论文部分内容阅读
在背景杂波被抑制后的图像序列中,残留样本为相互独立、服从高斯分布的条件下,首先论述了理想三维时空搜索检测算法,并对其性能进行分析。结果表明,虽然它具有最佳的检测性能,但是由于需要事先知道关于噪声及目标先验知识的缘故,无法进行实际应用。对此,研究了直接利用观测样本来估计噪声及目标的一、二阶矩,从而无须事先知道噪声统计特性的三维时空搜索检测算法及其详细步骤,推导了二元判决统计量所服从的概率分布函数(结果为t分布),对比分析了算法的检测性能,并给出了理想算法和本算法中共存的问题及其相应的改进方案。此种算法在连续多帧做任意运动的流星、人造卫星以及其他运动目标的光学检测与跟踪中应用广泛。
In the image sequence after the background clutter is suppressed, the residual samples are independent and subject to Gaussian distribution, the ideal 3D space-time search detection algorithm is discussed and its performance is analyzed. The results show that although it has the best detection performance, it can not be used in practice due to the prior knowledge of noise and target prior knowledge. In this regard, we study the direct use of the observed samples to estimate the first and second moments of the noise and the target, so that the three-dimensional spatio-temporal search detection algorithm and its detailed steps without prior knowledge of the statistical properties of the noise are derived and the probability obeyed by the binary decision statistics Distribution function (the result is t distribution), the detection performance of the algorithm is compared and analyzed, and the ideal algorithm and the coexistence of the algorithm and the corresponding improvement scheme are given. This algorithm is widely used in the optical detection and tracking of meteors, artificial satellites and other moving targets that make any movement in continuous multi-frame.