针对光电经纬仪跟踪测量过程的红外高炮序列图像的弹点目标,进行了相关研究。结合目标弹道轨迹规律,动态预测下一弹点位置,以预测点为中心的小范围图像区域内,使用最大类间差OSTU阈值法分割图像,对分割后的二值化图像进行连通性分析,提取出目标的脱靶量。以分辨率640×512的1000帧红外序列图像为实验数据,能够在2min内自动处理完成,且目标提取准确度99%以上。
The projectile point target in infrared image sequences during the photoelectric theodolite tracking process is researched in this paper. Based on the regularity of the target trajectory, the next bomb position is predicted. The predicting point is in the center 11 × 11 pixels, and within this range we use OSTU threshold segmentation algorithm to connectively analyze the segmented image to extract the target's miss distance. 1000 frame infrared image sequences with resolution of 640 × 512 as the experimental data can be completed within two minutes.