改进SIFT融合五官特征的旋转人脸检测算法

针对传统PID控制算法对四旋翼飞行器的姿态角进行控制时,其控制参数很难随着环境的变化进行自整定,进而影响四旋翼飞行器的稳定性的问题,提出了一种改进的PID控制算法。该算法通过改进单神经元中的K值公式提高了Kp、Ki、Kd的学习速率,从而提高了系统的响应速度；通过增加超调惩罚措施,通过适当的放大或缩小超调,可以使系统超调达到最小。通过与增量型PID、标准二次型单神经元PID算法进行比较,并且进行仿真实验。结果表明：所提方法具有参数自整定能力强且快速响应、鲁棒性强及稳定性好。

Face Detection Rotation Based on Improved SIFT Integration of facial features

When PID controller is adopted to control quadrotor attitude angle, one of difficult problems is how to self-tune parameters of controller with the variance of circumstance, which affects the stability of quadrotor. An improved single neuron PID controller is proposed. The algorithm improved K value formula of single neuron PID control to improve the Kp, Ki, Kd learning rate, which increase the response speed of the system; by increasing the overshoot punitive measures, through appropriate zoom in or out overshoot, overshoot can be minimized. By comparing with standard PID controller and standard quadratic single neuron PID algorithm and simulating, the result shows that the algorithm has a strong self-tuning, quick response, strong robustness and a good stability.