的分辨率特征图像矢量。仿真结果表明,该方法能够用于基于内容的图像检索。Abstract: A wavelet-radial algorithm for invariant feature extraction is given by the wavelet techniques and the edge information is obtained using the radial symmetric functions.This method does not use direction filters and the images in polar form for rotation invariance,while it is suitable for both the texture and non-texture image and easily implemental.The parameter in the algorithm is mathematically proved,and justified through computations,that can generate the invariant discriminative feature vectors.Simulation results demonstrate that it can be used for the application of content based image retrieval.