电子科技
電子科技
전자과기
IT AGE
2014年
9期
25-28
,共4页
行为识别%行为检测%目标跟踪%帧差法
行為識彆%行為檢測%目標跟蹤%幀差法
행위식별%행위검측%목표근종%정차법
action recognition%behavior detection%target tracking%frame difference
针对传统人体行为检测算法完整性差、计算量大等问题，提出了一种背景消减法与帧差法相结合的检测算法。为提高人体行为跟踪算法的稳定性与准确性，提出了基于直方图与区域的Kalman跟踪算法。仿真实验表明，该算法在人体行为检测与跟踪过程中完整性好、准确率高。
針對傳統人體行為檢測算法完整性差、計算量大等問題，提齣瞭一種揹景消減法與幀差法相結閤的檢測算法。為提高人體行為跟蹤算法的穩定性與準確性，提齣瞭基于直方圖與區域的Kalman跟蹤算法。倣真實驗錶明，該算法在人體行為檢測與跟蹤過程中完整性好、準確率高。
침대전통인체행위검측산법완정성차、계산량대등문제，제출료일충배경소감법여정차법상결합적검측산법。위제고인체행위근종산법적은정성여준학성，제출료기우직방도여구역적Kalman근종산법。방진실험표명，해산법재인체행위검측여근종과정중완정성호、준학솔고。
In view the deficiency of the traditional human behavior detection algorithm in integrity and calculat -ed amount , the interframe difference method is combined with background reduction division to detect the moving hu -man target .And a tracking method of Kalman based on adjacent frames area match and grayscale match is proposed to improve stability and veracity of target tracking algorithm .Simulation experiments show that this method has better integrity and accuracy rate in the human action detection and tracking .
