论文部分内容阅读
在复杂背景下,目标和背景的亮度或对比度变化以及光照不均匀等因素都会影响目标分割的结果,导致跟踪不稳定或失败。针对这些问题,介绍了基于相位一致性检测的目标跟踪算法。该算法利用log Gabor滤波器获取多方向相位一致图像,并根据最小尺度下6个方向log Gabor小波滤波图像的中值平方确定目标响应最大的3个方向,进行目标分割;利用分割结果确定目标大小、质心位置,进而实现波门自适应跟踪。实验表明,该算法不受亮度和对比度变化影响,采用一个固定的分割阈值,就能得到很好的分割结果,实现了对目标的稳定跟踪,具有较高的跟踪精度。
In complex contexts, changes in the brightness or contrast of the target and the background, as well as uneven lighting, can affect the outcome of the target segmentation, resulting in erratic or unsuccessful tracking. To solve these problems, the target tracking algorithm based on phase consistency detection is introduced. The algorithm uses log Gabor filters to obtain multi-directional phase-consistent images. According to the median squared of 6 directions log Gabor wavelet filters in the smallest scale, the maximum three directions of target response are determined and the target segmentation is performed. The segmentation result is used to determine the target size , Centroid position, and then to achieve self-adaptive Bowman track. Experiments show that this algorithm is not affected by the change of brightness and contrast. With a fixed segmentation threshold, the segmentation result can be well obtained, and the target can be tracked steadily with high tracking accuracy.