【摘要】 针对半自动CAMShift算法存在小目标颜色建模能力弱,计算窗口自适应能力不足和易受相似色目标干扰的问题,首先结合运动检测提取跟踪目标,并对颜色重采样以增强感兴趣对象,然后利用其在帧间运动的尺度、速度变化来优化计算窗口以减小计算量,最后提出一种遮挡抑制的思想解决相似色干扰问题.实验结果验证了所设计方法的正确性和鲁棒性.更多还原

【Abstract】 CAMShift is a semi-automatic algorithm which is weak in color modeling for small objects,lacks of adaptive scale of calculation window as well as becomes vulnerable under the interference of similar color objects.This paper takes advantage of motion detection to extract the target and resample the color feature for augment.Then,the scale and velocity changes between inter frames are combined to optimize the calculation window so as to reduce computations.Finally,an idea of occlusion suppression is put forward to overcome distractors problem.Experimental results validate the correctness and robustness of the raised methods.更多还原