基于双谱奇异值和最优参数支持向量机的目标识别算法

提出了基于双谱奇异值和最优参数支持向量机的目标识别算法。首先从目标距离像的双谱中提取双谱奇异值特征；然后通过实验的方法确定支持向量机参数取值的范围，利用遗传算法获取支持向量机的最优参数；最后应用最优参数支持向量机对3种地面坦克目标进行识别。仿真实验表明，即使在较低的信噪比下该算法仍然可以取得比较高的识别率。

A target recognition algorithm based on singular values of bispectrum and support vector machine with optimum parameters

A target recognition algorithm based on the singular values of bispectrum and the support vector machine （SVM） with optimum parameters is put forward. Firstly, the singular value features are extracted from the bispectrum of the target range profile. Secondly, the parameter span of the SVM is determined through experiments, and the optimum parameters of it are obtained through genetic algorithm. Finally, the SVM with optimum parameters is applied to recognize three kinds of ground tanks. The simulation of the algorithm shows that high recognition rate can still be obtained even with low SNR.