基于ARINC661和遗传融合的手势识别算法研究

Research on gesture recognition algorithm based on ARINC661 and genetic fusion

传统的手势识别算法通常受到复杂手势模式和噪声干扰等因素影响,导致其精度受到了限制,且不符合航电系统规范,故提出一种符合ARINC661规范的蚁群非支配顺序遗传融合神经网络算法。通过蚁群优化算法优化初始种群,使用第三代非支配排序遗传算法挑选种群下一代个体,保留种群多样性。引入蚁群优化后的变异与交叉策略以及种群寻优更新策略,提高算法的收敛速度;并对神经网络权值和阈值进行全局优化,提高估计精度以及手势识别系统的鲁棒性。实验结果表明,相较于现有算法,所提算法显著提高了准确率和收敛速度,并降低了平均标准误差,为应用于航电系统中的手势识别精度不足问题提供了有效的解决方案。

Traditional gesture recognition algorithms are usually affected by complex gesture patterns and noise interference,which leads to limited accuracy and does not conform to avionics system specifications.Therefore,an ant colony non-dominated sorting genetic algorithm-back propagation neural network conforming to ARINC661 specifications is proposed.The ant colony optimization(ACO)algorithm was used to optimize the initial population,and the third generation non-dominant sorting genetic algorithm was used to select the next generation individuals to preserve the diversity of the population.The mutation and crossover strategy after ACO and the population optimization update strategy are introduced to improve the convergence speed of the algorithm,and the weight and threshold of the neural network are optimized globally to improve the estimation accuracy and the robustness of the gesture recognition system.The experimental results show that,in comparison with existing algorithms,this algorithm can significantly increase the accuracy and convergence speed,and reduce the average standard error,providing an effective solution to the problem of insufficient gesture recognition precision in avionics systems.