基于复数小波变换的景像匹配算法

讨论了一种基于复数小波变换的景像匹配算法 ,并以此算法为基础给出了一个高效的景像匹配模型。依据生物视觉的空间频域通路思想 ,利用复数小波移动不变性 (shiftinvariance)和实数离散小波快速算法 (Mallat快速算法 )将图像分解成多层 ,模拟生物视觉的感知过程 ,分别在高、低空间频域内进行了由粗到精的快速匹配。通过实验证明 ,与传统的景像匹配算法相比具有速度快、匹配概率高、抗干扰能力强等优点。

A Scene Matching Algorithm Based on Complex-Valued Wavelet Transform

A scene matching algorithm based on complex valued wavelet transform is discussed in this paper. We propose an effective scene matching model based on this algorithm. According to the thinking of frequency field paths in vision, we decompose images into some levels by using shift invariance of complex valued wavelet and Mallat's fast algorithm. We simulate the process of apperceiving and complete the fast image matching from coarse to fine scale in high and low frequency fields. Compared with conventional methods, results show that the proposed method has advantages of faster, higher matching probability and higher ability of anti jamming.