CAJ | 학술논문

针对红外图像序列信噪比低，缺少边缘及纹理信息，以及人体目标运动自主性强且姿态复杂多变的特点，选用Mean Shift目标跟踪算法，对红外人体目标进行跟踪。针对传统Mean Shift目标跟踪算法不能对尺度变化的红外目标进行有效跟踪的问题，对原有算法进行改进，使用边缘加权直方图与模板中心加权直方图计算Bhattacharyya系数的方式，实现核函数带宽的自适应，并利用界面友好的LabVIEW2011平台进行实现。通过两组实验证明了改进后算法的可行性和有效性。
침대홍외도상서렬신조비저，결소변연급문리신식，이급인체목표운동자주성강차자태복잡다변적특점，선용Mean Shift목표근종산법，대홍외인체목표진행근종。침대전통Mean Shift목표근종산법불능대척도변화적홍외목표진행유효근종적문제，대원유산법진행개진，사용변연가권직방도여모판중심가권직방도계산Bhattacharyya계수적방식，실현핵함수대관적자괄응，병이용계면우호적LabVIEW2011평태진행실현。통과량조실험증명료개진후산법적가행성화유효성。
Aiming at low SNR infrared image sequences , the lack of edge and texture information ,as well as inde-pendent and the characteristics of strong human target mo-tion and gesture complex , the paper choosese Mean Shift tracking algorithm to track the target for infrared body .In-frared target can not scale effectively tracking changes in the traditional Mean Shift tracking algorithm ,this paper has im-proved the original Mean Shift algorithm and by using edge-weighted histograms and histogram templates center-weigh-ted way to calculate Bhattacharyya coefficient to achieve nu-clear adaptive function of bandwidth and use friendly inter-face LabVIEW2011 platform to achieve , and has finally proved the feasibility and effectiveness of the improved algo-rithm through two experiments .