以DSP平台圆目标定位为研究对象，提出了连续阈值质心算法，利用算术平均原理对普通质心法进行了改进。结果表明，该算法减小了随机误差和系统误差。存在噪声的情况下，其定位精度能达到0.02~0.05 pixels，同等情况下与普通质心法相比较，精度提高了4~10倍，抗噪声能力更强，而计算量却没有增加，能够用于DSP平台的圆目标快速精确定位。

The location algorithm for circular target based on DSP requires less storage and computing cost, faster processing speed and higher accuracy.The continuous threshold centroid algorithm was presented.The ordinary centroid algorithm was inproved based on arithmetic average theory. The results of simulation and experiment validate that the continuous threshold centroid algorithm is effective. The results show that the random and system erros are decreased. The location precision of improved algorithm can reach 0.02 ~ 0.05 pixels.Compared with ordinary centroid algorithm under the condition of the same noise image, the improved algorithm has the advantages of strong antinoise ability. And the accuracy can be improved 4~10 times. However the speed of the improved algorithm is as high as ordinary centroid algorithm.It can be used in DSP platform to locate circular targets fast and accurately.