频域光学相干层析成像技术是一种新型的医学成像技术，其传统的图像重构算法主要是基于傅里叶变换。但这种重构算法的主要缺陷在于其纵向分辨率随着深度位置的变化而明显下降。为了使频域光学相干层析成像系统的纵向分辨率在整个成像深度内基本保持不变，提出了一种基于观察矩阵的图像重构法，并用该法重构了平面镜以及皮肤信号。结果表明，这种图像重构法能够使频域光学相干层析成像系统保持纵向分辨率不变。与文献报道相比，这种方法在保持系统简单性的同时，还保持了高的纵向分辨率，并在成像深度范围内使得纵向分辨率基本保持不变。

Spectral domain optical coherence tomography (SDOCT) is a new biological imaging technique. The traditional reconstruction method in SDOCT is based on Fourier transform. One disadvantage of this method is that the axial resolution of the system decreases with the increase of the imaging depth. A new reconstruction method is proposed which has the capability of maintaining the axial resolution over the all imaging depth. To prove the accuracy of this method, the cross-sectional images of the mirror and skin are obtained. The results demonstrate the effectiveness of the method. Compared with the methods reported, the new method has the advantage of not increasing the complexity of the system while having an almost constant axial resolution over the imaging depth.