论文部分内容阅读
针对目前图像跟踪器跟踪不稳定、跟踪精度不高及不能满足实时性要求等问题,提出了一种概率加权的质心跟踪算法。该算法首先对波门内的像素进行阈值分割,摒弃灰度低于阈值的背景像素,保留目标像素的灰度值,然后计算波门内目标区域的质心。实验结果表明:基于概率加权的质心跟踪方法能够有效降低复杂背景和噪声干扰,增强跟踪系统的抗干扰能力,减少传统跟踪系统中使用大量灰度梯度值带来的巨大计算量,从而提高跟踪器的精度和稳定性。创新点在于通过引入概率加权的方法,在计算初始时刻的目标质心时使用贝叶斯概率作为权重,而没有设置离散的阈值来分辨目标,减少了传统跟踪系统中使用大量灰度梯度值产生的计算复杂度。
Aiming at the problems such as unstable tracking of image tracker, low tracking accuracy and unable to meet the real-time requirements, a probability-weighted centroid tracking algorithm is proposed. The algorithm firstly thresholds the pixels in the wave gate, rejects the background pixel whose gray level is lower than the threshold value, keeps the gray value of the target pixel and then calculates the centroid of the target area in the wave gate. The experimental results show that the centroid tracking method based on probability weighting can effectively reduce the complexity of background and noise interference, enhance the anti-jamming ability of the tracking system and reduce the huge amount of computation caused by using a large number of gray gradient values in the traditional tracking system, Accuracy and stability. The innovation point is that by using the method of probability weighting, the Bayesian probability is used as the weight in calculating the target centroid at the initial moment without setting a discrete threshold to resolve the target, thereby reducing the amount of gray gradient generated in traditional tracking systems Computational complexity.