基于概率修正的机场场面运动航空器跟踪算法

基于光电设备的航空器识别跟踪是机场场面监视的重要手段。针对机场场面全景视频监视中运动航空器跟踪算法存在的计算效率低、目标丢失等缺点,提出了基于概率修正的场面运动航空器跟踪算法。首先,预先估计出运动航空器目标概率图,降低在搜索区域内搜索目标时的计算量。通过引入干扰项抑制后,降低对真实目标的干扰。其次,在当前搜索区域内滑动窗口依次计算候选窗口是航空器目标的得分,选取得分最高的候选窗口作为新的航空器目标位置。再次,根据航空器目标位置更新概率图。最后,在多个场面监控视频图像序列上进行测试,对比分析了算法对目标周围出现的相似的区域有较好的抵抗力,在跟踪所得目标区域与真实目标区域的重叠面积率方面具有显著优势,验证了算法能够在目标尺度变化较大的情况下较好的实现对运动航空器的连续准确跟踪,能够满足对场面运动航空器跟踪的有效性和稳定性要求。

Tracking Algorithm of Moving Aircraft in Airport based on Probability Correction

The identification and tracking of aircraft based on photoelectric equipment is an important means of field surveillance in airport. Aircraft tracking algorithm in panoramic video surveillance has some shortcoming, such as low computational efficiency, loss of target. The tracking algorithm of moving aircraft in airport based on probability correction is presented. Firstly, the target probability map of the moving aircraft is estimated, and the amount of searching computation is reduced in searching the target area. The disturbance of the real target is reduced by introducing the disturbance suppression. Secondly, the candidate window as the target of the aircraft is calculated by sliding window in the current search area, and the candidate window with the highest score as the new aircraft target position will be selected. Thirdly, the probability map is updated according to the aircraft position. Finally, a series of experiments are carried out on some sequence of video images. The results show that the algorithm has good resistance to the similar regions appearing around the target, and has a significant advantage in computing the overlapping area ratio between the identified and the real target region. It is proved that the algorithm can achieve continuous and accurate tracking of the moving aircraft under the condition that the target scale changes greatly, and can satisfy the requirement of the availabilityand stability of aircraft tracking.