【摘要】 文章从复杂背景下运动小目标成像特性与目标、背景关系出发,深入分析了此类目标识别定位的难点。综合目标区中值滤波和自适应二值化处理,将目标与背景和噪声分离。通过模板匹配,确定目标亚像元定位的计算区域。然后,基于优化的邻域像素插值方法计算运动小目标在图像中的精确坐标。针对目标实测图像序列,通过与形心算法和质心算法识别定位精度的比较实验,证明该复杂背景下运动小目标亚像元识别定位算法具有稳定、有效、高精度等优点,并为相关领域内目标识别定位算法提供了宝贵的借鉴价值。更多还原

【Abstract】 According to the characteristics of the complex backgrounds, stochastic noises and moving puny targets, difficulties of sub-pixel identification and location have been analyzed thoroughly in this paper. Taken template matching method to locate calculation area for sub-pixel computing, then combined with the technologies of target area median filtering and adaptive thresholding, the sub-pixel dimensions of this type of targets can be calculated of much better precision after pixel gray optimization in comparison to normal algorithms. Experiment results prove this new algorithm is fast, stable and efficient with other possible usage in related applications.更多还原