OCT systems are subjected to various kinds of noise and this kind of mixed noise significantly affects the image quality and makes the further image processing tasks difficult to be carried out. Most of denoising methods are based on filtering algorithms in spatial domain or the transform domain, and some details and texture characteristics of tissue are inevitably lost, which may affect the following image segmentation, feature extraction and pattern recognition based on texture. We proposed a method consisting of several kinds of processing algorithms combined to optimize noisy OCT images. The type of noise did not need to be distinguished and the mixed noise is taken into account as a whole to be reduced. Constrained least squares filtering is applied to minimize noise and restore the image. The sequential gray scale transform and image enhancement are further carried out in succession to smooth the boundaries and enhance the contrast. The effectiveness is proven with the processing results of degraded OCT images heavily polluted by noise. The method can not only effectively reduce the noise, but also reserve the detail features as well as possible. Evaluation parameters, the region contrast (RC), signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR), are also obviously improved after being processed. This method can be used for images from different OCT systems, only adjusting three parameters, alpha (the fitting parameter), beta (the stretching coefficient) and T (the threshold used to determine the boundary points) according to different conditions. The algorithms of denoising, restoration, smoothing and enhancement are independent, so we can add or delete other steps to further optimize OCT images according to different applications. For more information see recent Article . Courtesy of Yanmei Liang from the Nankai University.