科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2015年
8期
156-158
,共3页
智能手机%动作识别%特征
智能手機%動作識彆%特徵
지능수궤%동작식별%특정
smart phones%action recognition%feature
通过对智能手机动作识别算法改进,提高智能手机的嵌入式集成水平和面向对象能力.传统的智能手机动作识别算法采用静态姿势状态空间模型提取算法,当状态空间出现敏感性变化时,识别效果不好.提出一种基于特征敏感性控制的智能手机动作识别算法.基于特征敏感性控制,按翻转的方向,分向前、向后、向侧手翻三种,三个方向上的姿态特征信息采集特征向量,计算智能手机动作状态的变换矩阵分布强度,通过特征敏感性控制,得到的特征方程对模糊状态动作进行完全跟踪补偿,提高手机动作识别能力.仿真结果表明,该算法能有效实现对智能手机动作的实时跟踪和识别,对手机状态特征的追随性能较好,识别误差收敛于0,性能优越,使得智能手机更具有面向对象性,提高了对手机的控制精度.
通過對智能手機動作識彆算法改進,提高智能手機的嵌入式集成水平和麵嚮對象能力.傳統的智能手機動作識彆算法採用靜態姿勢狀態空間模型提取算法,噹狀態空間齣現敏感性變化時,識彆效果不好.提齣一種基于特徵敏感性控製的智能手機動作識彆算法.基于特徵敏感性控製,按翻轉的方嚮,分嚮前、嚮後、嚮側手翻三種,三箇方嚮上的姿態特徵信息採集特徵嚮量,計算智能手機動作狀態的變換矩陣分佈彊度,通過特徵敏感性控製,得到的特徵方程對模糊狀態動作進行完全跟蹤補償,提高手機動作識彆能力.倣真結果錶明,該算法能有效實現對智能手機動作的實時跟蹤和識彆,對手機狀態特徵的追隨性能較好,識彆誤差收斂于0,性能優越,使得智能手機更具有麵嚮對象性,提高瞭對手機的控製精度.
통과대지능수궤동작식별산법개진,제고지능수궤적감입식집성수평화면향대상능력.전통적지능수궤동작식별산법채용정태자세상태공간모형제취산법,당상태공간출현민감성변화시,식별효과불호.제출일충기우특정민감성공제적지능수궤동작식별산법.기우특정민감성공제,안번전적방향,분향전、향후、향측수번삼충,삼개방향상적자태특정신식채집특정향량,계산지능수궤동작상태적변환구진분포강도,통과특정민감성공제,득도적특정방정대모호상태동작진행완전근종보상,제고수궤동작식별능력.방진결과표명,해산법능유효실현대지능수궤동작적실시근종화식별,대수궤상태특정적추수성능교호,식별오차수렴우0,성능우월,사득지능수궤경구유면향대상성,제고료대수궤적공제정도.
Through the improvement of the intelligent mobile action recognition algorithm, improve the embedded integrat-ed level of smart phones and object oriented capabilities. Intelligent mobile action recognition algorithm of traditional ex-traction algorithm by using static posture state space model is constructed, when the state space sensitivity change, the rec-ognition effect is not good. Presents a new intelligent mobile action recognition algorithm based on feature sensitivity con-trol. Based on the characteristics of sensitivity control, in the reverse direction, divided into forward and backward, to a cart-wheel three, collecting the characteristic attitude feature information vectors in three directions, the intensity distribution of computational intelligent mobile action state transformation matrix, control through the characteristics of sensitivity, the characteristic equation is obtained by the fuzzy state action to complete tracking compensation, improve the mobile action recognition ability. Simulation results show that the algorithm can effectively realize the real-time tracking and recognition of intelligent mobile action, on the phone state characteristics follow performance is better, the identification error converg-es to 0, superior performance, the smart phone has a more object-oriented, improve the control precision of the mobile phone of.