基于Zernike矩和SVM的不变性目标识别

Invariance-based target recognition using Zernike Moment & SVM

在图像目标识别中,目标图像平移、尺度和旋转不变性是一个重要前提。文中采用稳定性好的低阶几何矩特征实现目标的平移、尺度不变性变换;然后利用Zernike矩提取目标的旋转不变性特征;最后在目标不变特征空间通过支持向量机(SVM)分类器实现目标识别。实验证实了算法的有效性。

In target recognition, the invariance in target image translation, scaling and rotation is a very important factor. A novel approach is proposed in this paper. First, stable low-order geometric moments are used to conduct the translation and scaling invariance. And then Zernike Moment is used to extract rotation invariant feature. Finally, target recognition is conducted based on Support Vector Machine（SVM）. The experimental results demonstrate the robustness and accuracy of the proposed algorithm.