科技通报
科技通報
과기통보
BULLETIN OF SCIENCE AND TECHNOLOGY
2015年
6期
43-45
,共3页
乒乓球%图像处理%眼动识别%动作预测
乒乓毬%圖像處理%眼動識彆%動作預測
핑퐁구%도상처리%안동식별%동작예측
table tennis%image processing%eye movement recognition%motion prediction
眼动的速度最高可达到600°/s，具有比人手更灵敏的反应，因此有更好的实时性和应激性，通过眼动识别可以准确反映人体的动作特征，实现动作的判断和预测。采用眼动识别的图像处理算法实现对乒乓球运动员的动作预测，提高攻球效能，改进乒乓球攻击和防御的目的性。传统方法中对乒乓球运动员的眼动识别算法采用边缘特征融合算法，对运动员的动作变换跟踪性能不好，提出一种基于多模态融合眼动识别的乒乓球运动员动作预测算法。提取边缘特征，进行虹膜定位设计，统计搜索区域灰度直方图分布，建立虹膜颜色与边缘联合特征模板，通过多模态融合眼动时变分析，采用从粗到细的处理方法，在减少匹配对应项的同时进行动作预测相关系数匹配，预测跟踪目标中心，实现动作预测，仿真结果表明，算法对运动员的动作预测准确度高，实现对乒乓球运动员动作的实时跟踪和识别。
眼動的速度最高可達到600°/s，具有比人手更靈敏的反應，因此有更好的實時性和應激性，通過眼動識彆可以準確反映人體的動作特徵，實現動作的判斷和預測。採用眼動識彆的圖像處理算法實現對乒乓毬運動員的動作預測，提高攻毬效能，改進乒乓毬攻擊和防禦的目的性。傳統方法中對乒乓毬運動員的眼動識彆算法採用邊緣特徵融閤算法，對運動員的動作變換跟蹤性能不好，提齣一種基于多模態融閤眼動識彆的乒乓毬運動員動作預測算法。提取邊緣特徵，進行虹膜定位設計，統計搜索區域灰度直方圖分佈，建立虹膜顏色與邊緣聯閤特徵模闆，通過多模態融閤眼動時變分析，採用從粗到細的處理方法，在減少匹配對應項的同時進行動作預測相關繫數匹配，預測跟蹤目標中心，實現動作預測，倣真結果錶明，算法對運動員的動作預測準確度高，實現對乒乓毬運動員動作的實時跟蹤和識彆。
안동적속도최고가체도600°/s，구유비인수경령민적반응，인차유경호적실시성화응격성，통과안동식별가이준학반영인체적동작특정，실현동작적판단화예측。채용안동식별적도상처리산법실현대핑퐁구운동원적동작예측，제고공구효능，개진핑퐁구공격화방어적목적성。전통방법중대핑퐁구운동원적안동식별산법채용변연특정융합산법，대운동원적동작변환근종성능불호，제출일충기우다모태융합안동식별적핑퐁구운동원동작예측산법。제취변연특정，진행홍막정위설계，통계수색구역회도직방도분포，건립홍막안색여변연연합특정모판，통과다모태융합안동시변분석，채용종조도세적처리방법，재감소필배대응항적동시진행동작예측상관계수필배，예측근종목표중심，실현동작예측，방진결과표명，산법대운동원적동작예측준학도고，실현대핑퐁구운동원동작적실시근종화식별。
Eye movement speed can reach a maximum of 600 degrees/s, is more responsive than a human hand, real-time and stress so had better, with eye recognition can accurately reflect the motion characteristics of the human body, the real?ization of judgment and prediction of movement. Using image processing algorithms to achieve eye recognition of table ten?nis athletes action prediction, improve the efficiency of improved objective to attack the ball, table tennis attack and de?fense. The traditional method of eye movements in the recognition algorithm of table tennis athletes using the edge feature fusion algorithm, motion tracking performance of athletes transformation is not good, put forward a prediction algorithm of multimodal fusion eye recognition based on table tennis athletes in action. The edge feature of iris localization design, statis?tical search regional gray histogram distribution, the establishment of a joint feature of iris color and edge template, through multimodal fusion eye time-varying, using the processing method from coarse to fine, in reducing the matching item at the same time should be action prediction correlation coefficient matching, prediction and tracking target center, the realization of motion prediction simulation results show that the algorithm, the action of athletes high prediction accuracy, real-time tracking and recognition of table tennis athletes in action.
