针对真随机数生成问题，提出基于大气湍流光闪烁图像的真随机数提取方法。利用相机获取波长为532 nm的激光经过大气湍流传输后的光斑图像，根据其闪烁特性，分别采用固定间隔选取和多步长选取的方式选取光斑图像，固定间隔选择50帧，多步长选择30帧、70帧和100帧，得到的光斑图像相关性很弱，相关系数均小于0.3，由于湍流效应的影响，图像中的像素点发生无规则变化，通过对像素点作组合计算以提取随机序列；通过NIST(national institute of science and technology)随机性测试的方法，对提取的随机序列进行测试。实验结果表明：固定间隔选取的随机序列随机效果一般，测试结果存在P值小于0.01的情况，而多步长选取测试的P值均大于0.01，可以通过随机性测试。

Aiming at the generation of true random numbers, a true random number extraction method based on light scintillation images in atmospheric turbulence was proposed. First, we used a camera to obtain the light spot images of the laser with a wavelength of 532 nm propagating through atmospheric turbulence. Then, according to its scintillation characteristic, we selected the light spot images by fixed-interval selection and multiple-step selection, respectively,the fixed interval was selected as 50 frames, the multiple steps were selected as 30 frames, 70 frames and 100 frames. The correlation of the obtained light spot images was weak, and the correlation coefficient was less than 0.3. Because of the influence of the turbulence effect, the pixels in the image changed irregularly. We calculated the combination of pixels to extract the random sequences. Finally, we tested the extracted random sequences by the national institute of science and technology(NIST ) randomness test method. Experimental results show that the random effect of random sequence with the fixed-interval selection is general and the test results have the P value less than 0.01, while the P value obtained by multiple-step selection are all greater than 0.01 and the random sequence can pass the randomness test.