针对传统CCD（charge coupled device）交汇立靶采集的图像中，枪弹位置提取时采用背景差分法、互相关法所存在通用性差、耗时长的问题。通过对CCD精度靶图像弹丸提取所存在的问题进行深入分析，提出了基于改进灰狼算法的CCD精度靶图像弹丸提取方法。首先，将子弹提取问题转化为在一定约束条件下，寻找灰度值最小连通区域问题。其次，建立了最小化区域灰度值模型、竖直光斑区域及低灰度区域剔除模型。然后，采用基于维度学习的狩猎（dimensional learning-based hunting, DLH）搜索策略的改进灰狼算法，来跳出局部最优解，进而提升求解性能。最后，在参数设定相同的条件下，采用改进的灰狼算法、灰狼算法、飞蛾扑火算法、互相关算法、背景差分法进行了对比试验。实验结果表明，在上述方案下，平均求解时间缩短至12 ms。同时，目标检测成功率达到了95%，相较其他对比算法，性能提升明显。

The background difference method and cross-correlation method used in the extraction of the bullet position in the images collected by the traditional CCD(charge coupled device) intersection stand-up target have the problems of poor versatility and long time-consuming. By analyzing the problems existing in CCD precision target image projectile extraction, a method for CCD precision target image projectile extraction based on IGWO (improved grey wolf optimizer) algorithm was proposed. The DLH (dimensional learning-based hunting) search strategy was used to update the position of each search factor through the neighborhood. Generate candid ate solutions, increase the diversity of search populations, and jump out of local optimal solutions. The bullet extraction problem was transformed into the problem of finding the minimum connected region of gray value under certain constraints. The minimization area gray value model, the vertical light spot area and the low gray area elimination model were established. Under the same parameter setting, the IGWO, GWO(grey wolf optimizer), MFO (moth-flame optimization) algorithm, cross-correlation algorithm, and background difference method were used to conduct comparative experiments. The experimental results show that the target detection success rate of the IGWO algorithm is much higher than other algorithms, reaching 95%, and the algorithm solution time is much lower than other algorithms, shortening to 12 ms.