针对光学相干层析（OCT）技术受散斑噪声影响严重的问题，研究了低噪声OCT成像技术。在系统中加入标定臂，利用标定信号对纵向扫描位置进行精确定位，降低检测过程中样品微位移所引入的误差；并在此基础上对各次纵向扫描信号进行曲波(curvelet)变换，采用自适应阈值法滤除图像散斑噪声，最后通过对多次纵向扫描信号的curvelet系数进行权重滤波和均值处理进一步降低散斑噪声。实验结果证明，与未经curvelet滤波的多次纵向扫描平均图像相比，低噪声OCT成像技术可以提高信噪比18.35%。

This paper describes a low noise optical coherence tomography (OCT) technology based on curvelet algorithm and a demarcating arm to decrease the speckle noise in OCT. The demarcating arm in the Michelson interferometer makes the precise positioning of A-scans possible and diminishes the displacement error caused by tissue moving. Besides, each A-scan is curvelet transformed and adaptive filtered, the curvelet coefficients are weighted, averaged and reconstructed to reduce speckle noise. Compared with traditional A-scans average method, the application of curvelet algorithm and demarcating arm in low noise OCT provides 18.35% increase in signal to noise ratio.