基于SIFT的目标跟踪算法研究

Target Tracking Method Based on SIFT

图像特征提取和描述是目标跟踪的一项关键技术。已经提出的许多特征提取算法精度稳定性不够,尤其对图像的变换不具有很强的鲁棒性,SIFT算法是目前最具鲁棒性的算法。将SIFT特征提取算法应用到目标跟踪系统中,使用重心算法计算匹配的特征点的重心作为目标的脱靶量,既加快了算法的精度也提高运算的速度。实验证明,算法对目标的旋转、遮挡、亮度变化具有很强的鲁棒性,并且跟踪速度满足实时性的要求。

It is be key technology that extract and descript image feature in target tracking algorithm. Though Many feature extracting methods have been proposed, precision and stabilization were very lower, especially when image is transformed. SIFT is the most robust algoritlun today. The proposed method uses SIFT to extract image features and computes target displacement through bary method which get the the bary of SIFT features. This method not only improves the speed of algorithm, but its precision. Experimental results show that this method has strong robustness to rotated, shielded, illumination changed and meet the real time requirement of target tracking system.