基于ABCshift算法的目标检测与跟踪

Target Detection and Tracking Based on ABCshift Algorithm

针对Camshift算法需要人为定位的问题,利用帧间差分法检测运动目标的初始位置,采用ABCshift算法连续更新背景模型,自动降低类目标色在颜色概率分布图中的密度值,从而实现准确的目标检测与跟踪。实验结果表明,该方法可以解决Camshift算法进入大面积类目标色背景下目标跟踪效果不理想的问题,在复杂背景下具有较好的适应性。

Aiming at the problem that Camshift algorithm needs artificial orientation, this paper uses difference in frame to detect the initial position of moving target. It uses Adaptive Background Camshifl（ABCshift） algorithm to update the background model, and reduces the probabilities that are the same colors with the target in color probability distribution, and achieves accurate target detection and tracking. Experimental results show that this method can overcome shortcoming of poor performance with Camshift algorithm when the target enters a large area with the same color background, and it has a good adaptability in complicated background.