大气湍流对实现扩展目标的高分辨率重建具有重要影响.针对此问题, 本文提出了一种改进的散斑成像算法.传统散斑成像算法在相位恢复计算中存在双谱数据量大和计算复杂等问题, 改进算法利用图像的厄米特对称性和查找表技术将相位恢复和双谱计算紧密结合, 通过计算截止频率内的每个空间频率点邻域双谱和添加双谱坐标约束使得双谱数据量减小.建立傅里叶频域相邻两象限共用的坐标查找表, 确定双谱和相位恢复计算顺序, 避免了双谱的对称操作从而使得整个计算简单易行.仿真实验结果表明: 改进算法相对于双谱截切法使得双谱数据量至少减少了24%并准确恢复出目标相位谱, 恢复相位谱经过傅里叶逆变换后清晰地显示了目标的轮廓和结构, 再结合Labeyrie-Kroff法得到了目标的高分辨率图像；最后对实际天文图像进行处理, 使恢复后图像的分辨率相对于原始图像得到明显提高, 并且改进算法以更少的计算时间获得了与双谱截切法几乎同样的恢复效果.

In order to overcome the effect of the atmosphere turbulence and reconstruct the high resolution image of the extended object, an improved speckle imaging algorithm based on bispectrum truncation method was proposed. There are two problems in phase recovery using the conventional algorithms: one is the enormous amount of the bispectrum data, and the other is the complicated computation. The improved algorithm combines phase recovery with the bispectrum computation by the Hermite symmetry of image and the look-up table technique. Only the bispectrum in neighbored domain of a spatial frequency below the cut-off frequency are calculated and some constraint is added to the bispectrum coordinates, therefor the improved algorithm makes a dramatic reduction in the bispectrum data. The coordinate look-up table, which is applicable to the two neighbored quadrants in Fourier domain, describes the computation sequence of the bispectrum and the phase recovery. The look-up table makes the phase recovery become simple, for the bispectrum and the object phase can be obtained just in order of the look-up table. The results on simulated demonstrate that the improved algorithm is able to recover the object phase correctly with a 24% reduction in the bispectrum data. The phase spectrum obtained by the improved algorithm displays the profile and structure of the object after inverse Fourier transform. The results on real world astronomical data demonstrate the improved algorithm can restore the image with higher resolution than the original image and it obtain almost the same final image as bispectrum truncation method with less computation time.