基于改进自适应正交匹配追踪算法的手势识别

Gesture Recognition Based on Modified Adaptive Orthogonal Matching Pursuit Algorithm

针对稀疏求解算法在稀疏度估计上的优势和增加固定步长的不足,提出改进的自适应正交匹配追踪算法。该算法引入稀疏度和变步长步骤,首先通过匹配测试来估计稀疏度初始值,以减少后续迭代次数,然后在不同阶段调整步长来筛选原子数,逼近真实稀疏度。实验结果表明,与其他贪婪算法相比,该算法有效提高了识别精度和效率。

A modified adaptive orthogonal matching pursuit（MAOMP）algorithm was proposed to guarantee advantages in sparsity estimation and overcome the disadvantages of increasing fixed step values in sparse solution.The algorithm introduced sparsity and variable step sizes.Initial value of sparsity was estimated by matching tests,and the numbers of subsequent iterations were decreased. Finally,the step sizes were adjusted to select atoms and approximate the true sparsity at different stages.Experimental results show that compared with other greedy algorithms,the proposed algorithm improves the recognition accuracy and efficiency.