基于相位信息的图像特征检测算法:对称相位一致性

相位一致性是利用相位信息检测图像特征的成功算法，但其概念本身存在固有的局限性，即不能检测单频信号中的特征，而且存在“余震”现象，为了克服相位一致性的局限性，研究了4个对称相位的一致性，并利用相邻像素响应的符号来确定对称相位的位置，最后得到了新的对称相位一致性图像特征检测算法，实验证明，该算法不但能克服相位一致性的局限性，还提高了特征的定位精度和抗噪性能。

Image Feature Detection Technique Based on Phase Information:Symmetry Phase Congruency

Phase congruency is one of the successful image feature detection algorithm using phase information. However, phase congruency has some inherent limitations,namely, its inability to detect the features from single frequency signals and the "aftershock" behavior. To overcome the limitations of phase congruency, four symmetry phases are studied. Signs of responses of neighboring pixels are used to find the location of symmetry phases. Finally, the new image feature detection algorithm-symmetry phase congruency is presented. Experiments show that this algorithm not only can overcome the limitations of phase congruency, but also can improve the location accuracy and anti-noise ability.