在光电跟踪中,对于低信噪比(SNR)的非对称激光目标图像,采用对Z. Zeev的方法改进过的宏像素质心迭代法来定位目标可明显提高抗噪声能力。该方法利用最大宏像素概念对目标区域进行初步估计,减少了计算量,提高了目标覆盖率。通过改进权值函数和比例函数,进一步提高了抗噪声能力。对于128 pixel×128 pixel的远场红外光斑图像,大小约为4 pixel×4 pixel的非对称目标,当信噪比为-5 dB时,利用该方法得到的质心误差仅为0.3 pixel,平均每帧仅需时50 ms。

In optical tracking system,the macropixels method which is improved by method of Z. Zeev can enhance anti-noise ability to localize target. The computation complexity is reduced and target percentage of coverage of target is increased. Especially by using manimal macropixels to estimate the target region,the noise sensitivity to localization can be reduced for unsymmetry target because of improved method. It is proved that the method can improve precision and speed effectively. The centroid precision is only 0.3 pixel when signal-noise-ratio (SNR) is -5 dB,and the processing time is 50 ms for the unsymmetry target of 4 pixel×4 pixel in infrared spot image of 128 pixel×128 pixel.